Peer Review of the Greenhouse Gas
Emissions Model (GEM) and EPA's
Response to Comments
&EPA
United States
Environmental Protection
Agency
-------
Peer Review of the Greenhouse Gas
Emissions Model (GEM) and EPA's
Response to Comments
Assessment and Standards Division
Office of Transportation and Air Quality
U.S. Environmental Protection Agency
Sections prepared for EPA by
Research Triangle Institute International
EPA Contract No. EP-C-08-008
Work Assignment No. 3-02
NOTICE
This technical report does not necessarily represent final EPA decisions or
positions. It is intended to present technical analysis of issues using data
that are currently available. The purpose in the release of such reports is to
facilitate the exchange of technical information and to inform the public of
technical developments.
United States
Environmental Protection
Agency
EPA-420-R-11-007
August 2011
-------
Table of Contents
I. RTFs Technical Report of the Peer Review of EPA's Greenhouse gas Emissions Model
(GEM)
1. Background
2. Description of Peer Review Process
3. Summary of Peer Review Comments
4. B ackground Materi al s
II. Appendices
1. Appendix A: Charge Questions
2. Reviewer Resumes
3. Cover Letters
4. Individual Peer Reviews
i. Aris Babajimopoulos
ii. Daniel L. Flowers
iii. Shawn Midlam-Mohler
iv. Elliott Ortiz-Soto
III. EPA's Response to Peer Reviewer Comments
Page 1 of 104
-------
Memorandum
INTERNATIONAL
TO: Kent Helmer, Chi en Sze, U.S. Environmental Protection Agency, Office of
Transportation and Air Quality (OTAQ)
FROM: Tony Lentz, Paramita Sinha, and Karen Schaffner, RTI
DATE: January 28, 2011
SUBJECT: Peer Review of EPA's Heavy-Duty Greenhouse Gas Emission Model (GEM)
1. Background
EPA and the National Highway Traffic Safety Administration are considering a first-ever
program to reduce GHG emissions and improve fuel efficiency in the heavy-duty highway
vehicle sector. This broad vehicle sector, ranging from large pickups to sleeper-cab tractors
(Classes 2b through 8), represents the second largest contributor to transportation GHG
emissions after light-duty passenger cars and trucks. The agencies are proposing to evaluate both
fuel consumption and CO2 emissions from heavy-duty highway vehicles through a whole-vehicle
operation simulation model.
EPA has created a model called "Greenhouse Gas Emissions Model (GEM)," which is
specifically tailored to predict truck GHG emissions. As the model is designed for the express
purpose of vehicle compliance demonstration, it is less configurable than similar commercial
products and its only outputs are GHG emissions and fuel consumption. This approach gives a
simple and compact tool for vehicle compliance without the overhead and costs of a more
sophisticated model.
To assure that the regulated community gets the highest quality predictive tools that EPA can
provide and to assure its stakeholders that the proposed model structure (and overall
development process) will result in a tool that is simple, accurate and well suited for
certification, EPA sought an independent peer review of its GEM model.
2. Description of Review Process
EPA's Office of Transportation and Air Quality (OTAQ) contacted RTI International in October
of 2010 to facilitate the peer review of its Heavy-Duty Greenhouse Gas Emission Model (GEM).
" KU International is a trade name of Research Triangle Institute
-------
Kent Helmer
January 28, 2011
Page 2
RTI began the review process on October 19, 2010 and concluded January 21, 2011, a period of
slightly more than 3 months.
EPA provided a short list of subject matter experts from academia and the public sector
(Appendix B of the work assignment 3-02) to RTI, and this served as a "starting point" from
which we assembled the list of subject matter experts. RTI selected three1 independent (as
defined in Sections 1.2.6 and 1.2.7 of EPA's Peer Review Handbook) subject matter experts to
conduct the requested reviews. Subject matter experts familiar with MATLAB, Simulink,
Stateflow and Visual Basics software, as well as having expertise in vehicle operations and
analysis, linkages between mobile source emission modeling and transportation modeling and
planning, or application of current mobile source emissions models for analysis for regulatory
purposes were selected.
To ensure that the review process was conducted in a timely manner, RTI contacted potential
reviewers within 10 days of submitting the work plan and determined that each reviewer would
be able to perform work during the period of performance. To make the review process as
credible as possible, RTI did not consult the Agency in the final determination of reviewers. RTI
obtained the resumes of the selected reviewers, and these are included in Appendix B.
EPA provided RTI with the necessary model review material via email on November 12, 20102.
This was forwarded to the reviewers; and in addition to the review material, RTI forwarded a set
of charge questions prepared by the EPA (these questions were later revised).
On November 17, 2010, RTI organized and held a teleconference between EPA, the three
reviewers, and RTI to provide an opportunity to the panel to discuss any questions or concerns
they may have regarding the material provided and expected deliverables. The call concluded
when all participants' questions and concerns were addressed and a mutually agreed upon
deliverable date was set. Based on the discussion during the call, EPA sent RTI an updated set of
charge questions on November 23, 2010 and this was forwarded to the reviewers on November
29, 2010. These charge questions are included in Appendix A of this memorandum.
Following the first bi-monthly progress report call between RTI and EPA (November 16, 2010)
and subsequent email correspondence (November 18, 2010), it was also agreed upon that a
fourth subject matter expert should be identified and selected as a reviewer. EPA sent RTI an
expanded short list from which the fourth reviewer was identified and contacted by RTI. The
1 Initially, RTI identified 3 subject matter experts to serve as reviewers. Following the first bi-monthly progress report call
between RTI and EPA, it was agreed upon that a fourth subject matter expert should be identified and selected as a reviewer.
2 EPA distributed all the necessary review material to RTI in an email, which contained hyperlinks ("weblinks") to the review
material along with weblinks to both, the executable and MATLAB/Simulink, versions of the GEM and an accompanying
model guide.
Page 3 of 104
-------
Kent Helmer
January 28, 2011
Page3
necessary material and charge questions were forwarded to him upon his acceptance to
participate in this peer review.
Completed reviews from the panel were sent to EPA on Wednesday, December 22, 2010. These
reviews included the response to the charge questions and any additional comments the reviewer
may have had (e.g., margin notes on review materials). RTI also obtained a cover letter from
each reviewer stating the reviewer's name; the name and address of their organization if
applicable; and a statement of any real or perceived conflict(s) of interest. The cover letters and
reviews are included in Appendices C and D, respectively. EPA's comments in response to the
reviewers' assessments are included in Appendix E.
3. Summary of Review Comments
Aristotelis Babajimopoulos (University of Michigan, College of Engineering), Dan Flowers
(Lawrence Livermore National Laboratory, Combustion and Alternative Fuels, E Program),
Shawn Midlam-Mohler (Ohio State University, Department of Mechanical Engineering), and
Elliott Ortiz-Soto (University of Michigan, College of Engineering) reviewed EPA's GEM. This
section provides a summary of the comments received from them.
3.1 EPA'S OVERALL APPROACH TO THE STATED PURPOSE OF
THE MODEL (MEET AGENCIES' COMPLIANCE
REQUIREMENTS) AND WHETHER THE PARTICULAR
ATTRIBUTES FOUND IN RESULTING MODEL EMBODIES
THAT PURPOSE.
All four reviews addressed the model's ability to meet the agencies' compliance requirements.
One reviewer explicitly detailed whether the model's particular attributes embody the stated
purpose of the model. In general, the reviewers reported that the model does an acceptable job
testing different vehicle configurations from different vehicle manufacturers for compliance
purposes.
Dr. Flowers comments, "Overall, the concept of using a generic vehicle model has merit to limit
the need to test the myriad possible vehicle configurations. The use of a generic powertrain
(engine and transmission) is problematic because a well-integrated powertrain can significantly
improve vehicle performance." (Additional discussion of powertrain issues are provided in
subsection 3.2.1 below.)
Page 4 of 104
-------
Kent Helmer
January 28, 2011
Page 4
Dr. Babajimopoulos remarks, "GEM is a very detailed vehicle simulation that could capture with
reasonable accuracy the impact of changes in aerodynamic drag coefficient, tire rolling
resistance and tire weight reduction on overall vehicle fuel economy and CO2 emissions. The
model itself is almost too detailed for this purpose, but this should not be a problem, provided
that not all details of the model are discussed in such great length with the users. However... it is
hard to envision a compliance tool that does not account for fuel economy improvements coming
from the development of advanced combustion technologies by the engine manufacturers."
Dr. Midlam-Mohler addresses the five modeling attributes3 needed for the model to serve as a
primary compliance tool. Regarding each attribute, he comments:
1. "The model fidelity of the type proposed should be capable of achieving the desired
objectives. The model reviewed, however, has a number of issues which cast doubt
upon the specific implementation of the model. Specifically, a number of issues were
found in the electrical subsystem as well as the engine subsystem."
2. "Providing source code as a Simulink diagram is necessary for this objective but not
sufficient. Additional documentation on the equations and references behind the
Simulink code should be developed and released to the public."
3. "The compiled version of the code is free and easy to use. The Simulink version
requires a Matlab license which is not free but fairly common in industry."
4. "The current structure satisfied this objective."
5. "By releasing an official and unalterable executable version of the model this
objective is met."
While Mr. Ortiz-Soto provides specific comments on multiple aspects of the model (including
comments on inputs, outputs, model and submodels, see sections below for details), he also notes
that "In general, the rest of the model looks good." Mr. Ortiz-Soto reports, "Overall, the model is
in great shape and should be a strong starting point for a dedicated simulation oriented to
compliance purposes."
3 The five attributes listed in EPA's updated charge questions are: 1) capable of modeling a wide array of medium- and heavy-
duty vehicles over different drive cycles; 2) contains open source code, providing transparency in the model's operation; 3)
freely available and easy to use by any user with minimal or no prior experience; 4) contains both optional and preset
elements; and 5) managed by the Agencies for compliance purposes.
Page 5 of 104
-------
Kent Helmer
January 28, 2011
Page 5
3.2 THE APPROPRIATENESS AND COMPLETENESS OF THE
CONTENTS OF THE OVERALL MODEL STRUCTURE AND ITS
INDIVIDUAL SYSTEMS, AND THE COMPONENT MODELS, IF
APPLICABLE (i.e., USING THE MATLAB/SIMULINK VERSION)
This section is broken down into 4 subsections with each subsection containing one or more
comments from the reviews. In general, each reviewer commented on one or more of the
following subsections.
3.2.1 The Elements of Each System to Describe Different Vehicle
Categories
The GEM model has 6 systems (Driver, Ambient, Electric, Engine, Transmission, and Vehicle)
used to describe different vehicle categories. There was little, to no, discussion among the
reviewers concerning 2 systems, the ambient and driver systems. In the remaining four systems
(electric, engine, transmission, and vehicle), the reviewers found errors and identified issues that
raised questions about the overall ability of the systems to accurately depict different vehicle
categories.
Dr. Babajimopoulos details issues with the component models of the engine, transmission and
vehicle systems. He comments, "... .engine fuel maps and drivetrain parameters are hardwired in
the model and the user has no option of changing them. However, it seems counterintuitive that a
tool for determining compliance with emissions standards would ignore efforts on the part of the
manufacturers to make improvements on the engine itself. Moreover, in order to take full
advantage of any improvements in combustion and engine-out emissions, the vehicle
transmission needs to be optimized for a particular vehicle/engine/driving schedule combination,
so that the engine can operate near its optimum efficiency points at all times."
Dr. Flowers noted that in the GEM model, "the engine and transmission is not optimized to the
vehicle" and "The use of a generic powertrain (engine and transmission) is problematic because
a well-integrated powertrain can significantly improve vehicle performance." "In practice, the
engine and transmission can be appropriately sized to best take advantage of the reduced overall
vehicle load. By requiring only one engine and transmission be used, drag reduction efforts could
be penalized."
Mr. Soto also commented on the engine fuel maps. He stated that "One of the most important
input dat[um] for a fuel economy drive-cycle simulation is the engine mechanical load and fuel
consumption maps. The mechanical load maps are usually simple because only the WOT (or
Diesel equivalent) values are required, but obtaining full range fuel consumption values is much
more difficult. Several engine maps appear to be available for each vehicle class, but making
Page 6 of 104
-------
Kent Helmer
January 28, 2011
Page 6
these completely standard with a prescribed displacement volume and operating range might be a
limiting factor for some manufacturers. A more flexible approach would be to have normalized
load and fuel consumption maps, given in BMEP and BSFC values. The current maps can be
easily converted into BMEP and BSFC with the data available. The user could then provide the
engine displacement and possibly another key parameter such as rated torque or power and the
engine speed, and an algorithm could automatically manipulate the normalized maps to obtain
more representative absolute values for the engine in question. Even though this compliance tool
assumes that the engines have already been certified, the fuel economy and CC>2 values that the
simulation predicts are directly related to the maps given, and manufacturers might want to
ensure the engines in their vehicles are properly accounted for."
Dr. Flowers conducted a comparison of the GEM model output values and direct calculated
values for the same parameters for a particular vehicle configuration and drive cycles. He
determined that the direct calculated torque is 3 percent lower than the GEM-modeled torque,
and noted a possible explanation may be due to the speed variation during the constant desired
speed portion of the drive cycle. Referring to the chassis component model contained in the
vehicle system, Dr. Flowers reports that the powertrain inertial mass should be zero during a
certain drive cycle. He states, "The GEM model uses an "effective mass" formulation that
includes powertrain inertial effects. In the GEM code, the vehicle static mass
(vehicle.chsmass_static) is added to the representative powertrain inertial mass (tire_mass_out).
For steady speed vehicle operation the powertrain inertial mass should be zero." Dr. Flowers
compared GEM model output values and calculated values for fuel usage, fuel consumption, and
GHG emissions (using GEM output values for torque and speed), and he noted that errors were
small (less than 0.3 percent).
Dr. Midlam-Mohler summarizes his comments of the model systems and their underlying
components models by stating, "The overall approach of using a relatively simple model
structure based in Matlab-Simulink is sound provided that models are calibrated and validated to
a sufficient level."
Dr. Midlam-Mohler stated that some issues in the Engine subsystem need to be addressed and he
stated "The method of handling negative brake torques in the model does not seem to be
appropriate." Dr. Midlam-Mohler notes that "A map-based engine model should be sufficient to
achieve the desired objectives. The engine model implemented in the current version of the
software does not appear to be as well implemented as it could be. Given the importance of this
in the overall objectives of the simulator this needs to be addressed. Using fuel maps which have
torque indices ranging from a negative brake torque to the maximum rated torque would alleviate
much of the uncertainty in the model. Driver accelerator requests should then be linearly scaled
from minimum value to the maximum value on this map with the exception of idle conditions in
Page 7 of 104
-------
Kent Helmer
January 28, 2011
Page 7
which alternative measure must be taken. This approach also automatically takes into account
deceleration fuel cut-off as well."
Dr. Midlam-Mohler notes some recommendations for the Vehicle subsystem, stating "The most
serious item is considered to be the fact that the "Vehicle Weight Reduction" parameter is
specifically cited as being able to model light-weight wheels. The existing model structure would
not accurately do this as it does not take into account the inertial aspect of the wheels which
would have a greater impact on the vehicle."
Dr. Midlam-Mohler noted that in the Driver subsystem, the PID values are fixed in the GEM
model but that it may be worth adding this as an advance feature or using a more sophisticated
control concept, such as augment the current PID control with a feedforward component. He did
note that large errors in velocity tracking were not observed in the model.
Mr. Ortiz-Soto notes that "Control for most of the vehicle components seems to be achieved by
fairly standard PID controllers. Usually the gains for these controllers are tuned to a specific
plant, but in this case they remain fixed for all the vehicle configurations. Were these gains tuned
for all the plants individually and then somehow averaged to account for all of them, or were
they computed for a single vehicle? Although for the test cases do not show any major problems
with following the prescribed velocity profile, simulation of some vehicles or with a different set
of parameters could possibly suffer if the controller gains are not appropriate. For the driver, for
example, more elaborate, robust and reusable driver models exist, and it might [be] useful to
investigate the possibility of incorporating one of these in order to avoid possible issues with the
simulations."
3.2.2 The Performance of Each Component Model, Including the
Reviewer's Assessment of the Underlying Equations and/or
Physical Principles Coded into That Component
Four of the GEM model systems (electric, engine, transmission and vehicle) are made up of
underlying component models. The reviewers assessed the performance of those component
models, including the equations and physical principles of the component model, and reported
their findings. Each reviewer noted that one or more of the component models performed
inadequately and recommended these component model inadequacies be addressed to improve
the robustness of the compliance tool. Additionally, a couple of the reviewers identified non-
trivial errors in the equations of some of the component models. For example, one reviewer
states, "A number of errors were found in models within GEM. None of these errors are expected
to contribute to larger errors to the output results but should be corrected nonetheless."
Page 8 of 104
-------
Kent Helmer
January 28, 2011
PageS
Three of the reviewers commented on flaws in the electric system.
Dr. Babajimopoulos reports, "The model of the electric subsystem is particularly detailed and
convoluted. GEM includes submodels for the starter, alternator, battery and electric accessories.
This complexity seems unnecessary for the stated purposes of GEM. Careful examination of the
results reveals that the starter has almost zero effect on overall fuel economy and CC>2 emissions.
Moreover, the overall effect of the electrical system on fuel economy and CC>2 emissions is
almost negligible."
Dr. Midlam-Mohler comments, "Very significant issues were found in the electric subsystem
which require attention. In particular, the battery model appears to [have] an error which causes
battery voltage to decrease with battery state of charge which is exactly opposite of the desired
behavior. Furthermore, it appears that the sign convention used for the starter, accessories,
alternator have the wrong sense. The alternator generates negative current which decreases SOC.
The other two currents, which are current sinks, actually increase the SOC of the battery. Even
with the above issues aside, the alternator model appears to not consider the mechanical to
electrical efficiency of the device and the control is naive of actual alternator capabilities and
control."
Mr. Ortiz-Soto comments, "The electric components and EES seem to be fixed for all the
vehicles in the simulation, but in reality the electrical system is probably designed for a given
application to account for the particular load requirements. It is understandable that due to the
complexity of acquiring parameters such as these, the system model is standardized, but it could
also result in simulation inaccuracies. It might be more appropriate to provide at least some basic
scaling capability for the overall electrical system so that with one or two additional inputs, the
electrical components and EES are scaled to match the actual setup more closely." "A similar
observation can be made regarding the starter and alternator models." While these are not critical
components, a scaling factor should be applied.
Dr. Babajimopoulos found that the density of air value in the ambient system "seems to be rather
low" and could impact model results in a non-trivial fashion depending on the cycle.
3.2.3 The Input and Output Structures and How They Interface with
the Model to Obtain the Expected Result; i.e., Fuel Consumption
and CO2 over the Given Driving Cycles
Overall, the reviewers commented the input and output structures interfaced well with the model
to obtain the expected results. Several reviewers offered minor suggestions that could help the
end user when using the model. These suggestions are found in subsection 3.6.1 of this report.
Page 9 of 104
-------
Kent Helmer
January 28, 2011
Page 9
3.2.4 The Default Values Used for the Input Files, as Shown in the GEM
User Guide
All of the reviewers commented that the default values for the input files should be allowed to
change to reflect manufacturer improvements. The reviewers' comments reflect a concern that
the model does not allow for sufficient flexibility in certain respects. For example, Dr. Flowers
expresses his concern about standardization when he remarks, "My main concern with the
overall approach is the standardization of the vehicle and powertrain combination. This seems to
have the potential to devalue efforts towards vehicle and powertrain integration and optimization
towards GHG reduction."
Dr. Midlam-Mohler recommends that EPA allow some of the model parameters to change with
respect to vehicle class. He suggests, "A number of parameters were noted which should change
with respect to the vehicle class. The reviewer is certain that there are others that were not noted
in this review. It is recommended that the EPA investigate this and take an appropriate action. In
many cases, these components will not have a serious impact on the overall performance of the
vehicle. By way of example, many of the inertias simulated in the model will not have a large
impact on the results in contrast to the large inertia of the vehicle. If this is the case, then these
inertias could be discarded from the model with little impact on performance. If the detailed
inertias remain in the model, then they should accurately reflect the vehicle class."
Given the overall importance of fuel consumption and CC>2 emissions to the model's objective,
three reviewers specifically address the engine maps default values.
Dr. Babajimopoulos commented, "If the assumption is that engines will be relatively similar for
the same class vehicles coming from different manufacturers, then it is safe to assume that GEM
would be an appropriate tool for determining compliance with fuel economy and CC>2 emissions
standards based on vehicle design changes alone. Nevertheless, it would be proper to allow for
the provision to change the engine fuel map and transmission characteristics used by GEM."
Mr. Ortiz-Soto and Dr. Midlam-Mohler provide comments on the default values for the engine
fuel maps in subsection of 3.2.1 of this report.
3.3 USING THE STANDARD OF GOOD ENGINEERING
JUDGMENT, THE PROGRAM EXECUTION IS OPTIMIZED BY
THE CHOSEN METHODOLOGIES
One reviewer commented that he interprets this statement to be referring to "the performance of
the code as an effective tool for this application [regulatory application]." The reviewer states the
code seems to be developed in such a way that it provides detail on both the vehicle and
Page 10 of 104
-------
Kent Helmer
January 28, 2011
Page 10
powertrain dynamics. Because the model is complex and is a "highly interconnected system," he
expresses concern about the model documentation and believes more detail should be provided
about the physical models implemented. He feels that transparency in the details of the model is
important for regulatory application and the model may suffer without sufficient detailing of the
underlying physics and engineering assumptions.
3.4 CLARITY, COMPLETENESS AND ACCURACY OF THE
OUTPUT/RESULTS (CO2 EMISSIONS OR FUEL EFFICIENCY
OUTPUT FILE)
Two reviewers stated the model output was clear and one commented that it was complete. One
reviewer added, "The four tab format with the first tab being summary data and others being
cycle data was sufficient." A second review concluded, "The model reports the individual drive-
cycle results and weighted average results, which is what is most important to the end user." The
reviewer added, "All the inputs needed to reproduce the results are reported."
Two reviewers express concern about the clarity of the results with respect to the output file
naming scheme.
Mr. Ortiz-Soto comments, ".. .naming the files based on date and time is not very useful or
descriptive. When multiple simulations are performed, it becomes difficult to determine what file
you should be looking into, unless you actually open it. The file names should include at least
some sort of indication of what the simulation configuration was. The second problem I found
was the lack of flexibility to specify where these output files are saved. There should be an
option allowing the user to browse and select the main directory where these files are to be
saved. As a final comment on this, there is really no reason for each of these files to be saved to a
different folder if there is just a single output file. This simply adds an unnecessary layer to the
file structure."
Dr. Babajimopoulos raises a similar concern when he remarks, "It would be good if the message
indicating where the results will be stored also include the drive (C:) in the path (e.g.,
' C:\GEM_Results\December_l4_2010-013 5PM instead of \GEM_Results\December_l4_2010-
0135PM).'"
Regarding the accuracy of the output/results, Dr. Flowers indicates, "accuracy of the results is
difficult to assess, since that requires specific comparison to experimental data to evaluate the
performance of the model. Based on my testing efforts and experience, the results seem of
reasonable magnitude for these kinds of vehicles." Dr. Flowers concludes:
Page 11 of 104
-------
Kent Helmer
January 28, 2011
Page 11
"The model is quite detailed with regard to powertrain and vehicle dynamics. There is
a danger here that imbedded assumptions can affect results in unexpected and
undesirable ways. The example of the 3% difference in torque for analytical versus
GEM simulation calculated torque for steady state operation may be indicative of
these kinds of issues."
"Detailed description of the physics and assumptions imbedded in the models and
submodels should be documented and made available to users."
"It may be worth considering if the model could be streamlined to provide greater
clarity and transparency while still providing a tool for quantitatively estimating fuel
consumption and GHG emissions."
3.5 ANY RECOMMENDATIONS FOR SPECIFIC IMPROVEMENTS
TO THE FUNCTIONING OR THE QUALITY OF THE OUTPUTS
OF THE MODEL
The reviewers made several recommendations for improving the functioning and quality of the
outputs. Two reviewer recommendations have been detailed in section 3.4 regarding the output
file naming scheme. Additional reviewer recommendations are detailed below:
6. One reviewer recommends including additional results in the output. He believes, "It
would be informative to have the fraction of each drive-cycle used in the average
reported somewhere in the output."
7. Dr. Midlam-Mohler suggests, "End users will likely want to see more detail in the
output file then just the vehicle target speed and achieved speed. Making a limited
number of "internal" parameters available to allow end users a glimpse inside the
model without having to use Matlab-Simulink would be sufficient. These should be
limited to things relevant to their inputs, such as aerodynamic drag over the cycle,
rolling losses over the cycle, etc."
8. Mr. Ortiz-Soto offers a couple recommendations for improving the quality of the
outputs. He suggests for compliance purposes, ".. .it would be good to see the actual
target value next to the simulation result, and probably some sort of percentage
difference between these. It would give the manufacturer/user an idea of how their
product performs with respect to the expected regulation standard." Mr. Ortiz-Soto
also believes some additional results will be helpful when he recommends, "... some
additional results might be helpful for manufacturers to determine if the simulation is
representative of their vehicle. Because many model parameters and vehicle operating
strategies have been standardized using internal assumptions and algorithms, the
overall behavior of the vehicle in question could end up being very different from
what the vehicle manufacturer actually observes. This can result in a significant over-
estimation of fuel consumption and CO2 emissions, and possibly non-compliance. For
this reason, it is fair that the manufacturer be able to assess the validity of the
Page 12 of 104
-------
Kent Helmer
January 28, 2011
Page 12
simulation without having to investigate the model in detail. This could be achieved
by providing a series of additional results, which could be related to the engine
operation over the drive-cycles, the shifting strategy, the electrical system, etc." He
noted that it not practical to have to close each plot in order to see the next one or to
run another simulation; he suggested that a small table with drive output would be
useful to see along with the plots. He suggested that plots of the engine map and
shifting strategy be included, along with various drive-cycle visitation points plotted
on the engine map.
3.6 OTHER COMMENTS
The following subsections contain additional reviewer comments.
3.6.1 "Input" Format
Multiple reviewers suggested improvements to the input boxes to streamline its ease for the user;
suggestions included:
"The coefficient of aerodynamic drag can only be specified with a pull-down list of
values from 0.50 to 0.85, with step 0.05. As a result, not all intermediate values for
Cd can be specified, including the recommended values provided by EPA in Table 5
(e.g. 0.69, 0.76, 0.81 etc.). Considering the significant impact of Cd on fuel economy
and its importance in achieving compliance, the value of Cd should be allowed to be
entered in a textbox."
"... it is not clear why there should be a dropdown menu for the "Coefficient of
Aerodynamic Drag" parameter. Furthermore, the dropdown menu allows the values
to be overwritten by the user, so the dropdown menu has no real purpose... A better
approach would be to just provide a sample value in the parameter name to give the
user an idea of what would be an expected input in the box. Basically, it should look
something like the "Steer Tire RR" and "Drive Tire RR" input boxes."
One reviewer suggests that input boxes should become inactive ("grayed out") when
it is not desirable for those input values to be changed.
"The windows executable version has predefined values for C_d in a dropdown menu
with preset values in increments of 0.02. The C_d value should just be an entry box,
like the C_rr values."
"The inputs for weight reduction, speed limiter, and idle reduction are not consistent
between the matlab version and the windows executable. For example in the matlab
version. In matlab, zero "Weight Reduction" defaults to "N/A," which causes an error
in the windows version. The windows version does not accept "N/A" for idle
reduction."
"The location of the "Vehicle Model Year" dropdown menu is not intuitive. This is
one of the most important parameters of the simulation and it is part of the inputs that
Page 13 of 104
-------
Kent Helmer
January 28, 2011
Page 13
affects the results, but it has been grouped with the identification parameters. These
should be separated as they currently are, but somehow the "Vehicle Model Year"
was left in the top section."
"Having radial buttons with all of the vehicle configurations in the "Regulatory
Class" section is not necessary. It occupies space and reduces the GUI's flexibility to
add other parameters in the future. This type of list is probably better addressed
through the use of a drop down menu. It would reduce the profile of this parameter
list, and it would show much more clearly what vehicle type is being used. Currently,
closer attention has to be paid to the GUI to notice which radio button of the ten
available is selected, whereas with the dropdown menu it is only necessary to read
what is displayed."
3.6.2 Further Validation of the GEM Model
Two reviewers remarked that further validation is needed to ensure confidence of the model
results.
Dr. Midlam-Mohler addresses model validation in remarking, "Based on the issues noted in (2)
[Parameter values for Different Vehicle Classes] above, it is important to validate the model
across vehicle classes. Because the model structure is relatively low-fidelity it has a greater
burden of proof when "extrapolating" results. To have confidence in the model some further
level of validation should be conducted."
Dr. Flowers comments, "It should be confirmed whether the various controllers in the GEM
model are well tuned and result in a vehicle response consistent with empirical data."
3.6.3 Uncertainty/Sensitivity Analysis
One reviewer suggests that a sensitivity analysis should be conducted to better understand the
propagation of error in the input parameters. He recommends that, "It would be useful to have a
better understanding [of] the propagation of error in the input parameters. For the proposed
configuration for the class 8 high-roof sleeper cab the sensitivity of the CC>2 result to errors in Cd
is approximately 50%. This implies that a 10% error in Cd will result in a 5% error in prediction
of CC>2 emissions. For rolling resistance, the impact of a 10% error in the tire rolling resistance
causes a 2.3% error in prediction of CC>2 emissions. These sensitivities should be compared to
the reduction in CC>2 emissions required as well as the accuracy of the key input parameters in
the model. This analysis would also be useful in determining which parameters might be
superfluous with respect to the desired output. As discussed above, there are some models which
likely have more complexity then necessary."
The reviewer concludes, "A rigorous study of the sensitivity of key input parameters should be
conducted. Our ability to measure and estimate input parameters is not perfect, hence, the output
Page 14 of 104
-------
Kent Helmer
January 28, 2011
Page 14
of the model is affected by this uncertainty. If our ability to measure the coefficient of drag is +/-
x.y % then that has an impact on the model output. This uncertainty can then be compared to
required accuracy to make a judgment on the validity of this method at estimating green house
gas emissions or fuel economy."
3.6.4 Complexity, Detail, Depth of Some Parts Seem Unnecessary
A couple reviewers note that they believe the model has a level of detail and complexity that
may be unnecessary for the stated purpose of the model.
Mr. Ortiz-Soto provides a couple of examples of detail that seem unnecessary. He reports,
"Some blocks go into deeper levels unnecessarily. Examples can be found in the electrical
system and in the driver models. Although the approach used in this model of grouping models
into blocks based on their physical components or functionality is fairly intuitive, adding extra
layers can also make the model more difficult to follow if done excessively." Adding to this, he
comments, "Some models, such as the electrical system, appear to be extremely complex and
detailed for this type of dedicated simulation. Unless there is a particular reason, such as future
extensions to GEM for hybrid-electric trucks or different drive-cycles, where such details are
necessary, then the electrical system model can probably be stripped down substantially without
sacrificing much fidelity in the simulation."
Dr. Midlam-Mohler similarly reports, ".. .that there is a higher than necessary level of fidelity in
many of the models." He suggests, "EPA could reduce the complexity of many of the models
with little impact on the accuracy of the simulation - this would then lead to a reduced set of
parameters that v[a]ry with vehicle class and therefore need to be determined." Following up on
this he concludes, "Several of the sub-models had complexity that far outweighed their impact
on the results. The battery was one such sub-model which also contained some serious errors in
its formulation. Many of these models could be simplified which will also reduce the number of
parameters required.
3.6.5 User Guide
One reviewer provides comments on the user guide. The reviewer believes that the model
description, as presented in the user guide, is too detailed and may "generate unnecessary
confusion to the users of GEM." He provides examples of "features of the model that are
irrelevant and outside the scope of GEM, even though these features are present in the model."
4. References
Environmental Protection Agency (EPA). "Greenhouse Gas Emissions Model (GEM) User
Guide." EPA-420-B-10-039, October 2010.
Page 15 of 104
-------
Kent Helmer
January 28, 2011
Page 15
Environmental Protection Agency (EPA). "Draft Regulatory Impact Analysis: Proposed
Rulemaking to Establish Greenhouse Gas Emissions Standards and Fuel Efficiency
Standards for Medium- and Heavy-Duty Engines and Vehicles." EPA-420-D-10-901,
October 2010.
Environmental Protection Agency (EPA) and National Highway Traffic Safety Administration
(NHTSA). "Preamble: Greenhouse Gas Emissions Standards and Fuel Efficiency
Standards for Medium and Heavy-Duty Engines and Vehicles." EPA-HQ-OAR-2010-
0162; NHTSA-2010-0079, October 2010.
Environmental Protection Agency (EPA). "EPA and NHTSA Propose First-Ever Program to
Reduce Greenhouse Gas Emissions and Improve Fuel Efficiency of Medium- and Heavy-
Duty Vehicles: Regulatory Announcement." EPA-420-F-10-901, October 2010.
Page 16 of 104
-------
APPENDICES
Page 17 of 104
-------
Appendix A. Elements to be addressed in the Review of EPA's GEM model
(The model and its documentation can be downloaded from the EPA website at:
http://www.epa.gov/otaq/climate/regulations. Background information, the pre-publication draft
Preamble, Regulations and Regulatory Impact Analysis, can also be found on the same website.)
EPA's vehicle simulation model, GEM, was created to serve as the primary tool to certify Class
7/8 combination tractors and Classes 2b - 8 vocational vehicles in meeting EPA's and NHTSA's
proposed vehicle GHG emission levels and fuel efficiency requirements. As both agencies'
proposed compliance tool, GEM needed the following modeling attributes:
1) capable of modeling a wide array of medium- and heavy-duty vehicles over different
drive cycles;
2) contains open source code, providing transparency in the model's operation;
3) freely available and easy to use by any user with minimal or no prior experience;
4) contains both optional and preset elements; and
5) managed by the Agencies for compliance purposes.
The design of GEM parallels the proposed regulations, which focus on the application of
technologies having the largest impact on reducing vehicle GHG emission reductions or fuel
consumption in the 2014-2017 timeframe. For the given timeframe, the model would allow
various inputs to characterize a vehicle's properties (e.g., weight, aerodynamics, and rolling
resistance) and predict how the vehicle would behave when it to be operated over a particular
driving cycle.
EPA has validated GEM based on the chassis test results from "SmartWay"-certified tractors
tested at the Southwest Research Institute. Since many aspects of one tractor configuration (such
as the engine, transmission, axle configuration, tire sizes, and control systems) are similar to
those used on a manufacturer's sister models, the validation work conducted on these vehicles is
representative of the other Class 8 tractors.
The input values needed for the simulation model (e.g., drag coefficient, tire rolling resistance
coefficients, tire/wheel weight reduction, vehicle speed limiter and extended idle reduction
technologies) are obtained as manufacturer testing or model default values. At the present time,
the agencies are proposing test procedures for generating aerodynamic drag and tire rolling
resistance coefficient inputs. Likewise, the agencies are proposing a range for vehicle speed
limiter and default extended idle reduction technology benefit variables. All other aspects of
vehicle conformation as defined by the agencies are fixed within the model and are not variable
for the purpose of compliance.
After parameters are input to the graphical user interface, GEM predicts the individual and cycle-
weighted fuel consumption and CC>2 emissions for three proposed test cycles - a Transient cycle,
a 55 mph steady-state cruise cycle, and a 65 mph steady-state cruise cycle. The model can also
be used to determine a level of technology necessary for a vehicle to meet a specified GHG
standard and allows a manufacturer to estimate the benefits and costs of those changes to a
Page 18 of 104
-------
particular vehicle for that level of GHG reductions.
In general, EPA is looking for the reviewer's opinion of the concepts and methodologies upon
which the model relies and whether or not the model can be expected to execute these algorithms
correctly. Toward this end, we suggest that reviewers comment on the following items:
1) EPA's overall approach to the stated purpose of the model (meet agencies' compliance
requirements) and whether the particular attributes found in resulting model embodies
that purpose.
2) The appropriateness and completeness of the contents of the overall model structure
and its individual systems, and their component models, if applicable (i.e., using the
MATLAB/Simulink version), such as:
a) The elements of each system to describe different vehicle categories;
b) The performance of each component model, including the reviewer's assessment
of the underlying equations and/or physical principles coded into that component.
c) The input and output structures and how they interface with the model to obtain
the expected result, i.e., fuel consumption and CC>2 over the given driving cycles; and
d) The default values used for the input file, as shown in the GEM User Guide.
3) Using the standard of good engineering judgment, the program execution is optimized
by the chosen methodologies;
4) Clarity, completeness and accuracy of the output/results (CCh emissions or fuel
efficiency output file); and
5) Any recommendations for specific improvements to the functioning or the quality of the
outputs of the model.
In making comments to the model, reviewers should distinguish between recommendations for
clearly defined improvements that can be readily made based on data or literature reasonably
available to EPA and improvements that are more exploratory or dependent on information not
readily available to EPA. Any comment(s) should be sufficiently clear and detailed to allow a
thorough understanding by EPA or other parties familiar with the model. EPA would appreciate
the reviewers not releasing any peer review materials or their comments to the public until the
Agency makes its GEM model and supporting documentation public.
If a reviewer has questions as to what is required to complete this review or needs additional
background materials, please have that person contact the RTI project manager. If a reviewer has
a question about the EPA peer review process itself, please have that person contact Ms. Ruth
Schenk in EPA's Quality Office by phone (734-214-4017) or e-mail schenk.ruth@epa.gov.
Page 19 of 104
-------
Curriculum Vitae: Aris Babajimopoulos
Personal
Degrees
- Ph.D. in Mechanical Engineering, February 2005 (conferred April 2005), The
University of Michigan, Ann Arbor, MI
- M.Sc. in Mechanical Engineering, August 2002, The University of Michigan, Ann
Arbor, MI
- Diploma in Mechanical Engineering, October 1998, Aristotle University of
Thessaloniki, Thessaloniki, Greece
Positions at the University of Michigan
- Assistant Research Scientist, January 2006 - present
- Research Fellow, February 2005 - January 2006
Honors and Awards
- 2006 SAE Excellence in Oral Presentation Award
Teaching
Ph.D. Committees
1. Mr. Michael Mosburger (member, chair: Volker Sick, Mechanical Engineering,
expected 2011)
2. Mr. Jerry Fuschetto (member, chair: Dennis Assanis, Mechanical Engineering,
expected 2011)
3. Dr. Jason Martz (member, chair: Dennis Assanis, Mechanical Engineering, October
2010)
4. Dr. Seung Hwan Keum (member, co-chairs: Dennis Assanis and Hong Im,
Mechanical Engineering, February 2009)
5. Dr. Heejun Park (member, co-chairs: Dennis Assanis and Dohoy Jung, Mechanical
Engineering, February 2009)
M.Sc. Committees
1. Mr. Prasad Shingne ((co-chair with Dennis Assanis, Mechanical Engineering,
August 2010)
2. Mr. Janardhan Kodavasal (co-chair with Dennis Assanis, Mechanical Engineering,
May 2010)
Page 20 of 104
-------
3. Mr. Elliott Ortiz-Soto (co-chair with Dennis Assanis, Mechanical Engineering, May
2010)
4. Mr. Sotiris Mamalis (co-chair with Dennis Assanis, Mechanical Engineering,
December 2009)
5. Mr. Anastasios Amoratis (co-chair with Dennis Assanis, Mechanical Engineering,
June 2009)
6. Mr. Prasad Challa (co-chair with Dennis Assanis, Mechanical Engineering,
December 2008)
ME590 Students Supervised
1. Mr. Chang-Ping Lee (Winter term, 2009)
2. Mr. Sourabh Goel (Winter term, 2009)
UROP Students Supervised
1. Mr. Joshua Busuito (2 credits/term, September 2009 - April 2010)
2. Ms. Christine Siew (3 credits/term, September 2006 - April 2007)
Courses
1. ME 538: Advanced Internal Combustion Engines, Winter semester 2009 (Three
lectures. Also prepared the three homework sets for the class and advised student
teams on their class projects).
Research
Grants and Contracts:
1. "Advanced Combustion Controls - Enabling Systems and Solutions (ACCESS) for
High Efficiency Light Duty Vehicles," funded by the Department of Energy,
National Energy Technology Laboratory, 9/30/2010 - 9/29/2014, total funding
-$25,000,000 including 50% cost share, CO-PI (PI: Hakan Yilmaz, Robert Bosch
LCC).
2. "A University Consortium on Efficient and Clean High Pressure Lean Burn
(HPLB) Engines," funded by the Department of Energy, Office for Energy
Efficiency and Renewable Energy, 9/1/2009 - 12/31/2012, total funding
$3,750,000, CO-PI (PI: Dennis Assanis).
3. "Development of a High Fidelity Vehicle Energy Assessment Simulation and
Implementation in the GT-Drive Framework," funded by the TARDEC FED/CAS SI
Team, 9/1/2009 - 12/31/2010, total funding $525,000, CO-PI with a share of
approximately $50,000 (PI: Dennis Assanis).
4. "Thrust Area 4: JP-8 Engine Combustion Modeling," plus-up project funded by the
Department of Defense (U.S. Army) through the Automotive Research Center,
1/1/2009 - 12/31/2009, $100,000, PI.
5. General Motors-University of Michigan Engine Systems Research Collaborative
Research Lab (GM/UM ESR CRL), Phase 3, funded by General Motors
Page 21 of 104
-------
Corporation, 1/1/2009 to 12/31/2013, total funding $3,750,000, CO-PI with a share
of approximately $450,000 (PI: Dennis Assanis).
5. "Modeling and Experimental Study of the Boosted HCCI Engine," funded by
General Motors Corporation, 7/1/2007 - 12/31/2009, total funding $1,365,000, CO-
PI with a share of approximately $350,000 (PI: Dennis Assanis).
6. "Simulation-Based and Experimental Assessment of Variable Valve Timing
Strategies for HCCI Engines," funded by Borg Warner, 1/1/2007 - 12/31/2009, total
funding $300,000, CO-PI (PI: Dennis Assanis).
7. "University Consortium on Low Temperature Combustion for High-Efficiency,
Ultra-Low Emission Engines," funded by the Department of Energy, 1/1/2006 -
12/31/08, total funding $4,697,000, CO-PI (PI: Dennis Assanis).
8. "Numerical Assessment of a Free Piston Linear Alternator for Series HEV (Hybrid
Electric Vehicle)," funded by General Motors Corporation, 6/1/2005 - 5/31/07, total
funding $525,000, CO-PI (PI: Dennis Assanis).
Publications:
- Articles in Journals and Transactions
1. Ortiz-Soto, E., Assanis, D.N. and Babajimopoulos, A. (2010) A Comprehensive
Engine to Drive-Cycle Modeling Framework for the Evaluation of Future Engine
and Combustion Technologies. Submitted to InternationalJournal of Engine
Research.
2. Keum, S., Park, H., Babajimopoulos, A., Assanis, D.N. and Jung, D. (2010)
Modeling of Heat Transfer in Internal Combustion Engines with Variable Density
Effect. Submitted to InternationalJournal of Engine Research.
3. Martz, J.B., Middleton, R.J., Lavoie, G.A., Babajimopoulos, A. and Assanis, D.N.
(2010) A Computational Study and Correlation of Premixed Isooctane-Air
Laminar Reaction Front Properties under Spark Ignited and Spark Assisted
Compression Ignition Engine Conditions. Accepted for publication in
Combustion and Flame.
4. Mamalis, S., Nair, V., Andruskiewicz, P., Assanis, D. and Babajimopoulos, A.
(UM); Wermuth, N. and Najt, P. (GM) (2010) Comparison of Different Boosting
Strategies for Homogeneous Charge Compression Ignition Engines - A Modeling
Study. SAE International Journal of Engines, vol. 3, no. 1, pp. 296-308
(Presented as SAE Paper 2010-01-0571 at the SAE World Congress, Apr 13-15,
2010, Detroit, MI).
5. Babajimopoulos, A., Challa, P., Lavoie, G.A. and Assanis, D.N. (2010) Model-
based assessment of two variable cam timing strategies for HCCI engines:
Recompression vs. rebreathing. Accepted for publication in the ASME Journal of
Engineering for Gas Turbines and Power (also presented at the ASME Internal
Combustion Engine Division 2009 Spring Technical Conference, May 3-6, 2009,
Milwaukee, WI, as ASME Paper ICES2009-76103).
Page 22 of 104
-------
6. Babajimopoulos, A., Lavoie, G.A., Assanis, D.N. (2007) On the role of top dead
center conditions in the combustion phasing of homogeneous charge compression
ignition engines. Combustion Science and Technology., vol. 179, no. 9, pp. 2039-
2063.
7. Chang, K., Babajimopoulos, A., Lavoie, G.A., Filipi, Z.S., Assanis, D.N. (2006)
Analysis of load and speed transitions in an HCCI engine using 1-D cycle
simulation and thermal networks. 2006 SAE Transactions, Journal of Engines,
vol. 115, no. 3, pp. 621-633 (Presented as SAE Paper 2006-01-1087 at the SAE
World Congress, Apr 3-7, 2006, Detroit, MI).
8. Babajimopoulos, A., Assanis, D.N., Flowers, D.L., Aceves, S.M., Hessel, R.P.
(2005) A fully coupled computational fluid dynamics and multi-zone model with
detailed chemical kinetics for the simulation of premixed charge compression
ignition engines. International Journal of Engine Research, vol. 6, no. 5, pp. 497-
512.
9. Aceves, S.M., Flowers, D.L., Espinosa-Loza, F., Babajimopoulos, A., Assanis,
D.N. (2005) Analysis of premixed charge compression ignition combustion with a
sequential fluid mechanics-multizone chemical kinetics model. 2005 SAE
Transactions, Journal of Engines, vol. 114, no. 3, pp. 252-262 (Presented as SAE
Paper 2005-01-0115 at the SAE World Congress, Apr 11-14, 2005, Detroit, MI).
10. Sjoberg, M., Dec, I.E., Babajimopoulos, A., Assanis, D.N. (2004) Comparing
enhanced natural thermal stratification against retarded combustion phasing for
smoothing of HCCI heat-release rates. 2004 SAE Transactions, Journal of
Engines, vol. 113, no. 3, pp. 1557-1575 (Presented as SAE Paper 2004-01-2994 at
the Powertrain and Fluid Systems Conference and Exhibition, Oct 25-28, 2004,
Tampa, FL).
11. Babajimopoulos, A., Assanis, D.N., Fiveland, S.B. (2002) Modeling the effects of
gas exchange processes on HCCI combustion and an evaluation of potential
control through variable valve actuation. 2002 SAE Transactions, Journal of
Fuels and Lubricants, vol. Ill, no. 4, pp. 1794-1809 (Presented as SAE Paper
2002-01-2829 at the Powertrain and Fluid Systems Conference and Exhibition,
Oct 21-24, 2002, San Diego, CA; also included in Homogeneous Charge
Compression Ignition (HCCI) Engines - Key Research and Development Issues,
SAE PT-94, 2003).
- Articles in Refereed Conference Proceedings
1. Manofsky, L., Vavra, J., Babajimopoulos, A. and Assanis, D. (2010) Bridging the
Gap between HCCI and SI: Spark-Assisted Compression Ignition. Submitted to
the 2011 SAE World Congress.
2. Delorme, A., Rousseau, A., Wallner, T., Ortiz-Soto, E., Babajimopoulos, A. and
Assanis, D. (2010) Evaluation of Homogeneous Charge Compression Ignition
(HCCI) Engine Fuel Savings for Various Electric Drive Powertrains. To be
presented at the 25th World Battery, Hybrid and Fuel Cell Electric Vehicle
Symposium & Exhibition, Nov 5-9, 2010, Shenzhen, China.
Page 23 of 104
-------
3. Shingne, P., Assanis, D. and Babajimopoulos, A. (UM); Keller, P., Roth, D. and
Becker, M. (2010) Turbocharger Matching for a 4-Cylinder Gasoline HCCI
Engine Using a ID Engine Simulation. SAE Paper 2010-01-2143, Presented at
the SAE Powertrains, Fuels and Lubricants Meeting, Oct 25-27, 2010, San Diego,
CA.
4. Mamalis, S. and Babajimopoulos, A. (2010) Model-Based Estimation of
Turbocharger Requirements for Boosting an HCCI Engine. ASME Paper
ICEF2010-35122. Proceedings of the ASME Internal Combustion Engine
Division 2010 Fall Technical Conference, Sep 12-15, San Antonio, TX.
5. Lee, C.-P., Goel, S., and Babajimopoulos, A. (2010) The Effects of Stroke-to-
Bore Ratio on HCCI Combustion. SAE Paper 2010-01-0842, Presented at the
SAE World Congress, Apr 13-15, 2010, Detroit, MI.
6. Hessel, R., Foster, D., Aceves, S., Davisson, M., Espinosa-Loza, F., Flowers, D.,
Pitz, W., Dec, J., Sjoberg, M. and Babajimopoulos, A. (2008) Modeling iso-
octane HCCI using CFD with multi-zone detailed chemistry; Comparison to
detailed speciation data over a range of lean equivalence ratios. SAE Paper 2008-
01-0047, Presented at the SAE World Congress, Apr 14-17, 2008, Detroit, MI.
7. Murotani, T., Hattori, K., Sato, E., Chryssakis, C., Babajimopoulos, A. and
Assanis, D.N. (2007) Simultaneous reduction of NOx and soot in a heavy-duty
diesel engine by instantaneous mixing of fuel and water. SAE Paper 2007-01-
0125, Presented at the SAE World Congress, Apr 16-19, 2007, Detroit, MI.
8. Chang, K., Lavoie, G.A., Babajimopoulos, A., Filipi, Z.S. and Assanis, D.N.
(2007) Control of a multi-cylinder HCCI engine during transient operation by
modulating residual gas fraction to compensate for wall temperature effects. SAE
Paper 2007-01-0204, Presented at the SAE World Congress, Apr 16-19, 2007,
Detroit, MI.
9. Aceves, S.M., Flowers, D.L., Chen, J.-Y., Babajimopoulos, A. (2006) Fast
prediction of HCCI combustion with an artificial neural network linked to a fluid
mechanics code. SAE Paper 2006-01-3298, Presented at the Powertrain and Fluid
Systems Conference and Exhibition, Oct 16-19, 2006, Toronto, Canada.
10. Flowers, D., Aceves, S., Babajimopoulos, A. (2006) Effect of charge non-
uniformity on heat release and emissions in PCCI engine combustion. SAE Paper
2006-01-1363, Presented at the SAE World Congress, Apr 3-7, 2006, Detroit, MI.
11. Babajimopoulos, A., Lavoie, G.A. and Assanis, D.N. (2003) Modeling HCCI
combustion with high levels of residual gas fraction - A comparison of two VVA
strategies. SAE Paper 2003-01-3220, Presented at the Powertrain and Fluid
Systems Conference and Exhibition, Oct 27-30, 2003, Pittsburgh, PA.
12. Babajimopoulos, A., Lavoie, G.A. and Assanis, D.N. (2003) Numerical modeling
of the effects of temperature and composition stratification on HCCI combustion
for high levels of residual gas fraction. Proceedings of the 6th International
Conference on Engines for Automobile, Sep 14-19, 2003, Capri, Italy.
Page 24 of 104
-------
- Patents
1. Assanis, D.N., Babajimopoulos, A., Filipi, Z.S., Kuo, T.-W., Lavoie, G.A., Najt,
P.M., and Rask, R.B. (2009) Hybrid Powertrain System Using Free Piston Linear
Alternator Engines. Patent application filed to the U.S. Patent Office, US Serial
No. 12/504502, filed 16-JUL-2009.
- Book Chapters
1. Aceves, S.M, Flowers, D.L., Dibble, R.W. and Babajimopoulos, A. (2007)
Overview of modeling techniques and their application to HCCI/CAI engines. In
Zhao, H. (Ed.) HCCI and CAI engines for the automotive industry, Chap. 18, pp.
456-474, Woodhead Publishing Limited, Cambridge, England.
- Articles in Non-Refereed Conference Proceedings
1. Hessel, R., Foster, D., Aceves, S., Flowers, D, Pitz, B., Dec, J., Sjoberg, M. and
Babajimopoulos, A. (2007) Modeling HCCI using CFD and detailed chemistry
with experimental validation and a focus on CO emissions. Proceedings of the
17f International Multidimensional Engine Modeling User's Group Meeting, Apr
15, 2007, Detroit, MI.
2. Babajimopoulos, A., Assanis, D.N., Flowers, D.L., Aceves, S.M., Hessel, R.P.
(2005) A fully integrated CFD and multi-zone model with detailed chemical
kinetics for the simulation of PCCI engines. Proceedings of the 15th International
Multidimensional Engine Modeling User's Group Meeting, Apr 10, 2005, Detroit,
MI.
- Presentations in Conferences without Proceedings
1. Babajimopoulos, A., Manofsky, L., Shingne, P., Spater, J., Lavoie, G., and
Assanis, D. (2009) UM FFVA Engine: Experimental Recompression Results and
GT-Power Modeling. HCCI University Working Group Meeting at USCAR,
October 8, 2009, Southfield, MI.
2. Assanis, D.N., Lavoie, G., and Babajimopoulos, A. (2009) Advanced Combustion
for High Efficiency Ultra-Clean Engines. Keynote Lecture, 6* US National
Combustion Meeting, May 17-20, 2009, Ann Arbor, MI.
3. Babajimopoulos, A., Nair, V., Lavoie, G. and Assanis, D. (2009) Exploring
supercharged HCCI using GT-Power based simulation tool. HCCI University
Working Group Meeting at Sandia National Laboratories, February 12, 2009,
Livermore, CA.
4. Babajimopoulos, A., Challa, P., Mamalis, S., Lavoie, G., Filipi, Z. and Assanis,
D. (2008) System modeling of turbocharging for HCCI. HCCI University
Working Group Meeting at USCAR, August 21, 2008, Southfield, MI.
5. Babajimopoulos, A., Lavoie, G., Filipi, Z., Challa, P., Mamalis, S. and Assanis,
D. (2008) System modeling of valve actuation strategies and turbocharging for
HCCI. HCCI University Working Group Meeting at Sandia National
Laboratories, March 20, 2008, Livermore, CA.
Page 25 of 104
-------
6. Babajimopoulos, A., Lavoie, G., Filipi, Z., Challa, P., and Assanis, D. (2007)
System modeling of valve actuation strategies for HCCI with a realistic
combustion model. HCCI University Working Group Meeting at USCAR, October
4, 2007, Southfield, MI.
7. Babajimopoulos, A., Lavoie, G., Filipi, Z., Mo, Y., Chang, K. and Assanis, D.
(2007) A CFD-based HCCI combustion correlation for use in system models.
HCCI University Working Group Meeting at Sandia National Laboratories,
February 8, 2007, Livermore, CA.
8. Babajimopoulos, A., Lavoie, G.A., Assanis, D.N. (2006) On the role of top dead
center conditions in the combustion phasing of homogeneous charge compression
ignition engines. Oral only presentation at the Powertrain and Fluid Systems
Conference and Exhibition, October 16-19, 2006, Toronto, Canada.
9. Chang, K., Lavoie, G, Babajimopoulos, A., Filipi, Z., and Assanis, D. (2006)
Simulation of a multi-cylinder HCCI engine during transient operation by
modulating RGF - Compensating for the wall temperature effects. HCCI
University Working Group Meeting at USCAR, June 29, 2006, Southfield, MI.
10. Babajimopoulos, A., Lavoie, G., Filipi, Z., Im, H., Mo, Y., Chang, K.,
Hamosfakidis, V. and Assanis, D. (2006) HCCI modeling: System modeling of
coolant temperature effect and update on CFD results. HCCI University Working
Group Meeting at Sandia National Laboratories., February 9, 2006, Livermore,
CA.
11. Babajimopoulos, A., Lavoie, G. and Assanis, D. (2005) Scaling of HCCI
combustion phasing: The role of TDC conditions. HCCI University Working
Group Meeting at USCAR, September 15, 2005, Southfield, MI.
12. Babajimopoulos, A., Lavoie, G., Filipi, Z., Mo, Y., Chang, K. and Assanis, D.
(2005) Recent HCCI modeling results and application to system models. HCCI
University Working Group Meeting at Sandia National Laboratories, February 3,
2005, Livermore, CA.
13. Assanis, D., Filipi, Z., Lavoie, G., Babajimopoulos, A., Chang, K. and Mo, Y.
(2004) Modeling HCCI for control and system simulation. HCCI University
Working Group Meeting at USCAR, June 24, 2004, Southfield, MI.
14. Babajimopoulos, A., Lavoie, G., Mo, Y. and Assanis, D. (2004) Developments in
HCCI modeling at the University of Michigan. HCCI University Working Group
Meeting at Sandia National Laboratories, January 29, 2004, Livermore, C A.
15. Assanis, D., Filipi, Z., Lavoie, G., Babajimopoulos, A. and Chang, J. (2003)
Progress in HCCI thermo-kinetic Modeling and engine experiments. HCCI
University Working Group Meeting at USCAR, June 26, 2003, Southfield, MI.
16. Babajimopoulos, A., Assanis, D. and Fiveland, S. (2002) Sequential use of an
open cycle CFD code and a multi-zone model for assessment of VVA control
strategies. HCCI University Working Group Meeting at USCAR, June 12, 2002,
Southfield, MI.
Page 26 of 104
-------
- Invited Seminars
1. Babajimopoulos, A. (2008) An introduction to Homogeneous Charge
Compression Ignition (HCCI). Dept. of Naval Engineering, National Technical
University of Athens, September 22, 2008, Athens, Greece.
2. Babajimopoulos, A. (2008) An introduction to Homogeneous Charge
Compression Ignition (HCCI) and the ongoing work at the University of
Michigan. Graduate Seminar Series, Dept. of Mechanical Engineering, Marquette
University, April 10, 2008, Milwaukee, WI.
Service
Co-organizer for the Kinetically Controlled CI Combustion (including HCCI) session,
SAE 2011 World Congress, April 12-14, 2011, Detroit, MI
Co-organizer for the Kinetically Controlled CI Combustion (HCCI) session, SAE 2010
World Congress, April 13-15, 2010, Detroit, MI
Review coordinator for the Low Temperature Combustion session, ASME Internal
Combustion Engine Division 2009 Fall Technical Conference, September 27-30, 2009,
Lucerne, Switzerland
Co-organizer for the Homogeneous Charge Compression Ignition session, SAE 2009
International Powertrains, Fuels and Lubricants Meeting, June 15-17, 2009, Florence,
Italy
Co-chair for the Multi-dimensional Modeling session, ASME Internal Combustion
Engine Division 2009 Spring Technical Conference, May 3-6, 2009, Milwaukee, WI
Co-organizer for the Homogeneous Charge Compression Ignition session, SAE 2009
World Congress, April 20-23, 2009, Detroit, MI
Judge, UM Engineering Graduate Student Symposium, November, 2006
Reviewer for
1. SAE/JSAE
2. ASME/IMECE
3. The Combustion Institute
4. Transactions of the ASME - Journal of Engineering for Gas Turbines and Power
5. International Journal of Engine Research
6. Proceedings of the Institution of Mechanical Engineers, Part D, Journal of
Automobile Engineering
7. IEEE/ASME Transactions on Mechatronics
8. Combustion and Flame
9. Journal of Energy Resources Technology
Page 27 of 104
-------
10. Combustion Science and Technology
11. Energy
Page 28 of 104
-------
Daniel L. F'lowers
Brief Biography
Daniel Flowers is the Associate Program Leader for Combustion and Alternative Fuels at
Lawrence Livermore National Laboratory, where his work focuses on experimental and
analytical research in thermal sciences and combustion. He has been working in the
area of Homogeneous Charge Compression Ignition (HCCI) engine combustion since
joining LLNL in 1998. Flowers leads several combustion research projects at LLNL in the
areas of HCCI, hydrogen and Diesel combustion. On leave from LLNL Flowers led
research and development at Cleeves Engines, an energy research startup company.
Flowers served as Associate Technical Editor of the ASME Journal of Energy Resource
Technologies in 2007 and 2008. Flowers holds Ph.D. (2001), M.S. (1997), and B.S.
(1996) degrees in Mechanical Engineering from the University of California, Davis.
Work History
Lawrence Livermore National Laboratory, September 1998 to present
Title: Principal Investigator/Project Leader
Responsibilities:
Principal Investigator- DOE OFCVT Combustion and Fuels Programs ($1M/FY07)
o Leading Ongoing LLNL activities in HCCI research, developing multidimensional
modeling tools
o Leading collaborations with Universities and Other National Labs
o Program highly ranked at annual program review
o Integral part of a world recognized team that has developed the most advanced
analysis tools for HCCI combustion
o Extending analysis tools for HCCI combustion - Continuing to advance
multidimensional HCCI combustion modeling tool
o Continuing development of massively parallel tool for simulation of
multidimensional HCCI and PCCI combustion
o Collaborating with US auto industry partners to guide development of new
combustion systems
o Investigating HCCI applications for biofuels and non-standard fuels: biodiesel,
"wet ethanol," "trash gas"
Principal Investigator - DOE NETL ($300K/FY07)
o Separate project on HCCI working with International Engine Company funded by
NETL
o Modeling to support International Engine's HCCI Engine Development program
Principal Investigator- DOE/OFCVT
o Modeling of Hydrogen Spark Ignition Combustion ($150K FY06)
o Modeling of Smokeless Rich Diesel Combustion ($150K FY06)
Principal Investigator/Project Leader - HCCI engines for stationary power generation
(California Energy Commission, 3 years $2M)
o Leading development of an experimental HCCI engine for stationary power
generation applications
o Completed 2006
Cleeves Engines Incorporated (San Carlos, CA), February 2008 to June 2009
Title: Senior Combustion Engineer
Responsibilities:
Leading Research and Development activities on an advanced technology concept.
Concept development advanced operating strategies for an advanced internal
combustion engine strategy
Developing test cell hardware, methods, and protocols for demonstration of engine
-1-
Page29of 104
-------
Daniel L. F'lowers
concept
Guiding and conducting numerical analysis activities for prototype engine
development, including CFD (Fluent), Engine dynamic modeling (GT-Power), and
FEA (Cosmos, Ansys)
Cleeves Engines is an Energy Technology startup developing advance internal
combustion engine concepts
Education
Ph.D in Mechanical Engineering - University of California, Davis
Dissertation: Combustion in Homogeneous Charge Compression Ignition Engines:
Experiments and Detailed Kinetic Modeling
M.S. in Mechanical Engineering - University of California, Davis
Thesis: Application of Morphology Dependent Resonance Spectroscopy to Droplet
Sizing
B.S. in Mechanical Engineering - University of California, Davis (Highest Honors)
Professional Activities
Associate Technical Editor, ASME Journal of Energy Resources Technology
The Combustion Institute, Member (Alternate on Executive Committee)
Society of Automotive Engineers, Member
American Society of Mechanical Engineers, Member
Symposium Co-Chair, Advanced Energy Systems, ASME IMECE, 2004
Session Organizer and Chair, Advanced Energy Systems, ASME IMECE, Multiple years
Session Chair, Society of Automotive Engineers, Multiple years
Mentoring and Education
Mentor to LLNL Graduate Research Fellows and LLNL Graduate Student Employees
Long-standing collaboration with Profs. Robert Dibble and J.Y. Chen at UC Berkeley.
Direction and Research Guidance to UC Berkeley Graduate Students.
Mentor and project leader to several LLNL Undergraduate Engineering Interns.
Invited Mini-course, Universidad de Guanajuato, Mexico: Introduction to KivaSv.
Patent
Daniel L. Flowers, "Controlling and Operating Homogeneous Charge Compression
Ignition (HCCI) Engines," U.S. Patent 6,923,167
Book Chapter
S. M. Aceves, D. L. Flowers, R. W. Dibble and A. Babajimopoulos, "Overview of
modeling techniques and their application to HCCI/CAI engines," in HCCI and CAI
engines for the automotive industry, Hua Zhao, Ed., in press.
Peer Reviewed Publications
Journal Papers
Killingsworth, N.J., Aceves, S.M., Flowers, D.L., Espinosa-Loza, F.J., and Kristic, M.,
"HCCI Engine Combustion Timing Control: Optimizing Gains and Fuel Consumption Via
Extremum Seeking," IEEE Transactions on Control Systems Technology, in press
(2008).
-2-
Page 30 of 104
-------
Daniel L. F'lowers
D.L. Flowers, S.M. Aceves, R.W. Dibble, "Effect of Laser-induced Excitation of Oxygen
on Ignition in HCCI Engines Analyzed by Numerical Simulations," Combustion Theory
And Modeling, Vol. 11, No. 3, 2007: 455-468.
J.H. Mack, B.A. Buchholz, D.L. Flowers and R.W. Dibble. "Using Biofuel Tracers to
Study Alternative Combustion Regimes," Nuclear Instruments & Methods B (2006) in
press.
Aristotelis Babajimopoulos, Dennis Assanis, Daniel Flowers, Salvador Aceves, Randy
Hessel"A Fully Integrated CFD and Multi-zone Model with Detailed Chemical Kinetics for
the Simulation of PCCI Engines" International Journal of Engine Research, Volume 6,
Number5, October 2005, pp. 497-512(16).
Parag Mehresh, Daniel Flowers, Robert Dibble, "Experimental and Numerical
Investigation of Effect of Fuel on Ion Sensor Signal to Determine Combustion Timing in
HCCI Engines," International Journal of Engine Research, Volume 6, Number 5, October
2005, pp. 465-474(10).
Parag Mehresh, Jason Souder, Daniel Flowers, Uwe Riedel, Robert Dibble,
"Combustion Timing in HCCI Engines Determined by Ion-Sensor: Experimental and
Kinetic Modeling," Proceedings of the Combustion Institute, Vol 30, Part 2, 2005: 2693-
2700.
Hunter Mack, Bruce Bucholtz, Daniel Flowers, Robert Dibble, "Investigation of HCCI
Combustion of Diethyl Ether and Ethanol Mixtures Using Carbon 14 Tracing and
Numerical Simulations," Proceedings of the Combustion Institute, Vol 30, Part 2, 2005:
2701-2709.
Martinez-Frias J, Aceves SM, Flowers D, Smith JR, Dibble R, "Thermal charge
conditioning for optimal HCCI engine operation"ASME Vol. 124 No. 1, March 2002: 67-
75.
Flowers, D. L, Aceves, S.M., Martinez-Frias, J., and Dibble, R. W., "Prediction of
Carbon Monoxide and Hydrocarbon Emissions in Isooctane HCCI Engine Combustion
Using Multi-Zone Simulations," Proceedings of the Combustion Institute, Vol 29, Part 1,
2002: 687-694.
Flowers, D. L., Aceves, S. M., Westbrook, C. K., Smith, J. R., Dibble, R. W., "Detailed
Chemical Kinetics Simulation of HCCI Combustion Gas Composition Effects and
Investigation of Control Strategies," ASME Journal of Engineering for Gas Turbines and
Power, Vol. 123, No. 2, April 2001.
Santangelo, P. J., Flowers, D. L., and Kennedy, I. M., "Demonstration of droplet size
and vaporization rate measurements in the near field of a two-phase jet with droplet
lasing spectroscopy," Applied Optics, 1998 Aug 20, Vol 37 No. 24: 5573-5578.
SAE Transactions Papers
Hessel, R.P, Aceves, S.M., Flowers, D.L., "A Comparison on the Effect of Combustion
Chamber Surface Area and In-Cylinder Turbulence on the Evolution of Gas Temperature
Distribution from IVC to SOC: A Numerical and Fundamental Study," SAE Paper 2006-
01-0869 SAE Transactions, Journal of Engines, 2006.
-3-
Page31 of 104
-------
Daniel L. F'lowers
Aceves, S.M., Flowers, D.L., "A Detailed Chemical Kinetic Analysis of Low-
Temperature, Non-Sooting Diesel Combustion," SAE Paper 2005-01-0923, SAE
Transactions, Journal of Engines, 2005.
Aceves, S.M., Flowers, D. L., Espinosa-Loza, F.J., Babajimopoulos, A., Assanis, D.,
"Analysis of Premixed Charge Compression Ignition Combustion With a Sequential Fluid
Mechanics-Multizone Chemical Kinetics Model," SAE Paper 2005-01-0115, SAE
Transactions, Journal of Engines, 2005.
Salvador Aceves, Daniel Flowers, "Analysis of the Effect of Geometry Generated
Turbulence on HCCI Combustion by Multi-Zone Modeling" SAE Paper 2005-01-2134,
SAE Transactions, Journal of Engines, 2005.
Parag Mehresh, Daniel Flowers, Robert Dibble, "EGR effect on Ion Signal in HCCI
Engines," SAE Paper 2005-01-2126, SAE Transactions, Journal of Engine, 2005.
Hunter Mack, Bruce Bucholtz, Daniel Flowers, Robert Dibble, "Effect of the Di-Tertiary
Butyl Peroxide (DTBP) additive on HCCI Combustion of Fuel Blends of Ethanol and
Diethyl Ether" SAE Paper 2005-01-2135, SAE Transactions, Journal of Fuels and
Lubricants, 2005.
Salvador M. Aceves, Daniel L. Flowers, Francisco Espinosa-Loza, Joel Martinez-Frias,
John E. Dec, Magnus Sjoberg, Robert W. Dibble and Randy P. Hessel, "Spatial Analysis
of Emissions Sources for HCCI Combustion at Low Loads Using a Multi-Zone Model,"
SAE Paper 2004-01-1910, SAE Transactions, Journal of Fuels and Lubricants, 2004.
Daniel L. Flowers, Salvador M. Aceves, Joel Martinez-Frias, Randy Hessel, and Robert
W. Dibble, "Effects of Mixing on Hydrocarbon and Carbon Monoxide Emissions,
Predictions for Isooctane HCCI Engine Combustion Using a Multi-Zone Detailed Kinetics
Solver," SAE Paper 2003-01-1821, SAE Transactions, Journal of Fuels and Lubricants,
2003.
Salvador M. Aceves, Daniel L. Flowers, Francisco Espinosa-Loza, Joel Martinez-Frias,
Robert W. Dibble, Magnus Christensen, Bengt Johansson, Randy P. Hessel, "Piston-
Liner Crevice Geometry Effect on HCCI Combustion by Multi-Zone Analysis," SAE
Paper 2002-01-2869, SAE Transactions, Journal of Engines, Volume 111, pp. 2691-
2698, 2002.
Aceves, S.M., Martinez-Frias, J., Flowers, D. L., Smith, J. R., Dibble, R. W., "A
Decoupled Model of Detailed Fluid Mechanics Followed By Detailed Chemical Kinetics
for Prediction of Iso-Octane Hcci Combustion, " SAE Paper 2001-01-3612, SAE
Transactions, Journal of Fuels and Lubricants, 2001.
Aceves, S. M., Flowers, D. L., Martinez-Frias, J., Smith, J. R., Westbrook, C. K., Pitz,
W. J., Dibble, R. W. "Multi-Zone Analysis of Propane HCCI Combustion," SAE Paper
2001-01-1027, SAE Transactions, Journal of Engines, 2001.
Martinez-Frias, J. M., Aceves, S. M., Flowers, D. L., Smith, J. R., Dibble, R. W., "HCCI
Engine Control by Thermal Management," SAE Paper 2000-01-2869, SAE Transactions,
Journal of Engines, 2000.
.4.
Page 32 of 104
-------
Daniel L. F'lowers
Aceves, S.M, Flowers, D. L., Westbrook, C. K., Smith, J. R., Pitz.W., Dibble, R. W.,
Christensen, M., and Johansson, B., 2000, "A Multi-Zone Model for Prediction of HCCI
Combustion and Emissions," SAE paper 2000-01-0327, SAE Transactions, Journal of
Engines, 2000.
Other SAE Papers
Hessel, R.P., Foster, D., Steeper, R., Aceves, S.M., Flowers, D.L., "Pathline Analysis of
Full-cycle Four-stroke HCCI Engine Combustion Using CFD and Multi-Zone Modeling,"
SAE Paper 2008-01-0048.
Hessel, R., Babajimopoulos, A., Foster, D., Aceves, S., Davisson, M, Espinosa-Loza,
F.J., Flowers, D.L., Pitz, W., Dec, J., Sjoberg, M.," Modeling Iso-octane HCCI using
CFD with Multi-Zone Detailed Chemistry; Comparison to Detailed Speciation Data over a
Range of Lean Equivalence Ratios, 2008-01-0047.
Flowers, D.L., Aceves, S.M, Martinez-Frias, J., "Improving Ethanol Life Cycle Energy
Efficiency by Direct Utilization of Wet Ethanol in HCCI Engines," SAE Paper 2007-01-
1867/JSAE Paper 20077037.
S.M. Aceves, D.L.Flowers, J.Y. Chen, A. Babajimopoulos, "Fast Prediction of HCCI
Combustion With an Artificial Neural Network Linked to a Fluid Mechanics," SAE Paper
2006-01-3298.
R.P. Hessel, N. Abani, S. Aceves, D. Flowers, "Gaseous Fuel Injection Modelling Using
a Gaseous Sphere Injection Methodology," SAE Paper 2006-01-3265.
Flowers, D.L., Aceves, S.M., Babajimopoulos, A., "Effect of Charge Non-uniformity on
Heat Release and Emissions in PCCI Engine Combustion," SAE Paper 2006-01-1363.
Salvador M. Aceves, Daniel Flowers, Joel Martinez-Frias, Francisco Espinosa-Loza,
William J. Pitz, Robert Dibble, "Fuel and Additive Characterization for HCCI
Combustion," SAE paper 2003-01-1814.
Salvador M. Aceves, Joel Martinez-Frias, Daniel Flowers, J. Ray Smith, Robert Dibble,
J.Y. Chen, "A Computer Generated Reduced Iso-Octane Chemical Kinetic Mechanism
Applied to Simulation of HCCI Combustion," SAE Paper 2002-01-2870.
James W. Girard, Robert W. Dibble, Daniel L. Flowers, Salvador M. Aceves, "An
Investigation of the Effect of Fuel-Air Mixedness on the Emissions from an HCCI
Engine," SAE Paper 2002-01-1758.
Martinez-Frias, J., Aceves, S.M., Flowers, D. L., Smith, J. R., Dibble, R. W.,
"Equivalence Ratio-Egr Control of Hcci Engine Operation and the Potential for Transition
to Spark-Ignited Operation, " SAE Paper 2001-01-3613.
Flowers, D. L., Aceves, S. M., Martinez-Frias,J., Smith, J. R., Au, M. Y., Girard, J. W.,
Dibble, R. W., 2001, "Operation of a Four-Cylinder 1.9 L Propane Fueled Homogeneous
Charge Compression Ignition Engine," SAE Paper 2001-01-1895.
-5-
Page33of104
-------
Daniel L. F'lowers
Au, M., Girard, J. W., Dibble, R. W., Seibel, C., Maas, U., Aceves, S. M., Flowers, D. L.,
Martinez-Frias, J., Smith, J. R., 2001, "Four-Cylinder HCCI Engine Operation with
Exhaust Gas Recirculation," SAE Paper 2001-01-1894.
Flowers, D. L., Aceves, S. M., Smith, J. R., Torres, J., Girard, J., and Dibble, R. W.,
"HCCI in a CFR Engine: Experiments and Detailed Kinetic Modeling," SAE paper 2000-
01-0328.
Aceves, S. M., Smith, J. R., Perkins, L. J., Haney, S. W., Flowers, D. L., "Optimization
of a CNG Series Hybrid Concept Vehicle," SAE paper 960234.
Other Conference Papers
Killingsworth NJ, Aceves SM, Flowers DL, Krstic M. "A simple HCCI engine model for
control." IEEE Conference on Computer Aided Control System Design, 2006 IEEE
International Conference on Control Applications, 2006 IEEE International Symposium
on Intelligent Control. IEEE. 2006, pp. 6.
Walther, D.C., Fernandez-Pello, A.C., Dibble, R., Aceves, S.M., Flowers, D. "The use of
hydrogen combustion for power generation" 3rd International Energy Conversion
Engineering Conference, v 3, Collection of Technical Papers - 3rd International Energy
Conversion Engineering Conference, 2005, p 1919-1938.
Flowers, Daniel L., Martinez-Frias, Joel, Espinosa-Loza, Francisco, Killingsworth, Nick,
Aceves, Salvador M., Dibble, Robert, Kristic, Miroslav, Bining, Avtar, "Development and
testing of a 6-cylinder HCCI engine for distributed generation," Proceedings of the 2005
Fall Technical Conference of the ASME Internal Combustion Engine Division, 2005, p
643-651.
Martinez-Frias, Joel, Flowers, Daniel, Aceves, Salvador M., Espinosa-Loza, Francisco,
Dibble, Robert, "Thermal management fore-cylinder HCCI engine: Low cost, high
efficiency, ultra-low NOx power generation," Proceedings of the 2004 Fall Technical
Conference of the ASME Internal Combustion Engine Division, 2004, p 833-839.
Martinez-Frias, Joel, Aceves, Salvador M., Flowers, Daniel, Smith, J. Ray, Dibble,
Robert, "Exhaust energy recovery for control of a homogeneous charge compression
ignition engine," American Society of Mechanical Engineers, Advanced Energy Systems
Division (Publication) AES, v 40, 2000, p 349-356.
Santangelo, P.J. (Univ of California Davis); Flowers, D.; Kennedy, I.M. "Measurements
of droplet size in the near field of a droplet laden jet using MDR spectroscopy," Chemical
and Physical Processes in Combustion, Fall Technical Meeting, The Eastern States
Section, 1997, p 265.
Thesis and Dissertation
Flowers, D. L, "Combustion in Homogeneous Charge Compression Ignition Engines:
Experiments and Detailed Chemical Kinetic Simulations" Ph.D. Dissertation, University
of California, Davis, 2001
Flowers, D. L., "Application of Morphology Dependent Resonance Spectroscopy to
Droplet Sizing" Masters Thesis, University of California, Davis, 1997.
-6-
Page 34 of 104
-------
CURRICULUM VITAE
SHAWN W. MIDLAM-MOHLER, PH.D.
3938NorbrookDr.
Columbus, Ohio 43220
(614) 307-4176
midlam-mohler. I
EDUCATION
Engineering Education
Ph.D. Mechanical Engineering 6/2005
The Ohio State University Columbus, OH
Dissertation Title: "Modeling, Control, and Diagnosis of a Diesel Lean NOX Trap Catalyst"
M.S. Mechanical Engineering 3/2001
The Ohio State University Columbus, OH
Thesis Title: "A Novel Fuel-Operated Heater for Automotive Thermal Management"
B.S. Mechanical Engineering Summa cum Laude 6/1999
Wright State University Dayton, OH
Senior Design Project: "Aerodynamic Design and Simulation of a Wind-Turbine"
Academic Fellowships
Graduate Automotive Technology Education Program - Ph.D. Studies Source: Dept. of Energy
Awarded to select graduate students conducting research supporting DOE goals for transportation research
University Fellowship - M.S. Studies Source: Ohio State University
Awarded in a university-wide search to attract high-caliber graduate students
RESEARCH I
EXPERIENCE I
Research Appointments
Research Scientist 10/2008 to present
Ohio State University Center for Automotive Research, Columbus, OH
Conduct research in the area of clean and efficient transportation, including emissions reduction, Diesel
engines, alternative combustion, hydrogen generation, heavy fuel atomization, and advanced powertrains
Directed and advised graduate students in this area of research
Senior Research Associate 11/2005 to 9/2008
Ohio State University Center for Automotive Research, Columbus, OH
Conducted research in the area of clean and efficient transportation
Directed and advised graduate students in this area of research
Research Associate II 2/2004 to 10/2005
Ohio State University Center for Automotive Research, Columbus,
Conducted research in the area of clean and efficient transportation
Page 35 of 104
-------
Research Intern 6/2003 to 9/2003
Ford Scientific Research Labs, Dearborn, MI
Conducted research on emissions reductions for gasoline hybrid-electric vehicles
Three-month assignment resulted in three Ford invention disclosures and two U.S. patents
Research Funding
As a PI or co-Pi, Dr. Midlam-Mohler has averaged over a hah0 million dollars in research per year since 2005.
These projects are identified in the following sections.
Projects as PI / Co-Pi:
$50,000/1 years
$40,000/0.5 years
$99,000/2 year
$2,000,000/3 years1
$943,108/4 years
$724,531/3 years
$234,760/2 years
$673,550/3 years
Title: Analysis of Secondary Powertrain Systems in HEVs Start: 10/2009
Source: CAR Industrial Consortium Role: PI
Title: Life Cycle Analysis of Landfill Derived Natural Gas Start: 4/2009
Source: FirmGreen Role: PI
Title: Fleet Studies of Plug-In Electric Hybrid Vehicles Start: 1/2009
Source: SMART@CAR Consortium Role: PI
Title: EcoCAR Challenge Hybrid Electric Vehicle Project Start: 6/2008
Source: US Department of Energy and numerous other sponsors Role: Co-Pi
Title: Coordinated Diesel Engine and Aftertreatment Control Start: 4/2008
Source: Cummins Role: PI
Title: Hierarchical Approach to Engine Modeling Start: 4/2007
Source: General Motors Role: Co-Pi
Title: Soot Filter Regeneration though External Heat Addition Start: 11/2005
Source: Tenneco Automotive Role: Co-Pi
Title: On-Board Fuel Reformation for Diesel Aftertreatment Start: 11/2005
Source: Tenneco Automotive Role: Co-Pi
Minor Projects as PI/co-PI:
$45,000
$22,500
Miscellaneous small projects
Source: Hi-Stat, Henkel
Miscellaneous small projects
Source: National Energy Technology Lab, Nextech Materials
Projects with Major Research Role (not co-PD:
$940,863/4 years
$1,327,954/5 years
Title: Next Generation Charge Estimation for 1C Engines
Source: General Motors
Title: Next Generation APR Control for 1C Engines
Source: General Motors
2009
2008
Start: 7/2004
Role: Researcher
Start: 7/2004
Role: Researcher
1 This is the estimated cost of the research conducted under this problem if funded from an external sponsor. This
project is heavily leveraged by the Department of Energy, General Motors, Ohio State University, and a number of
other sponsors through in-kind contributions as well as direct funding and fellowships.
Page 36 of 104
-------
TEACHING I
EXPERIENCE I
Instructional Appointments
Adjunct Assistant Professor 7/2009 to present
Ohio State University Department of Mechanical Engineering, Columbus, OH
Granted in recognition of significant educational service to the Mechanical Engineering Department
Service includes one-on-one student advising, student project advising, and supervision of undergraduate
research
Instructor 4/2007 to present
Ohio State University Department of Mechanical Engineering, Columbus, OH
Sole instructor of record for two applied thermal and fluids courses on internal combustion engines
Course Development
ME 631 - Powertrain Laboratory (3 CR) 1/2009
Ohio State University Department of Mechanical Engineering, Columbus, OH
Developed course material for two quarter hours of classroom lecture which reinforced lab work
Developed eight new lab experiments based on in-depth knowledge of the automotive industry
Facilitated donation of a gasoline engine from General Motors and a Diesel engine from Cummins, both
with a calibration system to provide students access to cutting-edge equipment
ME 730 - Internal Combustion Engine Modeling (3 CR) 4/2007
Ohio State University Department of Mechanical Engineering, Columbus, OH
Developed all new lecture material to bring in personnel research experience
Developed new homework assignments to better engage students by building a fully functioning engine
model in stages of greater fidelity and complexity
Facilitated the donation of industry-standard engine simulation software for use by students
Developed capstone project which allowed students to become engaged in a topic of interest
Seminar - Alternative Fuels Short Course 1/2007
Ohio State University Center for Automotive Research Distance Education Program
Developed 10 hours of lecture and lecture notes for industrial distance education program
Provided case studies of alternative-fueled vehicles to reinforce concepts for the industry audience
Teaching Experience
ME 631 - Powertrain Laboratory (3 CR) Sole Instructor of Record 1/2010
Overall Teaching Rating: 5.0/5.0 Class Size: 15
ME 730 - Internal Combustion Engine Modeling (3 CR) Sole Instructor of Record 4/2009
Overall Teaching Rating: 4.4/5.0 Class Size: 7
ME 631 - Powertrain Laboratory (3 CR) Sole Instructor of Record 1/2009
Overall Teaching Rating: 4.8/5.0 Class Size: 12
ME 730 - Internal Combustion Engine Modeling (3 CR) Sole Instructor of Record 4/2007
Overall Teaching Rating: 4.5/5.0 Class Size: 8
Page 37 of 104
-------
Academic Advising
Since 2005, Dr. Midlam-Mohler has become increasingly involved in student advising. He has served in an
advisory or supervisory capacity to the following students at the M.S. and Ph.D. level:
Student
Quiming Gong
Bernhard Grimm
John Davis
Jason Meyer
Katherine Bovee
John Davis
Ryan Everett
Kenny Pollen
Beth Bezaire
Brad Cooley
Chris Hoops
Ming Fang
Chris Hoops
Dave Ortiz
Rajaram Maringanti
Joshua Supplee
Adalbert Wolany
Sai Rajagopalan
Sergio Hernandez
Andrea Pezzini
Patrick Rebechi
Rhisee Bhatt
Simone Bernasconi
Josh Cowgill
Kenny Pollen
Courtney Coburn
Adam Vosz
Eric Snyder
Role
Research Supervisor
Research Supervisor
Co-Advisor
Research Supervisor
Acting Advisor
Acting Advisor
Acting Advisor
Research Supervisor
Acting Advisor
Acting Advisor
Acting Advisor
Acting Advisor
Acting Advisor
Supervisor
Acting Advisor
Acting Advisor
Supervisor
Committee Member
Acting co-advisor
Supervisor
Supervisor
Acting co-advisor
Supervisor
Acting co-advisor
Acting co-advisor
Acting Advisor
Acting Advisor
Acting co-advisor
Graduation Date or
Expected Graduation Date
2012
2010
2011
2011
2010
2010
2010
2010
2010
2010
2010
2009
2009
2009
2009
2009
2009
2009
2008
2008
2008
2007
2007
2007
2007
2006
2006
2005
Undergraduate Student Research Assistants:
Dr. Midlam-Mohler has supervised the following students on research outside of a formal degree program:
Student
John Macauley
Alixandra Keil
Jennifer Loy
Sean Ewing
David Griffin
Role
Supervisor
Supervisor
Supervisor
Supervisor
Supervisor
Year
2009-10
2009-10
2009-10
2009
2009
Page 38 of 104
-------
B.S. Ross Wang Supervisor 2009
B.S. Orlando Inoa Supervisor 2008-09
B.S. Al Godfrey Supervisor 2008-09
B.S. JohnLutz Supervisor 2008
B.S. KonradSvzed Supervisor 2008
B.S. Joshua Supplee Supervisor 2007
Mentor for Local High School Students
Dr. Midlam-Mohler has mentored six local high school students for 30 hours of activity per student since 2007.
Student Organization Advising
EcoCAR Challenge Hybrid Electric Vehicle Team 6/2008 - present
Ohio State University
Co-advise 40 member (-80% undergraduate) student design project team competing in U. S. Department
of Energy sponsored vehicle competition
Oversee day-to-day operation of team as they design, build, and test a hybrid electric SUV
Team won 1st place in first year, 4th place in second year
Nominated by team for "NSF Advisor of the Year Award"
Challenge-X Hybrid Electric Vehicle Team 8/2006 - 6/2008
Ohio State University
Co-advised primarily undergraduate team competing in Department of Energy Sponsored advanced
technology vehicle completion
Over the course of the four year competition from 2004 - 2008, OSU placed 3rd, 4th, 4th, and 3rd
respectively in the premier advanced technology vehicle competition
Professional Development and Service - Education
Lecturer for Groups Touring the Ohio State Center for Automotive Research 1/2007 - present
Ohio State University Center for Automotive Research
Provide 30-60 minute presentation and discussion on topic of energy use in transportation to groups
Reached over 500 individuals including elementary school students, college students, and community
groups
OSU Continuing Education Program, Columbus, OH 9/2010
Presenter
Provided one hour seminar to practicing engineers on green vehicle design
Lilly Conference on College Teaching 11/2009
Miami University Teaching Conference
Attended three day conference on college teaching
Teaching at Ohio State Orientation 9/2009
Ohio State University Faculty Development Workshop
Attended the following seminars over the course of two and a half days: Introduction to Teaching and
Learning; Fair and Efficient Grading; Designing Assignments Quizzes, and Tests; and Seven Habits of
Effective Teachers - Universal Design for Learning; and Developing Effective Presentation Skills
OSU Continuing Education Program, Columbus, OH 9/2009
Presenter
Provided one hour seminar to practicing engineers on green vehicle design
Page 39 of 104
-------
Ohio State Mechanical Engineering Curriculum Development Retreat, Columbus, OH 7/2009
Participant in Design Focus Group
Requested by Dept. Chair to serve in Design Focus Group
Participated with faculty colleagues and alumni to evaluate engineering curriculum as OSU
Summer Institute on Course Design 6/2009
Ohio State University Faculty Development Workshop
Attended 15 hour, hands-on seminar on effective course design
Learned structured techniques for developing courses
Defined course goals, learning objectives, course content, and methods of assessment for a course
PROFESSIONAL I
SERVICE I
Professional Service
Clean Fuels Ohio, Columbus, OH 9/2009 to present
Member of the Board of Directors
Elected to Board of Directors of Clean Fuels Ohio, a non-profit committed to cleaner transportation fuels
State of Indiana 4/2009
Proposal Reviewer
Reviewed multi-million dollar proposal for Indiana grant program in area of internal combustion engines
Natural Gas Fleet Stakeholders Meeting, Grove City, OH 11/2008
Panel Member
Served as panel technical expert on alternative vehicular fuels
Meeting attended by designees' from the Governor's office and from both of Ohio's U.S. Senators' staff
McMaster Fuel Ltd., Perrysburg, OH 9/2006 to 1/2007
Independent Consultant
Provided analysis of a hydrogen production technique against other methods of hydrogen production
Provided analysis of these techniques for emissions reduction
Assisted McMaster Fuel Ltd. in making strategic decisions regarding their technology
Publication Reviewer Continuous
Review numerous publications for conferences and journal submission of ASME, SAE, IEEE, etc.
PUBLICATIONS
Scholarly Publications
Journal Articles:
1. J. Meyer, S. Yurkovich, S. Midlam-Mohler, "An Approach for Cylinder Specific APR Prediction," in
preparation for submission to ASME Journal of Dynamic Systems, Measurements, and Controls.
2. S. Midlam-Mohler, R. Maringanti, M. Fang, "Inverse-Distance Interpolation Methods for Diesel Engine
Combustion Control," in preparation for submission to ASME Journal of Dynamic Systems, Measurements,
and Controls.
Page 40 of 104
-------
3. M. Canova, S. Midlam-Mohler, P. Pisu, A. Soliman, "Model-Based Fault Detection and Isolation for a Diesel
Lean NOx Trap Aftertreatment System," Control Engineering Practice, November 2009.
4. M. Canova, S. Midlam-Mohler, Y. Guezennec, G. Rizzoni, "Mean Value Modeling and Analysis of HCCI
Diesel Engines with External Mixture Formation," ASME Journal of Dynamic Systems, Measurement and
Control, Vol. 131, No. 11, 2009.
5. M. Canova, S. Midlam-Mohler, Y. Guezennec, G. Rizzoni, "Theoretical and Experimental Investigation on
Diesel HCCI Combustion with External Mixture Preparation," International Journal of Vehicle Dynamics,
Volume 44, Nos 1-2, 2007.
6. N. Szabo, C. Lee, J. Trimbolil, O. Figueroa, R. Ramamoorthy, S. Midlam-Mohler, A. Soliman, H. Verweij, P.
Dutta and S. Akbar, "Ceramic-Based Chemical Sensors, Probes and Field-Tests in Automobile Engines,"
Journal of Materials Science, November, 2003.
Conference Papers:
1. Qi. Gong, S. Midlam-Mohler, V. Marano, G. Rizzoni, Y. Guezennec, "Statistical analysis based PHEV fleet
data study", 2010 IEEE Vehicle Power and Propulsion Conference, September, 2010.
2. Kerem Bayar, Beth Bezaire, Brad Cooley, John Kruckenberg, Eric Schact, Shawn Midlam-Mohler, Giorgio
Rizzoni, "Design of an Extended-Range Electric Vehicle for the EcoCAR Challenge", ASME 2010
International Design Engineering Technical Conference, August, 2010.
3. J. Meyer, S. Yurkovich, S. Midlam-Mohler, "An APR Control Architecture Comparison: Phase Lock Loop
Versus Duty Cycle Control," 2010 American Controls Conference, June, 2010.
4. R. Maringanti, S. Midlam-Mohler, M. Fang, F. Chiara, M. Canova, "Set-Point Generation using Kernel-Based
Methods for Closed-Loop Combustion Control of a CIDI Engine," ASME DSCC2009, September, 2009.
5. J. Meyer, S. Rajagopalan, S. Midlam-Mohler, Y. Guezennec, S. Yurkovich, "Application of an Exhaust
Geometry Based Delay Prediction Modal to an Internal Combustion Engine," ASME DSCC2009, September,
2009.
6. M. Fang, S. Midlam-Mohler, R. Maringanti, F. Chiara, M. Canova, "Optimal Performance of Cylinder-by-
Cylinder and Fuel Bank Controllers for a CIDI Engine," ASME DSCC2009, September, 2009.
7. S. Midlam-Mohler, E. Marano, S. Ewing, D. Ortiz, G. Rizzoni, "PHEV Fleet Data Collection and Analysis,"
IEEE VPPC09, September 2009.
8. L. Headings, G. Washington, S. Midlam-Mohler, J. Heremans, "Thermoelectric Power Generation for Hybrid-
Electric Vehicle Auxiliary Power," Proc. SPIE Int. Conference on Smart Structures and Materials, 2009, Vol.
7290, No. 13.
9. M. Canova, S. Midlam-Mohler, G. Rizzoni, F. Steimle, D. Boland, M. Bargende, "A Simulation Study of an
E85 Engine APU for a Series Hybrid Electric Vehicle," 9th Stuttgart International Symposium on Automotive
and Engine Technology, Stuttgart, Germany, 2009.
10. S. Rajagopalan, S. Midlam-Mohler, S. Yurkovich, Y. Guezennec, K. Dudek, "Control Oriented Modeling of a
Three Way Catalyst Coupled with Oxygen Sensors," ASME Dynamic System and Controls Conference, Ann
Arbor, MI, 2008.
11. L. Headings, S. Midlam-Mohler, G. Washington, and J. P. Heremans, "High Temperature Thermoelectric
Auxiliary Power Unit for Automotive Applications," ASME Conference on Smart Materials, Adaptive
Structures and Intelligent Systems, 2008, Paper #610.
12. K. Sevel, M. Arnett, K. Koprubasi, C. Coburn, M. Shakiba-Heref, K. Bayar, G. Rizzoni, Y. Guezennec, S.
Midlam-Mohler, "Cleaner Diesel Using Model-Based Design and Advanced Aftertreatment," SAE 2008-01-
0868, 2008 International Congress, Detroit, MI, April 2008.
13. K. Dudek, B. Montello, J. Meyer, S. Midlam-Mohler, Y. Guezennec, and S. Yurkovich, "Rapid Engine
Calibration for Volumetric Efficiency and Residuals by Virtual Engine Mapping," International Congress on
Virtual Power Train Creation 2007, Munich, Germany, October 24-25, 2007.
14. M. Canova, S. Midlam-Mohler, Y. Guezennec, A. Soliman, and G. Rizzoni, "Control-Oriented Modeling of
NOx Aftertreatment Systems," SAE ICE'07 Conference, Capri, Italy, September 2007.
15. M. Canova, F. Chiara, J. Cowgill, S. Midlam-Mohler, Y. Guezennec, G. Rizzoni, "Experimental
Characterization of Mixed-Mode HCCI/DI Combustion on a Common Rail Diesel Engine," 8th International
Conference on Engines for Automobile (ICE2007), Capri, Italy.
16. M. Canova, F. Chiara, M. Flory, S. Midlam-Mohler, Y. Guezennec, G. Rizzoni, "Experimental Characterization
of Mixed Mode HCCI/DI Combustion on a Common Rail Diesel Engine," submitted to SAE ICE'07
Conference, Capri, Italy, September 2007.
Page 41 of 104
-------
17. M. Canova, M. Flory, Y. Guezennec, S. Midlam-Mohler, G. Rizzoni, and F. Chiara, "Dynamics and Control of
DI and HCCI Combustion in a multi-cylinder Diesel engine," Paper 44, submitted to 5th IF AC Symposium on
Advances in Automotive Control, Pajaro Dunes/Seascape, CA, August 2007.
18. A. Vosz, S. Midlam-Mohler, and Y. Guezennec, "Experimental Investigation of Switching Oxygen Sensor
Behavior Due to Exhaust Gas Effects," Proc. of IMECE '06, Paper IMECE 2006-14915, Chicago, IL,
November 2006.
19. S. Midlam-Mohler and Y. Guezennec, "A Temperature-Based Technique for Temporally and Spatially
Resolved Lean NOx Trap Catalyst NOx Measurements," Proc. of IMECE '06, Paper IMECE 2006-14887,
Chicago, IL, November 2006.
20. M. Canova, S. Midlam-Mohler, Y. Guezennec, G. Rizzoni, L. Garzarella, M. Ghisolfi, and F. Chiara,
"Experimental Validation for Control-Oriented Modeling of Multi-Cylinder HCCI Diesel Engines," Proc. of
IMECE '06, Paper IMECE 2006-14110, Chicago, IL, November 2006.
21. A. Soliman, S. Midlam-Mohler, Z. Zou, Y. Guezennec, and G. Rizzoni, "Modeling and Diagnostics of NOx
Aftertreatment Systems," Proc. FISITA '06, Yokohama, Japan, October 2006.
22. Z. Zou, S. Midlam-Mohler, R. Annamalai, Y. Guezennec, V. Subramaniam, "Literature Survey of On-Board
Hydrogen Generation Methods for Diesel Powertrains," Global Powertrain Conference, Novi, MI, Not Peer
Reviewed, September 2006.
23. K. Pollen, S. Midlam-Mohler, Y. Guezennec, "Diesel Paniculate Filter Regeneration with an External Burner,"
Global Powertrain Conference, Novi, MI, Not Peer Reviewed, September 2006.
24. S. Midlam-Mohler and Y. Guezennec, "Regeneration Control for a Bypass-Regeneration Lean NOx Trap
System," American Control Conference '06, Minneapolis, MN, Invited paper, June 2006.
25. A. Soliman, I. Choi, S. Midlam-Mohler, Y. Guezennec, G. Rizzoni, "Modeling and Diagnostics Of NOx After-
Treatment Systems," SAE Paper 2006-05-0208, 2006 International Congress, Detroit, MI, April 2006.
26. S. Midlam-Mohler and Y. Guezennec, "Design, Modeling and Validation of a Flame Reformer for LNT
External By-Pass Regeneration," SAE Paper 2006-01-1367, 2006 SAE International Congress, Detroit, MI,
April 2006.
27. S. Midlam-Mohler, and Y. Guezennec, "Modeling of a Partial Flow Diesel, Lean NOx Trap System," Proc. of
IMECE '05, Paper IMECE 2005-80834, Orlando, FL, November 2005.
28. M. Canova, L. Garzarella, M. Ghisolfi, S. Midlam-Mohler, Y. Guezennec, and G. Rizzoni, "A Control-Oriented
Mean-Value Model of HCCI Diesel Engines with External Mixture Formation," Proc. of IMECE '05, Paper
IMECE 2005-79571, Orlando, FL, November 2005.
29. A. Soliman, P. Jackson, S. Midlam-Mohler, Y. Guezennec, and G. Rizzoni, "Diagnosis of a NOx
Aftertreatment System," ICE 2005 7th International Conference on Engines for Automobiles, Capri, Italy,
September 2005.
30. M. Canova, L. Garzarella, M. Ghisolfi, S. Midlam-Mohler, Y. Guezennec, and G. Rizzoni, "A Mean-Value
Model of a Turbo-Charged HCCI Diesel Engine with External Mixture Formation," ICE 2005 7th International
Conference on Engines for Automobiles, Capri, Italy, September 2005.
31. M. Canova, R. Garcin, S. Midlam-Mohler, Y. Guezennec, and G. Rizzoni, "A Control-Oriented Model of
Combustion Process in HCCI Diesel Engines," American Control Conference '05, Portland, OR, June 2005.
32. C. Musardo, B. Staccia, S. Midlam-Mohler, Y. Guezennec, and G. Rizzoni, "Supervisory Control for NOX
Reduction of an HEV with a Mixed-Mode HCCI/CIDI Engine," American Control Conference '05, Portland,
OR, June 2005.
33. M. Canova, A. Vosz, D. Dumbauld, R. Garcin, S. Midlam-Mohler, Y. Guezennec, and G. Rizzoni, "Model and
Experiments of Diesel Fuel HCCI Combustion with External Mixture Formation," 6th Stuttgart International
Symposium on Motor Vehicles and Combustion Engines, Stuttgart, Germany, Not peer reviewed, February
2005.
34. S. Midlam-Mohler, S. Haas, Y. Guezennec, M. Bargende, G. Rizzoni, S. Haas, and H. Berner, "Mixed-Mode
Diesel HCCI/DI with External Mixture Preparation," Paper F2004V258, Proc. FISITA '04 World Congress,
Barcelona, Spain, May 2004.
35. Y. Guezennec, C. Musardo, B. Staccia, S. Midlam-Mohler, E. Calo, P. Pisu, and G. Rizzoni, "Supervisory
Control for NOx Reduction of an HEV with a Mixed-Mode HCCI/DI Engine," Paper F2004F233, Proc. FISITA
'04 World Congress, Barcelona, Spain, May 2004.
36. M. Gilstrap, G. Anceau, C. Hubert, M. Keener, S. Midlam-Mohler, K. Stockmeier, J-M Vespasien, Y.
Guezennec, F. Ohlemacher, and G. Rizzoni, "The 2002 Ohio State University FutureTruck - the
BuckHybrid002," 2003 SAE International Congress and Exposition, Detroit, MI, March 2003.
Page 42 of 104
-------
37. Y. Guezennec, S. Midlam-Mohler, M. Tateno, and M, Hopka, "A 2-Stage Approach to Diesel Emission
Management in Diesel Hybrid Electric Vehicles," Proc. 2002 IF AC Meeting, Barcelona, Spain, July 2002.
38. M. Hopka, A. Brahma, Q. Ma, S. Midlam-Mohler, G. Paganelli, Y. Guezennec, and G. Rizzoni, "Design,
Development and Performance of Buckeyebrid: The Ohio State Hybrid Electric FutureTruck 2001," SAE SP-
1701, Not peer reviewed, March 2002.
Scholarly Presentations Independent of Paper Publications:
1. S. Midlam-Mohler and Y. Guezennec, "Lean NOx Trap Modeling Based on Novel Measurement Techniques,"
CLEERS Conference Workshop 3, Not peer reviewed, May 4, 2006.
2. S. Midlam-Mohler, and Y. Guezennec, "Design, Modeling and Validation of a Flame Reformer for LNT
External By-Pass Regeneration," 2005 DEER Conference, Chicago, IL, Not peer reviewed, August 2005.
3. M. Canova, S. Midlam-Mohler, Y. Guezennec, and G. Rizzoni, "Control-Oriented Modeling of HCCI
Combustion," 2005 DEER Conference, Chicago, IL, Not peer reviewed, August 2005.
4. S. Midlam-Mohler and Y. Guezennec, 2004 DEER Conference, San Diego, CA, Not peer reviewed, August
2004.
5. S. Midlam-Mohler, Y. Guezennec, G. Rizzoni, M. Bargende, and S. Haas, "Mixed-Mode Diesel HCCI with
External Mixture Preparation," 2003 DEER Conference, Newport, R. I., Not peer reviewed, August 2003.
6. S. Midlam-Mohler, Y. Guezennec, "An Active, Thermo-Chemically Managed Diesel NOx After-Treatment
System," CLEERS Conference Workshop 2, Not peer reviewed, October 11, 2001.
Intellectual Property Activity
Issued Patents:
1. S. Midlam-Mohler, B. Masterson, "System System for Controlling NOx Emissions During Restarts of Hybrid
and Conventional Vehicles," U.S. Patent 7,257,493, awarded 3/21/07.
2. S. Midlam-Mohler, "System and Method for Reducing NOx Emissions after Fuel Cut-Off Events," U.S. Patent
7,051,514, awarded 5/30/06.
Patent Applications:
1. S. Liu, K. Dudek, S. Rajagopalan, S. Yurkovich, Y. Hu, Y. Guezennec, S. Midlam-Mohler, "Off-Line
Calibration of Universal Tracking Air Fuel Ratio Regulators," U.S. Patent Application 20090271093,
10/29/2009.
2. S. Rajagopalan, K. Dudek, S. Liu, S. Yurkovich, S. Midlam-Mohler, Y. Guezennec, Y. Hu, "Universal
Tracking Air-Fuel Regulator for Internal Combustion Engines, U.S. Patent Application 20090266052,
10/29/2009.
3. K. Dudek, S. Rajagopalan, S. Yurkovich, Y. Guezennec, S. Midlam-Mohler, L. Avallone, I. Anilovich, "Air
Fuel Ratio Control System for Internal Combustion Engines," U.S. Patent Application 20090048766,
2/19/2009.
4. Y. Guezennec and S. Midlam-Mohler, Shawn, "Fuel Preparation System for Combustion Engines, Fuel
Reformers and Engine Aftertreatment," U. S. Patent Application 20040124259, 7/1/04
5. S. Midlam-Mohler and B. Masterson, "System and Methods for the Reduction of NOx Emissions after Fuel
Cut-Off Events," U.S. Patent application 20060021326, filed 2/2/03.
6. S. Midlam-Mohler and B. Masterson, "Strategy for Controlling NOx Emissions During Hot Restarts for Hybrid
and Conventional Vehicles," U.S. Patent Application 20060021330, filed 2/2/03.
Patent Applications in Preparation:
1. J. Meyer, S. Midlam-Mohler, K. Dudek, S. Yurkovich, Y. Guezennec, Topic: Engine emissions control, Status:
submitted to patent office 9/09.
2. J. Meyer, S. Midlam-Mohler, K. Dudek, S. Yurkovich, Y. Guezennec, Topic: Engine emissions control, Status:
submitted to patent office 9/09.
3. S. Midlam-Mohler, S. Rajagopalan, K. Dudek, S. Yurkovich, Y. Guezennec, Topic: Catalyst modeling for
improved emissions control, Status: Patent application being prepared by outside counsel.
S. Midlam-Mohler C.V. Page 9
Page 43 of 104
-------
ADDRESS: 2373 LESLIE CIRCLE, ANN ARBOR, MI 48105 PHONE: 787-475-0241 EMAIL: EORTIZSO@UMICH.EDU
Curriculum Vitae
ELLIOTT ORTIZ-SOTO
EDUCATION
UNIVERSITY OF MICHIGAN-ANN ARBOR (U-M) In Progress Ann Arbor, MI
PhD Pre-Candidate in Mechanical Engineering (4** Year)
Relevant Graduate Coursework: Turbulent Combustion, Turbulent Flow, Combustion Processes, Advanced Internal
Combustion Engines, Hybrid Electric Vehicles, Gas Turbine Propulsion, Advanced Heat Transfer, Advanced Fluid
Mechanics, Advanced Thermodynamics, Computational Fluid Dynamics, Internal Combustion Engines, Heat
Transfer Physics, Partial Differential Equations, Probability & Statistics
UNIVERSITY OF MICHIGAN-ANN ARBOR May 2010 Ann Arbor, MI
Master of Science in Mechanical Engineering
Thesis: Dual-Mode SI-HCCI Operation for Improved Drive-Cycle Fuel Economy: Modeling Framework
Development and Implementation in Comparative Fuel-Economy Study
MASSACHUSETTS INSTITUTE OF TECHNOLOGY (MIT) June 2006 Cambridge, MA
Bachelor of Science in Mechanical Engineering GPA: 4.2/5.0
Language Concentration in German
Thesis: Design of Oil Consumption Measuring System to Determine the Effects of Evolving Oil Sump
Composition over Time on Diesel Engine Performance and Emissions
RESEARCH
WALTER E. LAY AUTOMOTIVE LAB (U - M) Fall 2007 - Present Ann Arbor, MI
Researching the physics behind novel combustion approaches, involving high pressures, ultra high dilution,
spark-assisted compression ignition (SACI) and alternative fuels, and began combustion modeling and coding
work for the implementation in GT-Power as user-developed subroutines.
Developed complete heat release analysis program in Matlab for improved experimental heat release analysis of
multi-mode combustion engines and future combustion model development.
*ป* Increased computational speed and functionality through full Matlab implementation
*ป* Superior accuracy in temperature, heat transfer and heat release calculations through:
o Better properties estimation using in-house properties and equilibrium functions (based on
JANAF tables).
o Updated residual estimation techniques for unconventional valve actuation strategies.
o Single-zone and two-zone heat release analysis options to account for various combustion modes.
*ป* Fully functional Matlab GUI for enhanced utility and ease of use.
Developed complete modeling and simulation framework for fuel-economy evaluation and mode transition
studies of Dual-Mode SI-HCCI engines involving:
*ป* Detailed system-level engine models of spark-ignition (SI) and HCCI engines using GT-Power
*ป* Experimental validation of engine, combustion, heat transfer, knock, and emissions submodels based on
Fully-Flexible Valve Actuation Engine at the U-M Auto Lab.
*ป* Full range SI and HCCI engine operating map generation using Design of Experiments optimization
*ป* Flexible architecture vehicle model using a coupled GT-Suite/Simulink approach for intuitive physical
modeling and improved controls development
*ป* Drive-cycle simulations to assess real fuel-economy benefits of Dual-Mode SI-HCCI operation over
conventional SI engines
Performed simulation study exploring the potential synergy between the HCCI engine system and three hybrid
electric vehicle (HEV) configurations, proposed the supervisory control strategy to maximize the benefits
combining the two technologies.
Page 44 of 104
1
-------
*ป* Developed Matlab/Simulink conventional, split-hybrid and parallel-hybrid vehicle models
*ป* Implemented fuel-consumption maps for SI and HCCI engines and created bsfc-optimized shifting
strategies for each engine operating mode.
*ป* Developed rule-based control strategy to maximize HCCI engine operation and minimize mode
transitions
Developed new HCCI engine cycle simulation using a zero-dimensional thermodynamic combustion approach
with detailed chemical kinetics within the Cantera-Matlab environment, and investigated the effects of engine
speed, fueling and variable valve actuation on ignition timing
Proposed practical design to achieve constant-volume combustion using advanced split-cycle engine concept and
performed a modeling study to compare efficiency benefits over conventional and other split-cycle engines.
OAK RIDGE NATIONAL LABORATORY (ORNL) Summer 2010 Oak Ridge, TN
Fuels, Engines and Emissions Research Center (FEERC)
Started work on improved experimental engine heat release analysis program for in-depth evaluation of multi-
mode combustion, model development and validation.
Researched current state-of-the-art flame propagation and chemical kinetics models for SI and HCCI
combustion simulation, and evaluated their possible implementation as simplified models for system-level
simulations.
OAK RIDGE NATIONAL LABORATORY (ORNL) Summer 2009 Oak Ridge, TN
Fuels, Engines and Emissions Research Center (FEERC)
Began work on comprehensive, physics-based Spark-Assisted HCCI model for use in system-level simulations.
Presented in detail components and implementation of the U-M HCCI Combustion correlation.
Developed improved GT-Power engine model of experimental single-cylinder engine with fully-flexible valve
actuation capable of multi-mode SI and HCCI operation.
Performed validation study of engine and combustion models with available experimental data.
SLOAN AUTOMOTIVE LABORATORY (MIT) Fall 2005 - Spring 2006 Cambridge, MA
Set up experimental single-cylinder diesel engine for emissions and oil consumption studies
Studied formation and evolution of inorganic emissions from different diesel fuel compositions and evaluated its
effect on diesel particulate filter performance
WORK EXPERIENCE
M RACING Fall 2009 - Present Ann Arbor, MI
Formula SAE Powertrain Division
Serving as experienced modeling consultant for development of improved engine model in GT-Power.
Current engine model capable of reproducing similar experimental engine behavior; expected improvements with
further model enhancements in near future.
FORD MOTOR COMPANY Summer 2008 Dearborn, MI
Intern Transmission/Driveline Research & Advanced Engineering
Performed hydraulic, transmission and vehicle level simulations (Matlab/Simulink & Ford Software), validated
models with experimental data for Stop-Start w I Assisted Direct Start (Micro-Hybrid) technology development.
Studied formation and evolution of inorganic emissions from different diesel fuel compositions and evaluated its
effect on diesel particulate filter performance.
FORD MOTOR COMPANY Summer 2007 Livonia, MI
Intern Automatic Transmission New Product Center (Electro-Hydraulic Components)
Assessed theoretical performance of competitive 6-speed automatic transmission pumps.
Established target comparison metrics and presented preliminary data suggesting design improvements for
increased efficiency.
ZF FRIEDRICHSHAFEN AG July 2006 - December 2006 Friedrichshafen, Germany
Intern Automatic Transmission New Product Center (Electro-Hydraulic Components)
Page 45 of 104
-------
Worked on new simulation approaches with Dymola (Modelica) and prepared training material for new users.
Researched new control techniques for disturbance reduction in future hybrid transmission systems.
Optimized powertrain/vehicle level models for real-time simulations (DSpace) used in pre-development and
serial production projects.
FORD MOTOR COMPANY Summer 2005 Dearborn, MI
Intern -Automatic Transmission New Product Center (Electro-Hydraulic Components)
Tested and analyzed competitive air induction system performance in environmental wind tunnels.
Presented data to recommend and support possible air induction system redesign/placement.
MIT MOTORSPORTS Fall 2005 - Spring 2006 Cambridge, MA
Formula SAE Powertrain Division
Redesigned complete formula race car air induction system.
PUBLICATIONS
Ortiz-Soto, E., Babajimopoulos, A., Lavoie, and G., Assanis, D., "A Comprehensive Engine to Drive-Cycle
Modeling Framework for the Evaluation of Future Engine and Combustion Tehcnologies," International Journal
of Engine Research (IJER). (Submitted)
Lawler, B., Ortiz-Soto, E., Gupta, R., Peng, H., and Filipi, Z.S, "Hybrid Electric Vehicle Powertrain and Control
Strategy Optimizatino to Maximize the Synergy with a Gasoline HCCI Engine," SAE Paper 11PFL-0963.
(Submitted)
Delorme, A., Rousseau, A., Wallner, T., Ortiz-Soto, E., Babajimopoulos, A., and Assanis, D., "Evaluation of
Homogeneous Charge Compression Ignition (HCCI) Engine Fuel Savings for Various Electric Drive
Powertrains," The 25th World Battery, Hybrid and Fuel Cell Electric Vehicle Symposium & Exhibition,
Shenzhen, China, November 5-9, 2010.
PRESENTATIONS
Ortiz-Soto, E., Babajimopoulos, A., Lavoie, G., and Assanis, D., "Dual-Mode SI-HCCI Operation for Improved
Drive-Cycle Fuel Economy: Engine Modeling and Map Generation Framework," USCAR, May 12, 2010
Ortiz-Soto, E., Babajimopoulos, A., Lavoie, G., and Assanis, D., "Dual-Mode SI-HCCI Operation for Improved
Drive-Cycle Fuel Economy: Modeling Framework Development," Low Temperature Combustion (LTC)
University Consortium Meeting, USCAR, October 7-8, 2010
PROFESSIONAL DEVELOPMENT
GT-Pon>er Advanced Training Seminar, November 2010
Direction in Engine-Efficiency and Emissions Research (DEER) Conference, September 2010
. High-Pressure Lean Burn (HPLB) Consortium Meeting, USCAR, October 2010
Princeton-CEFRC Summer Program on Combustion: 2010 Session, June 27 July 3, 2010
Low Temperature Combustion (LTC) Consortium Meeting, USCAR, October 2009
. WINDPOWER 2009 Conference, May 2009
Society of Automotive Engineering (SAE) World Congress, April 2008
High Performance Engine Design and Development Seminar, April 2008
U-M Graduate School Recruiter @ SHPE Conference, October 2007
Certified Engineering in Training (E.I.T)
AWARDS & ACHIEVEMENTS
2010 and 2009 ScholarPOWER Award for Master's Student Achievement, University of Michigan - Ann Arbor
Awarded GEM Fellowship for PhD (2009) and Master's (2007) studies in Mechanical Engineering
2006 and 2005 Lufthansa Award for Excellence in German Studies, Massachusetts Institute of Technology
PROFESSIONAL ORGANIZATIONS
Society of Automotive Engineers (SAE)
Page 46 of 104
-------
American Society of Mechanical Engineers (ASME)
Society of Hispanic Professional Engineers (SHPE)
Alliance for Graduate Education and the Professoriate (AGEP)
Latino Engineering Graduate Organization (LEGO)
SKILLS
Languages: Fully bilingual and bicultural (English and Spanish). Fluent in German.
Computer: Matlab/Simulink, GT-Power/GT-Suite, Fortran, Mathematica, Fluent, SolidWorks and MS Office
Page 47 of 104
-------
Aris Babajimopoulos
2325 Leslie Circle
Ann Arbor, MI 48105
12/22/2010
Tony Lentz
RTI International
3040 Cornwall! s Road
RTF, NC 27709
Dear Mr. Lentz,
Enclosed is my review of the EPA GEM model. In reviewing the material, I did not
encounter any real or perceived conflicts of interest. Please note that this review was
conducted outside of my normal job duties as an Asst. Research Scientist at the W.E. Lay
Automotive Laboratory of the University of Michigan.
I appreciate the opportunity to review the EPA GEM model and hope that my comments
are helpful. I would be happy to address any questions or concerns that may arise.
Sincerely,
Aris Babajimopoulos, PhD
enclosure
Page 48 of 104
-------
Daniel L Flowers, Ph.D.
San Leandro, CA 94577
dlfenergyconsulting@gmail.com
12/27/2010
Tony Lentz
RTI International
3040 Cornwallis Road
RTP, NC 27709
Mr. Lentz,
Enclosed is my review of the EPA GEM model. In reviewing the material, I did not encounter any real or
perceived conflicts of interest. This review was conducted as a private consultant outside of my normal
job duties as a member of the technical staff at Lawrence Livermore National Laboratory.
I appreciate the opportunity to review the model and hope that my comments are helpful to the review
process.
Sincerely,
Daniel L. Flowers
enclosure
Page 49 of 104
-------
3938 Norbrook Drive
Columbus, Ohio 43220
12/11/2010
Tony Lentz
RTI International
3040 Cornwallis Road
RTP, NC 27709
Mr. Lentz,
Enclosed is a review of the EPA GEM model. In reviewing the material, I did not encounter any real or
perceived conflicts of interest. This review was conducted outside of my normal job duties as a
Research Scientist at the Ohio State University Center for Automotive Research; however, my
experience from this position was invaluable for conducting the review.
I appreciate the opportunity to review the model and hope that my comments are helpful to the review
process.
Sincerely,
Shawn Midlam-Mohler
enclosure
Page 50 of 104
-------
2373 Leslie Circle
Ann Arbor, Ml 48105
01/11/2011
Tony Lentz
RTI International
3040 Cornwallis Road
RTP, NC 27709
Mr. Lentz,
Enclosed is a review of the EPA's Greenhouse-Gas Emissions Model (GEM). During the review process, I
did not encounter any real or perceived conflicts of interest. This peer review was conducted outside of
my normal job duties as a Research Assistant at the University of Michigan Walter E. Lay Automotive
Laboratory; however, my work at the lab has provided me with the knowledge and experience that was
indispensable for conducting the review.
I appreciate the opportunity to become part of the reviewer team and hope this review provides some
useful feedback in the development and improvement of the GEM compliance simulation tool.
Best regards,
Elliott Ortiz-Soto
enclosure
Page 51 of 104
-------
Review of Greenhouse gas Emissions Model (GEM)
Reviewer: Aris Babajimopoulos, PhD
Dept. of Mechanical Engineering
University of Michigan
Documents reviewed:
1. Greenhouse Gas Emissions Model (GEM) User Guide
EPA-420-B-10-039 (October 2010)
Filename: 420bl0039.pdf
2. gem-vl.0-executable.zip
3. gem-vl.0-matlab.zip
Page 52 of 104
-------
Introduction
The new Greenhouse gas Emissions Model (GEM), developed by EPA as a tool for the
determination of compliance with the newly proposed GHG emissions and fuel economy
standards for Class 7 and 8 combination tractors and Class 2b-8 vocational vehicles, is
essentially a very detailed and complex transient truck simulation. The stated goal of EPA
for GEM is the assessment of the impact of tractor cab design (through its effect on drag
coefficient) and/or truck tires (through changes in rolling resistance and weight reduction)
on a vehicle's compliance with the new standards. The main objectives of this review can
be summarized as follows:
1. Assess the model's completeness and functionality and check for errors (technical
or of implementation).
2. Comment on EPA's overall approach to the stated purpose of the model.
The review is organized as follows: First comments are offered on the main model
components and assumptions, as found in the Matlab/Simulink version of GEM, followed
by some general comments on the Matlab code. Then some comments on the GEM user
guide are provided, followed by comments on the GEM executable. Finally, there is a
discussion about the appropriateness of GEM as a tool for determining vehicle compliance.
Main model components and assumptions
The model is indeed very complete and covers all these components that affect overall
vehicle performance and fuel economy. The Simulink model is well organized. The use of
many Goto and From blocks allows for a clean model; however it makes it a little more
difficult to follow for someone who is not familiar with the model. The Matlab codes are
also very well structured, well documented and easy to follow.
After detailed examination of the model, I have reached the conclusion that the
assumptions used in the model are reasonable and the model itself is free of major errors
of implementation. I only have three comments: one addresses the complexity of the
submodel for the electric system; one touches on the fact that engine fuel maps,
transmission and final drive are prescribed in the model; and one has to do with the default
value for the density of air. To facilitate the discussion, Table 1 includes the inputs and
simulation results for a baseline vehicle (Class 8 - Sleeper cab - high roof, MY 2010), as
well as several results for runs with the same exact vehicle and single parameter variation.
Electric system submodel
The model of the electric subsystem is particularly detailed and convoluted. GEM includes
submodels for the starter, alternator, battery and electric accessories. This complexity
seems unnecessary for the stated purposes of GEM. Careful examination of the results
reveals that the starter has almost zero effect on overall fuel economy and C02 emissions.
Moreover, the overall effect of the electrical system on fuel economy and C02 emissions is
almost negligible. Table 1 shows that if the electric load is totally ignored (by overriding the
Page 53 of 104
-------
elec_torq signal and setting it to zero), the simulation results change by 0.2-0.4% for all
tests.
Unless it is intended in the future to use GEM for the simulation of hybrid vehicles, it is hard
to justify such model complexity, particularly since the load from the mechanical
accessories is simply modeled as a constant power demand. It would be just as simple to
model the electric load as a constant power demand. Doing something along these lines, i.e.
adding the value of electric . ace . power to engine . ace .power and then setting
electric .ace .power to zero, produces results very close to the baseline ones for the
steady state cycles and the weighted fuel consumption and is off by 0.5% for the transient
cycle.
Engine, transmission and final drive
It is stated in the GEM user guide, that EPA is primarily interested in assessing the impact
of aerodynamic drag coefficient, tire rolling resistance and tire weight reduction on fuel
economy and C02 emissions. For this purpose, engine fuel maps and drivetrain parameters
are hardwired in the model and the user has no option of changing them. However, it
seems counterintuitive that a tool for determining compliance with emissions standards
would ignore efforts on the part of the manufacturers to make improvements on the engine
itself. Moreover, in order to take full advantage of any improvements in combustion and
engine-out emissions, the vehicle transmission needs to be optimized for a particular
vehicle/engine/driving schedule combination, so that the engine can operate near its
optimum efficiency points at all times. To illustrate this point, Table 1 includes a
comparison of the baseline vehicle, with final drive ratio equal to 2.64:1, with the same
vehicle but with final drive ratio changed to 2.77:1. This relatively small change in final
drive ratio (~5%), results in worse fuel economy for all tests (on the order of 3%), simply
by forcing the engine to operate at a less efficient region of the fuel map.
This point will be revisited in the discussion on the appropriateness of GEM as a tool for
determining vehicle compliance.
Density of air
The specified air density value (1.1071 kg/m3) in 'ambient_param.m' seems to be rather
low. Using the gas constant for air (287 J/kg.K) and the specified temperature (293 K) and
pressure (101325 Pa), the density can be calculated to be 1.205 kg/m3. As it can be seen in
Table 1, this difference of around 8.8% in air density does not have a great impact on the
model predictions for the transient cycle (approximately 0.5%), however it changes the
results for the steady-state cycles by 4-5% (worse fuel economy, due to increased drag).
General comments on the Matlab code
Cross-platform compatibility: Using the hardwired file separator (\) makes the
model incompatible with platforms other than windows. Consider defining file
Page 54 of 104
-------
names using the Matlab function 'fullfile'. For example, instead of the command
load (' drive_cycles\ARB_Transient'), one can use the command
load (fullf lie ( 'drive_cycles ' , ' ARB_Transient' ) ). The Matlab model
contains a total of 3 load and 10 run statements using the \ separator. By making
the aforementioned change 13 times, I was able to execute the Matlab model on
both a Mac and a Linux machine.
Robustness: Code contains hooks for future additions, but some of them seem
unnecessary. For example, i_sim as a case number index is hardwired to be equal to
1. If the value changed, the code would network without modifications, considering
that the arrays it points to have not been defined as such (vehjype, c_d, c_rr_steer,
etc.). Currently the code works only because i_sim is always equal to 1.
General comments on the GEM user guide
The GEM user guide is overall well written and clear. There are only some minor issues:
The model description is too detailed, referring to features of the model that are
irrelevant and outside the scope of GEM, even though these features are present in
the model. For example, there is reference to the road gradient that can be specified,
although this is never actually done; it is stated that a either a 12 or 24 volt standard
lead acid can be modeled, although in reality only one type of battery is modeled; in
the description of the Mechanical Accessory Block, Power Take Off is mentioned.
This detailed description of the model may generate unnecessary confusion to the
users of GEM.
In the discussion of the Electric Accessory component of the model, it is stated that
"all vehicles have a number of electrical loads...and these are already taken into
account in the fuel map." This is a troubling statement. It is indeed standard practice
to include accessory loads (both electrical and mechanical) in the fuel map, by
effectively changing the brake torque of the fuel map. However, in GEM, the
electrical load is actually calculated at every time step, which means that the fuel
maps that are used should not include the impact of any accessory loads. Hopefully,
this is indeed the case and this statement was included by error.
GEM download, installation and execution
The download and installation instructions, which are included in the GEM user guide, are
very clear. Installation was absolutely trouble free.
Comments on GEM executable
The coefficient of aerodynamic drag can only be specified with a pull-down list of
values from 0.50 to 0.85, with step 0.05. As a result, not all intermediate values for
Cd can be specified, including the recommended values provided by EPA in Table 5
Page 55 of 104
-------
(e.g. 0.69, 0.76, 0.81 etc.). Considering the significant impact of Cd on fuel economy
and its importance in achieving compliance, the value of Cd should be allowed to be
entered in a textbox.
In the GEM user guide, it is stated that the coefficient of drag is only required for
combination tractors and no input is required for vocational trucks (page 8).
However, when one selects the Heavy Heavy-Duty vocational vehicle, it is still
possible to change the value of Cd through the pull-down menu. Moreover, the
selected value seems to indeed influence the fuel economy prediction. Similarly to
Cd, the GEM user guide states that the input for Vehicle Speed Limiter is only
available for combination tractors and no input is allowed for vocational trucks.
Nevertheless, it is indeed possible to enter a value for vehicle speed limiter for a
vocational truck and this value has an effect on overall fuel economy. Table 2 shows
the inputs and results for 3 simulated cases for a Heavy Heavy-Duty vocational
truck. The first case is the baseline case. The inputs for the second case are the same
as the baseline ones, except that Cd is changed from 0.8 to 0.6. Similarly, for the
third case, all inputs are the same as the baseline ones, except that the vehicle speed
is limited to 55 mph instead of 65 mph. It can be seen clearly in Table 2 that both Cd
and speed limiter influence the results for Cases 2 and 3. If this behavior is not
desired, either the underlying code should be modified to ignore these inputs for
vocational trucks and use the default values or the GUI should be modified, so that
when a vocational truck is selected from the regulatory class list, the corresponding
input fields should become inactive ("grayed out").
Minor comments:
o It would be good if the message indicating where the results will be stored
also include the drive (C:) in the path (e.g.
'C:\GEM_Results\December_14_2010-0135PM instead of
\GEM_Results\December_14_2010-0135PM'
o The fact that the three figures must be closed one after the other before the
program execution ends is a little confusing, at least initially. It would be nice
if this behavior could change.
GEM as a tool for determining vehicle compliance
GEM is a very detailed vehicle simulation that could capture with reasonable accuracy the
impact of changes in aerodynamic drag coefficient, tire rolling resistance and tire weight
reduction on overall vehicle fuel economy and C02 emissions. The model itself is almost
too detailed for this purpose, but this should not be a problem, provided that not all details
of the model are discussed in such great length with the users.
However, as mentioned in the discussion about the engine, transmission and final drive, it
is hard to envision a compliance tool that does not account for fuel economy improvements
coming from the development of advanced combustion technologies by the engine
manufacturers. If the assumption is that engines will be relatively similar for the same class
Page 56 of 104
-------
vehicles coming from different manufacturers, then it is safe to assume that GEM would be
an appropriate tool for determining compliance with fuel economy and C02 emissions
standards based on vehicle design changes alone. Nevertheless, if it were anticipated that
trucks of the same class from different manufacturers would use engines with significantly
different fuel maps, it would be proper to allow for the provision to change the engine fuel
map and transmission characteristics used by GEM.
Page 57 of 104
-------
Table 1. Effects of the variation of various model parameters on the simulation
results for a baseline vehicle (Class 8 - Sleeper cab - high roof, MY 2010)
Baseline
No electrical
accessories
Electrical
ace. power
added to
mechanical
ace. power
Final drive
equal to 2.77
instead of
2.64
Air density
equal to
1.205 instead
of 1.1071
Model Inputs
Coefficient of Aerodynamic Drag
Steer Tire Rolling Resistance [kg/metric ton]
Drive Tire Rolling Resistance [kg/metric ton]
Vehicle Speed Limiter [mph]
Vehicle Weight Reduction [Ibs]
extend edldleReductionLabel
0.69
7.8
8.2
N/A
N/A
N/A
Transient Cycle Simulation
Fuel Consumption for Entire Cycle [mpg]
CO2 Emissions [g/ton-mile]
3.51
152.86
3.51
152.52
3.49
153.64
3.38
158.68
3.49
153.41
55 mph Steady-State Cycle Simulation
Fuel Consumption during Steady State [mpg]
CO2 Emissions [g/ton-mile]
7.40
72.38
7.43
72.14
7.41
72.32
7.28
73.59
7.15
74.98
65 mph Steady-State Cycle Simulation
Fuel Consumption during Steady State [mpg]
CO2 Emissions [g/ton-mile]
6.19
86.52
6.21
86.22
6.20
86.40
6.01
89.22
5.91
90.67
Cycle-Weighted Results
Weighted Fuel Consumption [mpg]
-> in gal/1000 ton-mile
Weighted CO2 Emission [g/ton-mile]
6.17
8.70
88.57
6.19
8.67
88.27
6.17
8.69
88.49
5.99
8.97
91.29
5.90
9.08
92.40
Page 58 of 104
-------
Table 2. Impact of Cd and vehicle speed limiter on the simulation results for a
Heavy Heavy-Duty - Vocational Truck (Class 8)
Case 1
Case 2
Case 3
Model Inputs
Coefficient of Aerodynamic Drag
Steer Tire Rolling Resistance [kg/metric ton]
Drive Tire Rolling Resistance [kg/metric ton]
Vehicle Speed Limiter [mph]
Vehicle Weight Reduction [Ibs]
extendedldleReductionLabel
0.8
9
9
65
0
0
0.6
9
9
65
0
0
0.8
9
9
55
0
0
Transient Cycle Simulation
Percent Time Missed by 2mph [%]
Fuel Consumption for Entire Cycle [mpg]
CO2 Emissions [g/ton-mile]
1.51
3.51
152.74
1.5
3.55
150.89
1.51
3.51
152.74
55 mph Steady-State Cycle Simulation
Percent Time Missed by 2mph [%]
Fuel Consumption during Steady State [mpg]
CO2 Emissions [g/ton-mile]
0.23
6.47
82.75
0
7.24
74.04
0.23
6.47
82.75
65 mph Steady-State Cycle Simulation
Percent Time Missed by 2mph [%]
Fuel Consumption during Steady State [mpg]
CO2 Emissions [g/ton-mile]
0
5.34
100.41
0
6.18
86.69
0
6.48
82.75
Cycle -Weighted Results
Weighted Fuel Consumption [mpg]
> in gal/1000 ton-mile
Weighted CO2 Emission [g/ton-mile]
4.81
11.66
118.68
5.3
10.9
111
5.23
11.02
112.15
Page 59 of 104
-------
Review of EPA GEM model for as a tool for evaluation of medium and heavy-
duty vehicle GHG emissions
Daniel L. Flowers, Ph.D.
danflowers@gmail.com
19 Dec 2010
This report reviews the methodology developed by EPA for evaluating greenhouse
gas (GHG) emissions reductions from medium and heavy-duty road vehicles [1].
This model focuses on GHG emissions improvements based on vehicle drag
reduction and rolling resistance reduction, and would be used by EPA as a
regulatory tool to evaluate compliance by vehicle manufacturers.
In general, the goal of this program is to provide a framework for fairly evaluating
GHG emissions from medium and heavy-duty vehicles [2]. Thus, a key mission of
this review is evaluating how well the modeling approach developed serves as a
regulatory and compliance tool.
We reviewers have been asked to comment on the model with regard to 5 specific
items:
1] EPA's overall approach to the stated purpose of the model (meet agencies'
compliance requirements] and whether the particular attributes found in
resulting model embodies that purpose.
2} The appropriateness and completeness of the contents of the overall model
structure and its individual systems, and their component models, if
applicable (i.e., using the MATLAB/Simulink version], such as:
a] The elements of each system to describe different vehicle categories;
b] The performance of each component model, including the reviewer's
assessment of the underlying equations and/or physical principles
coded into that component.
c] The input and output structures and how they interface with the
model to obtain the expected result, i.e., fuel consumption and C02
over the given driving cycles; and
d] The default values used for the input file, as shown in the GEM User
Guide.
3] Using the standard of good engineering judgment, the program execution is
optimized by the chosen methodologies;
4] Clarity, completeness and accuracy of the output/results (C02 emissions or
fuel efficiency output file]; and
Page 60 of 104
-------
5} Any recommendations for specific improvements to the functioning or the
quality of the outputs of the model.
Detailed discussion of each of these items will be described in the following sections.
Item 1) Overall approach
Based on the description of the proposed use in rulemaking (EPA-420-D-10-901)
[1], the overall approach is to provide a neutral framework upon which different
vehicles from different manufacturers can be compared. The idea of this approach
is to eliminate manufacturer differences by looking only at the external vehicle loss
characteristics: drag coefficient and coefficients of rolling resistance. For vehicles
from different manufacturers in each regulatory subcategory, there are several
assumptions made about the vehicle characteristics:
1. The frontal area is the same
2. Accessory power required is the same
3. Vehicle mass is the same
4. Distribution of weight on drive, steering, and trailer tires is the same
5. The engine is the same
6. The transmission and driveline losses are the same
For a regulatory subcategory of vehicles (e.g. Class 8 Sleeper Cab High Roof),
assumptions 1, 2, and 3 are very reasonable. Frontal area is likely very similar for
subcategory vehicles, and vehicle mass is likely to be similar based on gross vehicle
regulated weight. Accessory loads vary from truck to truck and application-to-
application, so constant accessory load for all is a reasonable approximation.
Assumptions 4, 5, and 6 are not necessarily fully justified. With regard to
assumption 4, for non-vocational trucks, the overall rolling resistance is specified as
42.5% trailer, 42.5% drive wheels, and 15% steering wheels. This has a potential to
penalize a vehicle that has reduced cab mass and biased the load towards the trailer.
However, it is likely to be a small effect and does not seem likely to be frequently
significant.
Assumptions 5 and 6 are more problematic. The engine and transmission can be
suitably sized to the load characteristics. In this case, the engine and transmission is
not optimized to the vehicle. This issue will be discussed quantitatively and in
greater detail in the next section of this review. Consider a Class 8 tractor with a
drag coefficient of 0.69 that has the engine optimally sized for the engine and
transmission on the drive cycle. Reducing the drag coefficient by 13% to 0.60 will
reduce the load requirements, shifting the operation to lower load on the engine.
Diesel engine achieve highest efficiency at highest load and efficiency decreases with
decreasing load. Thus the lower drag vehicle may operate on a lower efficiency part
of the engine map.
Page 61 of 104
-------
In practice, the engine and transmission can be appropriately sized to best take
advantage of the reduced overall vehicle load. By requiring only one engine and
transmission be used, drag reduction efforts could be penalized.
The danger exists that the manufacturers would be encouraged to optimize vehicles
to meet the characteristics that will give the best performance with the simulation
tool, instead of optimizing the vehicle to achieve the true goals of reducing fuel
consumption and GHG emissions.
Overall, the concept of using a generic vehicle model has merit to limit the need to
test the myriad possible vehicle configurations. The use of a generic powertrain
(engine and transmission] is problematic because a well-integrated powertrain can
significantly improve vehicle performance.
Item 2) Functional aspects of the overall model and model components
This section focuses on verification that the model works as expected, as well as
how the model parameters and components affect the prediction of fuel
consumption and GHG emissions in context of regulatory use. The first step is a
sanity check on the results of the model compared with direct calculation.
Determining fuel consumption analytically requires working backwards from the
forces and accelerations on the vehicle to the engine fuel consumption map.
Equation 1 shows gross engine power in terms of vehicle parameters based on
working backwards from the forces on the vehicle. The full derivation with
description of the parameters is in the appendix.
Pengme,gr0ss=\crrmgC0^)V + ]-pcdAfV^ +mgsm(p)V + maV + ^(lkdkdk}\+Pacc (1)
^tr \ L k=\ I
At constant speed and zero grade, the net acceleration and gravity terms become
zero.
P =\c meV + oc A V3\+P (2\
1 engine,gross \<-rr"l&y ^ ~ 1^dnfy y l ace l^J
Tltr\ L I
Table 1 below shows a comparison between the output of the GEM simulation
model and torque based on calculating equation (2} for the same parameters. The
vehicle configuration used for Table 1 is from the GEM manual for the "Class 8
Combination - Sleeper Cab - High Roof [refj." Torque is compared for the constant
speed portions of the 55 mph drive cycle and the 65 mph drive cycle.
The GEM simulation code calculates engine torque and speed, not power directly. In
Table 1 the engine speed from the GEM simulation is used with the analytically
determined power to determine analytical engine torque.
Page 62 of 104
-------
Table 1 - Comparison of GEM simulation predictions to the calculations based on equations
derived in Appendix 1. Comparisons are for "veh_type(i_sim) = 1" "Class 8 Combination -
Sleeper Cab - High Roof." The engine is the first map "veh_year=l" in engine_map_455.m
Property
Units 55 mph 65 mph
Source
Mechanical Accessory Power
Electrical Accessory Power
Vehicle Speed
Vehicle Speed
Vehicle acceleration
Vehicle driving grade
Aerodynamic force on vehicle
rolling resistance force on
vehicle
Total resistive force on vehicle
Vehicle power requirement
Engine speed
Transmission efficiency
Engine Power required
Engine Torque
Engine Torque
Difference in analytical versus
GEM simulated torque
kg/mA3
m/sA2
mA2
kg
No
units
No
units
No
units
No
units
No
units
No
units
No
units
No
units
W
W
Mph
m/s
m/sA2
Degrees
N
N
N
kW
Rpm
No
units
kW
N-m
1.1071
9.8066
9.8
31978
0.69
0.0082
0.0078
0.006
0.425
0.15
0.425
0.007205
1000
360
55
24.6
0
0
2262.8
2259.5
4522.3
111.2
1266.5
0.98
114.8
865.7
1.1071
9.8066
9.8
31978
0.69
0.0082
0.0078
0.006
0.425
0.15
0.425
0.007205
1000
360
65
29.1
0
0
3160.5
2259.5
5419.9
157.5
1495.6
0.98
162.1
1034.8
GEM model: ambient_param.m
GEM model: ambient_param.m
GEM model: run_preproc.m
GEM model: run_preproc.m
Input
Input
Input
GEM model: run_preproc.m
GEM model: run_preproc.m
GEM model: run_preproc.m
GEM model: run_preproc.m
Calculated
GEM model: run_preproc.m
GEM model: run_preproc.m
GEM model: specified by drive cycle
(Mild_55_mph.mat, Mild_65_mph.mat)
Calculated
GEM model: constant speed section of drive cycle
used for analysis
GEM model: specified by drive cycle
(Mild_55_mph.mat, Mild_65_mph.mat)
Calculated
Calculated
Calculated
Calculated
Output from GEM Model: Simulink model
"GEM_manual_vl/ engine/ engine/ engine_fuel_fl'
Output from GEM Model: Simulink model
"GEM_manual_vl/ transmission/ gear/
gear_engaged"
Calculated
Calculated fusing engine speed from GEM simulat
Output from GEM Model: Simulink model
N-m 892.5 1066.2 "GEM_manual_vl/engine/engine/engine_fuel_flow"
3.0 3.0 Calculated
Page 63 of 104
-------
For this case, the analytical torque is 3% lower than the torque determined by the
GEM simulation model. A possible explanation of this discrepancy may come from
the formulation of the GEM model. In the GEM model the desired vehicle speed is
specified and the vehicle dynamic system responds to try to meet that by providing
needed engine torque. The vehicle speed is calculated in the subroutine of the
Simulink Model "GEM_manual_vl/vehicle/chassis/vehicle_speed" as shown in
Figure 2.
iปehicl4_forcซ
1ine_mass_out
Static Vehicle Mass (kg)
1>J>!
iปeh_mass_dvn
Figure 1 - Section of the GEM Simulink model where vehicle speed is calculated.
The vehicle speed comes from integrating the force balance.
f51 (3)
eff tff ^ I J
V(t) =
[4]
The GEM model uses an "effective mass" formulation that includes powertrain
inertial effects. In the GEM code, the vehicle static mass (vehicle.chsmass_static] is
added to the representative powertrain inertial mass (tire_mass_out). For steady
speed vehicle operation the powertrain inertial mass should be zero. Figure 2
shows the vehicle inertial mass (tire_mass_out] for the constant desired vehicle
speed period of the 55 mph drive cycle. The inertial mass of 1693 kg during the
steady speed demand region represents 5% of the static vehicle mass. Figure 2
shows that the inertial mass term is not zero during the constant-desired-speed
portion of the drive cycle. Figure 3 shows the vehicle chassis speed varies during
the constant speed period of the 65 mph drive cycle.
The 3% discrepancy between analytical and GEM simulated torque may be due to
the speed variation during this portion of the drive cycle. The consistency of the
model vehicle dynamics with actual vehicle dynamics is a possible way to assess
Page 64 of 104
-------
whether the model is representative of actual vehicle dynamics. Figure 4 shows
engine torque during the acceleration ramp leading up to the 55 mph steady speed
demand region of operation. Comparing actual engine torque response to this
dynamic torque response would be a way of assessing whether the dynamics are
reasonable or not. The quality of these response dynamics will be even more critical
for transient drive cycle analysis.
2000
1500
- 1000
500
100
200
300 400
total-sim-time (s)
500
600
Figure 2 - Inertial mass during the constant speed demand portion of the 55 mph drive cycle.
100
200
500
600
300 400
total-sim-time (s)
Figure 3 -Actual vehicle chassis speed during the constant speed demand portion of the 55
mph drive cycle.
Page 65 of 104
-------
20
40
100
120
140
60 80
total-sim-time (s)
Figure 4 - Engine torque during acceleration ramp in 55 mph drive cycle.
The next sanity check is whether the fuel consumption and GHG emissions are
correctly calculated based on the engine torque and speed. Figure 5 shows the
torque versus engine speed contour map. Table 2 shows a comparison of off-line
calculations of the output parameters from the 55 mph and 65 mph cases in table 1
to the output from the GEM simulation code. The torque and speed used for these
calculations are the torque and speed calculated by the GEM simulation code, not
analytically calculated torque and speed from table 1. Very small error (less than
0.3%} between off-line and GEM simulation calculations is seen. These differences
could be attributed to round off or the averaging used for off-line calculations.
3000
2500
fuel flow rate contours (kg/s
fl
0.025
0.02
0.015
0.01
10.005
600 800 1000 1200 1400 1600 1800 2000 2200
Engine Speed (rpm)
Figure 5 - contours of fuel flow rate versus engine speed and torque for 15L engine from
"engine_map_455.m".
Page 66 of 104
-------
Table 2 - Comparison of GEM simulation and direct interpolation of fuel flow for 55 mph and
65 mph cases in Table 1. Comparisons are for "veh_type(i_sim) = 1" "Class 8 Combination -
Sleeper Cab - High Roof." The engine is the first map "veh_year=l."
Property
Engine speed
Engine torque
Fuel flow rate
Fuel flow rate
Difference in calculated
versus GEM simulated fuel
flow
Fuel density
Volumetric fuel flow rate
Volumetric fuel flow rate
Vehicle speed
Fuel consumption
Fuel consumption
Difference in calculated
versus GEM simulated fuel
consumption
Payload
C02 to ton-mile conversion
C02 emissions
C02 emissions
Difference in calculated
versus GEM simulated CO 2
emission
Units
Rpm
N-m
kg/s
kg/s
%
kg/L
L/s
gal/hr
miles/hr
miles/gal
miles/gal
%
Ton
gC02/
[mpg*
payload)
g/(ton-mile)
g/(ton-mile)
55 mph
1266.5
892.5
0.00660
0.00661
0.2
0.847
0.00780
7.417
55
7.42
7.40
0.2
19
10180
72.21
72.38
0.2
65 mph
1495.6
1066.2
0.00932
0.00934
0.2
0.847
0.0110
10.5
65
6.21
6.19
0.3
19
10180
86.56
86.52
0.05
Source
GEM simulation
GEM simulation
GEM simulation
Interpolated from map in
engine_map_455.m
Calculated
From "engine. cyl.fuel_desity" in
engine_map_455.m
Calculated
Calculated
Desired steady state speed from drive
cycle
Calculated
GEM simulation results
Calculated
From run_preproc.m
From run_preproc.m
Calculated
GEM simulation results
Calculated
Following up on the earlier discussion of engine and vehicle integration, Table 3
shows an example of the effect of engine sizing on overall vehicle performance when
drag reductions are implemented. The comparison is again for the "Class 8 -
Sleeper Cab - High Roof" vehicle used for the calculations in Tables 1 and 2. Three
cases are shown: 1} base case with drag coefficient of 0.69 and engine_map_455.m
vehjrear=l engine, 2} base case with drag coefficient reduced to 0.60, and 3} base
case with drag coefficient reduced to 0.60, and engine downsized to 90% of original
engine. As an approximation of downsizing, the engine map, torque, and maximum
torque are scaled by a factor of 0.9. This scaling is representative of the
performance changes that could be achieved by, for example, reducing the
displacement of the engine, or changing the turbocharger parameters.
The results in Table 3 show that a generic engine has limitations demonstrating
benefits of drag reduction strategies. The vehicle with reduced drag and reduced
engine size has lower fuel consumption and lower C02 emissions than the vehicle
with just reduced drag coefficient. With a generic engine, this model would give a
manufacturer that reduces vehicle drag without consideration of vehicle, engine and
powertrain integration the same performance as a manufacturer that does further
optimization of the vehicle. This example is a very simplistic reduction. With
further effort greater performance benefits are likely to be realized.
Page 67 of 104
-------
Table 3 - Comparison of reduction of fuel consumption and C02 emissions due to drag
coefficient reductions and engine-vehicle integration. Comparisons are for
"veh_type(i_sim)=l" "Class 8 Combination - Sleeper Cab - High Roof." The engine is the first
map "veh^year=l" in engine_map_455.m. Calculated values come from the GEM simulation
code.
Property Units
Drag coefficient
Engine scaling
Steer wheels coefficient of rolling resistance
Drive wheels coefficient of rolling resistance
Base case Reduced drag
no units
no units
no units
no units
0.69
1
0.0078
0.0082
0.6
1
0.0078
0.0082
Reduced drag,
reduced
engine size
0.6
0.9
0.0078
0.0082
Fuel consumption, transient
Fuel consumption, 55 mpg steady
Fuel consumption,65 mpg steady
Fuel consumption, cycle weighted
Improvement in cycle weighted fuel consumption relative
to base case
C02 emissions, transient
C02 emissions, 55 mpg steady
C02 emissions, 65 mpg steady
CO 2 emissions, cycle weighted
Improvement in cycle weighted C02 relative to base case
Mpg
Mpg
Mpg
Mpg
g/ton-
mile
g/ton-
mile
g/ton-
mile
g/ton-
mile
3.51
7.40
6.19
6.17
0.00
152.47
72.38
86.52
88.55
0.00
3.53
7.80
6.66
6.60
6.97
151.67
68.65
80.48
82.98
6.29
3.64
7.96
6.70
6.66
7.94
147.15
67.27
69.92
82.15
7.23
Item 3) Program execution
The objective of this section is to evaluate if by "Using the standard of good
engineering judgment, the program execution is optimized by the chosen
methodologies." I interpret this to be asking about the performance of the code as
an effective and efficient tool for this application.
The code overall seems to be developed in a way that provides detail on the vehicle
and powertrain dynamics. The model, like the vehicle it simulates, is a complex and
highly interconnected system. There are many submodels in this code, and there
are many imbedded assumptions that are not directly apparent without a great deal
of reverse engineering. It is often difficult to test and verify submodels in isolation
because they are highly interconnected with the main model and significant effort
would be required to recreate inputs suitable for the submodel to run on its own. A
general rule in modeling is that the level of complexity of the model should be the
minimum level needed to answer the question posed.
The documentation available on the model does not provide a detailed description
of the physical models implemented. This kind of detailed documentation is needed
to fully understand the model and modeling assumptions involved.
Page 68 of 104
-------
Transparency in the details of the model is important for a regulatory application.
Transparency of this model may suffer without detailed supporting documentation
on the physics and engineering assumptions underlying each model and submodel.
Item 4) Clarity, completeness and accuracy of output
The model output is overall clear and complete. The model reports the individual
drive-cycle results and weighted average results, which is what is most important to
the end user. All the inputs needed to reproduce the results are reported. I would
suggest that the a code version also be included, so if the code is changed in the
future it will be clear from which version an output file evolved.
Accuracy of the results is difficult to assess, since that requires specific comparison
to experimental data to evaluate the performance of the model. Based on my testing
efforts and experience, the results seem of reasonable magnitude for these kinds of
vehicles.
Item 5) Recommendations for improvements
Following are small issues I noticed during my review of the code.
1} The syntax in the m-files is not compatible with unix, specifically the
directory backslash "\" vs forward slash "/".
2} The windows executable version has predefined values for C_d in a
dropdown menu with preset values in increments of 0.02. The C_d value
should just be an entry box, like the C_rr values.
3} The inputs for weight reduction, speed limiter, and idle reduction are not
consistent between the matlab version and the windows executable. For
example in the matlab version. In matlab, zero "Weight Reduction" defaults
to "N/A," which causes an error in the windows version. The windows
version does accept "N/A" for idle reduction.
4} It would be informative to have the fraction of each drive-cycle used in the
average reported somewhere in the output.
5} The fuel density variable is "engine.cyl.fuel_desity." For clarity and
consistency I would recommend changing this to "engine.cyl.fuel_density."
Conclusions
1} My main concern with the overall approach is the standardization of the
vehicle and powertrain combination. This seems to have potential to devalue
efforts towards vehicle and powertrain integration and optimization towards
GHG reduction.
2} The model is quite detailed with regard to powertrain and vehicle dynamics.
There is a danger here that imbedded assumptions can effect results in
unexpected and undesirable ways. The example of the 3% difference in
torque for analytical versus GEM simulation calculated torque for steady
state operation maybe indicative of these kinds of issues.
3} It should be confirmed whether the various controllers in the GEM model are
well tuned and result in a vehicle response consistent with empirical data.
Page 69 of 104
-------
4} Detailed description of the physics and assumptions imbedded in the models
and submodels should be documented and made available to users.
5} It may be worth considering if the model could be streamlined to provide
greater clarity and transparency while still providing a tool for quantitatively
estimating fuel consumption and GHG emissions.
References:
1} "Greenhouse Gas Emissions Model (GEM] User Guide," EPA-420-B-10-039,
October 2010.
2} Jeongwoo Lee, "Vehicle Inertia Impact on Fuel Consumption of Conventional
and Hybrid Electric Vehicles Using Acceleration and Coast Driving Strategy,"
Ph.D. Dissertation, Virginia Polytechnic Institute, 2009,
http://scholar.lib.vt.edu/theses/available/etd-09172009-
234744/unrestricted/ETD_PhD_Dissertation_Jeongwoo_Lee.pdf
3} http://www.epa.gov/otaq/climate/regulations.htm
4} Uwe Kiencke, Lars Nielsen, "Automotive Control Systems: For Engine,
Driveline, and Vehicle," Springer; 2nd edition, 2005.
Page 70 of 104
-------
Appendix: Derivation of vehicle and engine power formulas
Figure a6 shows a free-body diagram of the forces and accelerations on a vehicle.
This vehicle has mass m, acting about the center of gravity. Further reading on
these derivations is available in the literature [3, 4]. Gravitational acceleration is
treated separately here from the vehicle acceleration.
Figure a6 - Free body diagram showing forces and accelerations on a vehicle
The net forces on the vehicle in the direction of movement are:
77 _/7 _/7
tractive rr drag
[al]
Ftractive = required propulsive force on the vehicle
Frr = resistive force due to rolling resistance
Fdrag = resistive force due to aerodynamic drag
a = net vehicle acceleration in the direction of travel
m = vehicle mass (static vehicle mass]
g = gravitational acceleration
(3 = angle of vehicle travel relative to gravity normal direction.
The engine transmits torque through the powertrain to the wheels. At the wheels,
the torque transferred becomes the propulsive (or tractive] force. Figure a7 shows a
schematic of the transfer of torque from engine, through the rotating components of
the powertrain, to the force acting on the ground to propel the vehicle.
Page 71 of 104
-------
Q
N
tive^eff
Iv
-,/N-l , JJL^r-
v_
UA-
I,
k
T|oss,
-.Tk-i lUc
| :^ ., m m t u p. I
k
^1
I,
M
'loss
_ ,
-------
Equation (al] can also be used to determine power by multiplying the forces by
vehicle speed, V.
Ptractive = FJ + FdrJ + iปgsin(0)F + maV (a6)
tractive dr
The rolling resistance is defined in terms of a rolling resistance coefficient (crr] and
the normal force of the vehicle (N=mg cos(|3}}.
Frr = crrmgcos(l3) (a7]
Aerodynamic drag is defined in terms of air density (p], drag coefficient (cd], vehicle
frontal area (Af), and vehicle speed.
(a8)
rag
Combining (a5-a8), vehicle tractive power can be used relate engine power to
(a9)
Pengme = crrmgCos(PW + -pcdAfV3+mgsm(PW + maV + lk6k6k (alO)
Equation (alO] completely describes the power demand upon an engine due to
external forces and powertrain dynamics.
The engine may support vehicle accessory loads (e.g. air conditioning, lights], and
these accessory loads will be removed from the engine before the transmission.
Since accessory power (Pace] is removed before the transmission, accessory power
can be directly added to the engine power demand. Fuel consumption maps are
based on gross engine power (Pengme,gross) or torque and engine speed.
maV+> (Ik9k9k + Pacc (all)
. - _ . Z^V ซ ซ ซ/ acc v -1
'/tr V * k=1 /
A common practice is to simplify the powertrain inertia characteristics from the
final term in equation (alO) to a proportionality scaling of the vehicle acceleration
(maV) [refj. The effective mass (meff) can be calculated dynamically or
approximated.
(a!2)
Page 73 of 104
-------
Using this effective mass definition gives engine power in terms of five power
demand terms: rolling resistance, aerodynamic drag, gravity, acceleration, and
accessories.
Pengm^oss = (crrmgC0^W + ^pcdAfV3 +mgsm(p)V + meffaV\+Pacc (a!3)
r\tr\ ^ I
Page 74 of 104
-------
PEER REVIEW:
GEM Vehicle Model
Review Conducted for:
U.S. EPA
Review Conducted By:
Shawn Midlam-Mohler
Review Period:
11/19/2010-12/12/2010
Page 75 of 104
-------
Contents
Summary 4
Introduction 5
Objective 1: Capable of modeling a wide array of vehicles over different drive cycles 5
Objective 2: Contains open source code, providing transparency in the model 5
Objective 3: Freely available and easy to use by any user 6
Objective 4: Contains both optional and preset elements 6
Objective 5: Managed by the Agencies for compliance purposes 6
Model Structure Evaluation 7
Ambient Subsystem 7
Driver Subsystem 7
GEM Manual Misleading 7
Driver PID Values not Configurable 8
Gear Shifting Control 8
Electric Subsystem 8
Electrical System Parameters not Adjusted with Vehicle Class 8
Alternator Model - Current Regulation and Control 8
Alternator Model -Accessory Torque 9
Starter Model Complexity 10
Pb Battery Model Accuracy 10
Pb Battery Model Complexity 11
Electrical Accessories not Adjusted by Vehicle Class 12
Engine Subsystem: 12
GEM Manual is Unclear 12
Closed Throttle Engine Torque 12
Shawn Midlam-Mohler GEM Review Page 2
Page 76 of 104
-------
Engine Decel Fuel Cut-Off 12
Closed Throttle Engine Fuel Consumption 13
Overall Structure of Engine Model 13
Mechanical Accessories not Adjusted by Vehicle Class 14
Transmission Subsystem 14
Transmission Model Parameters should Change by Class: 14
Vehicle Subsystem 14
Vehicle Model Parameters should Change by Class 14
Vehicle Frontal Area 14
Vehicle Weight Reduction for Rotating Components: 15
Vehicle Weight Reduction not Implemented for Certain Classes of Vehicle 15
Vehicle Loss Parameters 15
GEM Input and Output Files 17
Format of Output File (xml) 17
Clarity of Output File 17
Content of Output File 17
Standard Input Values Specified in GEM Manual 17
Miscellaneous Comments 18
Adjustment of Model Parameters for Different Vehicle Classes 18
Model Fidelity 18
Sensitivity of Parameters 18
Conclusions 20
Shawn Midlam-Mohler GEM Review Page 3
Page 77 of 104
-------
Summary
The model fidelity of the type proposed should be capable of achieving the desired
objectives. The model reviewed, however, has a number of issues which cast doubt upon the
specific implementation of the model. Specifically, a number of issues were found in the
electrical subsystem as well as the engine subsystem. In many cases, it is felt that the level of
modeling used in subsystems, the electrical subsystem being one excellent example, are more
complicated the necessary given the relatively low impact on the desired outcome.
From the supporting material, it is clear that the model did an acceptable job at modeling
a Class 8 SmartWay truck. Further validation across the range of vehicles being modeled would
be appropriate to provide confidence to the end users and ensure the model is doing an
acceptable job at modeling green house gas emissions.
It is also recommended that a better understanding of the propagation of uncertainty in
the key model input parameters be evaluated. For instance, key parameters like the drag
coefficient and coefficient of rolling resistance can be measured with a certain degree of
uncertainty. It is possible to determine how these errors propagate through the model and impact
the end result of fuel consumption or greenhouse gas emissions. These results should be one part
of the overall evaluation of the model. This level of uncertainty should then be compared to the
end use of the model and the expected resolution required to distinguish between different
technologies.
Shawn Midlam-Mohler GEM Review Page 4
Page 78 of 104
-------
Introduction
The peer review directives suggested addressing a number of different issues. The first
topic was an overall assessment of the model to meet the stated objectives. In the following
subjections, there are some high-level comments on the ability of the proposed model to achieve
the five attributes listed in the peer review statement.
Objective 1: Capable of modeling a wide array of vehicles over different drive cycles
The model fidelity of the type proposed should be capable of achieving the desired
objectives. The model reviewed, however, has a number of issues which cast doubt upon the
specific implementation of the model. Specifically, a number of issues were found in the
electrical subsystem as well as the engine subsystem. In many cases, it is felt that the level of
modeling used in subsystems, the electrical subsystem being one excellent example, are more
complicated the necessary given the relatively low impact on the desired outcome.
From the supporting material, it is clear that the model did an acceptable job at modeling
a Class 8 SmartWay truck. Further validation across the range of vehicles being modeled would
be appropriate to provide confidence to the end users and ensure the model is doing an
acceptable job at modeling green house gas emissions.
Objective 2: Contains open source code, providing transparency in the model
Providing source code as a Simulink diagram is necessary for this objective but not
sufficient. Additional documentation on the equations and references behind the Simulink code
should be developed and released to the public. Even an experienced Simulink user finds it
difficult to follow somebody else's code. The code provided is actually laid out quite well but
more documentation is necessary to avoid confusion. Inexperienced Simulink users would not
be able to follow the code directly and thus would rely much more heavily on the supporting
documentation. In later sections there is come critique regarding the current GEM manual in
how it describes certain aspects of the model. These issues should be addressed as
documentation is refined.
Shawn Midlam-Mohler GEM Review Page 5
Page 79 of 104
-------
3: to by
The compiled version of the code is free and easy to use. The Simulink version requires
a Matlab license which is not free but fairly common in industry.
4:
The current structure satisfied this objective.
5: by the for
By releasing an official and unalterable executable version of the model this objective is
met. Providing only a "source-code" version (i.e. Simulink code) would be problematic from
many perspectives.
Shawn Midlam-Mohler GEM Review Page 6
Page 80 of 104
-------
The overall approach of using a relatively simple model structure based in Matlab-Simulink
is sound provided that models are calibrated and validated to a sufficient level. In the following
subsections, there are comments on various issues found in the various sub-models in the model.
The following is a summary of what follows:
1. Ambient Subsystem: No issues were found in this very simple subsystem.
2. Driver Subsystem: No major issues were found in this subsystem.
3. Electrical Subsystem: Several serious problems were found in this subsystem. Most
notably, there are serious flaws in the battery model, the alternator model, and
alternator control.
4. Engine Subsystem: There were problems found in this subsystem which need
addressed. The main concerns in this subsystem are from the method use to model
the engine at negative brake torque values.
5. Transmission Subsystem: No major issues were found in this very simple subsystem.
6. Vehicle Subsystem: Some issues were found in this subsystem.
The ambient subsystem contains only parameters to describe the ambient conditions.
There were no relevant comments on this subsystem.
The driver subsystem is typical of those found in other models of similar fidelity. There
were no major issues found within the Driver Subsystem. The following subsections contain
some comments on models or controls within this subsystem.
The manual describes that the driver block in a misleading fashion. Once sentence in
particular: "The search for the proper vehicle speed occurs at every simulation time step." This
seems to imply it is something other than a simple PID control.
Shawn Midlam-Mohler GEM Review Page 7
Page 81 of 104
-------
From experience, there are times when the PID gains for a driver may need to be adjusted in
order to drive a particular velocity profile. The PID values are fixed in the current model. If an
end-user has a vehicle in which the driver does a poor job there is no recourse to correct this. It
may be worth adding this as an "advanced feature" or using a more sophisticated control
concept. For example, the driving trace is known as are the overall vehicle characteristics for
each class, it would not be terribly difficult to augment the current PID control with a
feedforward component. This being said, large errors in velocity tracking were never observed
in exercising the model.
The gear shifting strategy was only evaluated by observation. It appears to follow the
prescribed shift schedule as desired.
Very significant issues were found in the electric subsystem which require attention. In
particular, the battery model appears to an error which causes battery voltage to decrease with
battery state of charge which is exactly opposite of the desired behavior. Furthermore, it appears
that the sign convention used for the starter, accessories, alternator have the wrong sense. The
alternator generates negative current which decreases SOC. The other two currents, which are
current sinks, actually increase the SOC of the battery. Even with the above issues aside, the
alternator model appears to not consider the mechanical to electrical efficiency of the device and
the control is naive of actual alternator capabilities and control. These issues and others of more
minor consequence are described below.
Many of the model parameters used in the electrical system are not changed based on
class of vehicle. Many of these would change based on the class of vehicle.
The alternator model is particularly difficult to follow from the Simulink code. It appears
that alternator current is directly a function of speed, which is not correct for modern alternators
Shawn Midlam-Mohler GEM Review Page 8
Page 82 of 104
-------
which can regulate voltage quite effectively. Figure 1 shows simulation results for a 45 mph
simulation case. This shows a few strange behaviors: 1) the voltage drops to 10 volts by 500
seconds (the vehicle starts moving at 375 sec.); 2) the behavior of the voltage is erratic and not
typical of what happen in practice. The second point is a direct result of the control that is
applied to the alternator model in that it turns the alternator on at full rated capacity until it
reaches a setpoint and then turns it off until voltage drops below a setpoint. This will result in
r\
much higher internal I R losses than a more appropriate and more realistic model/control that
allows the alternator to actually modulate current.
Class 8 Truck at 45 mph
-ง13
1500 2000 2500
!lme (s)
3500 4000
Figure 1: Irregular Voltage of Battery
It appears that the alternator torque is only a function of alternator electrical power demand
without accounting for the alternator efficiency. This part of the model is shown in Figure 2. If
this is the case then the model is underestimating the accessory torque required to operate the
alternator. In looking through the m-file associated with the alternator there was no obvious
efficiency parameter for the alternator which further raises doubt.
Shawn Midlam-Mohler GEM Review
Page 9
Page 83 of 104
-------
Figure 2: Possible Error in Alternator Mechanical Torque Calculation
11 i i ' i i , i '' i ,'
Given the relative unimportance of the starter in the overall performance of the model,
the starter model is quite complex. With this level of model, it is clear that parameters should
change with the engine class - this is currently not implemented in the model.
Investigating the battery model independently led to the discovery of extremely
disturbing behavior. With the battery removed and the SOC initialized at zero, a 1-C charge at
352 amps at 20 deg. C was simulated. The battery SOC moved from 0 to 100 in roughly 3600
seconds, which was expected. What was not expected was that the value of "ees_volts" behaves
exactly counter to what it would in an actual battery - with increasing SOC the voltage drops
very quickly to a minimum value and stays there. The open circuit voltage, which is map based,
behaves as expected. Figures showing the results of this test are shown in Figure 3. This
behavior was observed in the vehicle simulation as well although it is difficult to observe
because of the other dynamics involved.
1 -C Battery Charge from 0% SOC
1 -C Battery Charge from 0% SOC
Shawn Midlam-Mohler GEM Review
Page 10
Page 84 of 104
-------
1-C Battery Charge from 0% SOC
0? 03 (M 05 OS 07 06 OS
SOC
Figure 3: Junction of Three Currents
Further investigation of the electrical system model yielded further inaccuracies. In the
model, there is a junction of three different currents: starter, alternator, and accessory current
(Figure 4). If one disconnects the alternator current and leaves the starter and accessory current
connected (i.e. disable the ability to charge the battery) one finds that the battery SOC increases.
If one disconnects the loads and applies an alternator current manually (required because of the
alternator control and initial SOC) you find the SOC decreases. In both of these cases the
"ees_volt" value goes the opposite direction of the SOC.
Figure 4: Junction of Three Currents
The battery appears to be unnecessarily complicated with respect to the objective of the
model. In particular, modeling the thermal dynamics of the battery seems excessive. Over the
transient cycle for a Class 8 truck, the battery changes temperature by less than one degree.
Generally, more complicated models than necessary require more calibration parameters and
could be more prone to inaccurate results. This level of complexity seems unnecessary.
Shawn Midlam-Mohler GEM Review
Page 11
Page 85 of 104
-------
Ot'Tiilnl 'tcre^TOrik^; not Adjusted In* Fell if!** Oh<;ซ;
The electrical accessory load is not adjusted by vehicle class. The electrical accessory
loads are not constant between classes.
The issues found in the engine subsystem are not as serious as those in the electrical
subsystem, yet they still need to be addressed. The method of handling negative brake torques in
the model does not seem to be appropriate. Because the engine model is one of the most
important in the simulator it must be as accurate as possible. Although not a technical flaw,
many of the variable names in the model are confusing or irrational, such as "closing throttle
torque" and "closed throttle torque" - use of such language leads one to question the model
structure and calibration.
The manual's description of the engine model is misleading. In particular, the sentence
"This map is adjusted automatically by taking into account three different driving types:
acceleration, braking, and coasting." This text is not very descriptive of what is actually in the
model.
Closed "throttle" is an inappropriate way to describe this parameter for a Diesel engine.
Diesel engines can have throttles but they are used for purposes other than load control. The
values seem to be the identical for each of the engine classes as well as being contrived numbers
since it is precisely equal to -5. This would impact the rate of deceleration and potentially have
an impact on overall fuel economy predictions.
There is no implementation of fuel cut-off during decelerations. This is a feature that is
implemented on at least some heavy-duty Diesel engines. This can be observed by plotting the
fuel flow rate during the transient cycle.
Shawn Midlam-Mohler GEM Review Page 12
Page 86 of 104
-------
The method used to calculate "closed throttle" fuel consumption is not clear. The part of
the code which does this is shown in Figure 5. The use of variable such as "closed throttle
torque" and "closing throttle torque" do not inspire confidence in the model as they are non-
standard terms - particularly for a Diesel engine. It is difficult to understand exactly why this
calculation should result in a valid fuel flow.
It is possible in a lab setting to measure fuel consumption from max rated torque down to
zero brake torque. With a motoring dyno, it is then possible to measure fuel consumption at
negative brake torques until the engine reaches a condition where it injects a minimum amount
of fuel, or in many cases, absolutely no fuel. It is understood that not all engines will have these
"negative brake torque" fuel maps available, however, there are approximate ways of modeling
this, such as techniques based on the popular Willans Line method.
"-^ dojce
- 3^- *
d
d
ajte
cpS
3- j t- e
d-^>(T)
f
1
, qotf
:*"
^
Bl_flCtt] ^>-|
1
dOJB*
,->
n
db- j D e
P-
P
'
raduc
Figure 5: Closed-Throttle Fuel Flow Calculation
A map-based engine model should be sufficient to achieve the desired objectives. The
engine model implemented in the current version of the software does not appear to be as well
implemented as it could be. Given the importance of this in the overall objectives of the
simulator this needs to be addressed. Using fuel maps which have torque indices ranging from a
negative brake torque to the maximum rated torque would alleviate much of the uncertainty in
the model. Driver accelerator requests should then be linearly scaled from minimum value to the
maximum value on this map with the exception of idle conditions in which alternative measure
must be taken. This approach also automatically takes into account deceleration fuel cut-off as
well.
Shawn Midlam-Mohler GEM Review
Page 13
Page 87 of 104
-------
The mechanical accessory load is not adjusted by vehicle class. The mechanical
accessory loads are not constant between classes.
There were no serious model issues found in this subsystem. As with many other
subsystems, there are a number of parameters which should change with vehicle class.
There are many transmission parameters which currently do not depend on vehicle class,
such as clutch and gear inertias. In vehicles which across this range of classes the inertias are
likely much different. These parameters can be found in "transmission_manual_param.m".
There were no modeling errors noted in the vehicle subsystem, however, there are a
number of things which could be taken as recommendation. The most serious item is considered
to be the fact that the "Vehicle Weight Reduction" parameter is specifically cited as being able to
model light-weight wheels. The existing model structure would not accurately do this as it does
not take into account the inertial aspect of the wheels which would have a greater impact on the
vehicle.
There are many vehicle model parameters which currently do not change with vehicle class
and should. There are a number of driveline component inertias which do not appear to change
with vehicle class. These parameters can be found in "vehicle_param.m".
The impact of tractor cab design is one of the key technologies that this simulation is
intended to evaluate. The equations used to model drag is the typical 0.5 * Cd * A * velocity A
2. The proposed approach constrains the fontal area (A) to fixed values that depend on vehicle
class. This could dis-incentivize novel cab designs which result in smaller frontal areas for a
Shawn Midlam-Mohler GEM Review Page 14
Page 88 of 104
-------
given class of vehicle. It is recommended that allowing the frontal area be an input parameter to
the model. In certain disciplines it is common to parameterize a model using a lumped Cd*A
term because of their interrelation.
The "Vehicle Weight Reduction" parameter is described as a way to accommodate,
among other things, lighter weight wheels. Simply subtracting wheel weight from payload will
underestimate the impact that light-weight wheels will have on the vehicle because it neglects the
rotating inertia of the wheels. This could be accommodated given information on the rotating
inertia of the wheels and subtracting it from the appropriate tire inertias in the mode - this would
be in addition to the weight reduction already implemented.
The code used to adjust vehicle mass for weight reductions does not do so for many of the
vehicle classes. This is shown in Figure 6 below.
Figure 6: Vehicle Weight Reduction Code from run_preproc.m
The vehicle loss parameters used, mainly rolling resistance and drag coefficient, use very
basic models. Essentially, the rolling losses are characterized entirely by a single constant per
tire and a single drag coefficient is used to model the aerodynamic losses. Relying on a single
Shawn Midlam-Mohler GEM Review Page 15
Page 89 of 104
-------
parameter may not be sufficient to model these losses accurately. An alternative would be to
allow alternative forms of entering this data. It is understood that these standards are being
under development - but it is certainly possible that these parameters are not well modeled by a
constant.
Shawn Midlam-Mohler GEM Review Page 16
Page 90 of 104
-------
The directions provided for the peer review requested some specific information regarding the
input and output of the mode. The following subsections address these issues.
of
The .xml format used in the output file will be problematic for some users. Most
operating systems opt to open .xml files with programs (MS Word, internet browsers) which do
not meaningfully displace the results. The manual states clearly that MS Excel should be used to
open the file, however, certain users may not head this warning. It may be beneficial to remind
the user from the software after they click the "RUN" button on the compiled code.
of Fie
The formatting of the output file was clear. The four tab format with the first tab being
summary data and others being cycle data was sufficient.
of
End users will likely want to see more detail in the output file then just the vehicle target
speed and achieved speed. Making a limited number of "internal" parameters available to allow
end users a glimpse inside the model without having to use Matlab-Simulink would be sufficient.
These should be limited to things relevant to their inputs, such as aerodynamic drag over the
cycle, rolling losses over the cycle, etc.
in
It was requested that reviewers comment the proposed standard parameters for the
different vehicle classes shown in the GEM manual. Unfortunately, the reviewer does not have
the required expertise to make an assessment of the proposed values.
Shawn Midlam-Mohler GEM Review Page 17
Page 91 of 104
-------
The following subsections contain observations which did not fit into the previous sections.
for
A number of parameters were noted which should change with respect to the vehicle
class. The reviewer is certain that there are others that were not noted in this review. It is
recommended that the EPA investigate this and take an appropriate action. In many cases, these
components will not have a serious impact on the overall performance of the vehicle. By way of
example, many of the inertias simulated in the model will not have a large impact on the results
in contrast to the large inertia of the vehicle. If this is the case, then these inertias could be
discarded from the model with little impact on performance. If the detailed inertias remain in the
model, then they should accurately reflect the vehicle class.
One overall comment is that there is a higher than necessary level of fidelity in many of
the models. By way of example, the battery model is particularly complicated and contributes
very little to the outcome of the simulation. There are also a great number of relatively small
inertias that are modeled, such as the starter motor inertia. These inertias contribute very little to
the type of results that are sought after in this simulation.
The added level of detail also comes with an additional practical consideration in that the
models require a great deal more parameters to describe the vehicles in each class. By way of
example, the starter inertia of a Class 2b truck is much different than a Class 8 truck. If the
starter is modeled as a zero inertia element, then it does not need a defined inertia. If there is an
inertia parameter, then it should be a representative number even if it does not have a major
impact on the simulation. EPA could reduce the complexity of many of the models with little
impact on the accuracy of the simulation - this would then lead to a reduced set of parameters
that very with vehicle class and therefore need to be determined.
of
It would be useful to have a better understanding the propagation of error in the input
parameters. For the proposed configuration for the class 8 high-roof sleeper cab the sensitivity
Shawn Midlam-Mohler GEM Review Page 18
Page 92 of 104
-------
of the CO2 result to errors in Cd is approximately 50%. This implies that a 10% error in Cd will
result in a 5% error in prediction of CO2 emissions. For rolling resistance, the impact of a 10%
error in the tire rolling resistance causes a 2.3% error in prediction of CO2 emissions. These
sensitivities should be compared to the reduction in CO2 emissions required as well as the
accuracy of the key input parameters in the model. This analysis would also be useful in
determining which parameters might be superfluous with respect to the desired output. As
discussed above, there are some models which likely have more complexity then necessary.
Shawn Midlam-Mohler GEM Review Page 19
Page 93 of 104
-------
Conclusions
The overall modeling fidelity and structure of the GEM model should be able meet the objectives
of the EPA. There are a number of issues with the current version of the GEM model which
would need to be addressed to best meet the objectives. These issues were described in greater
detail above, but in summary fit mainly into the following major points:
1. Accuracy of Sub-Models Structure: A number of errors were found in models within
GEM. None of these errors are expected to contribute to larger errors to the output
results but should be corrected nonetheless.
2. Parameter values for Different Vehicle Classes: Throughout many of the subsystems
there are a number of minor parameters which should be changed with vehicle class.
Philosophically, if the model has a parameter that should change with vehicle class then it
should change with class. If it is determined that changing the parameter with vehicle
class has not significant impact on the model results then that parameter should be
considered for elimination.
3. Model Complexity: Several of the sub-models had complexity that far outweighed their
impact on the results. The battery was one such sub-model which also contained some
serious errors in its formulation. Many of these models could be simplified which will
also reduce the number of parameters required which impacts the comment in (2) above.
4. Sensitivity Analysis: A rigorous study of the sensitivity of key input parameters should
be conducted. Our ability to measure and estimate input parameters is not perfect, hence,
the output of the model is affected by this uncertainty. If our ability to measure the
coefficient of drag is +/- x.y % then that has an impact on the model output. This
uncertainty can then be compared to required accuracy to make a judgment on the
validity of this method at estimating green house gas emissions or fuel economy.
5. Model Validation at other Classes: Based on the issues noted in (2) above, it is important
to validate the model across vehicle classes. Because the model structure is relatively
low-fidelity it has a greater burden of proof when "extrapolating" results. To have
confidence in the model some further level of validation should be conducted.
Shawn Midlam-Mohler GEM Review Page 20
Page 94 of 104
-------
EPA GEM Peer Review
Elliott Ortiz-Soto
PhD Pre-Candidate
Mechanical Engineering
University of Michigan -Ann Arbor
The following document reviews the Greenhouse Gas Emissions Model (GEM) from a user's
point of view, as well as providing a more detailed evaluation of the modeling approach and
assumptions. The review first addresses the executable version of GEM as a black-box
simulation from an end-user standpoint, commenting and suggesting improvements on the
overall GUI layout, usability and output. The second part looks into the Matlab/Simulink
version of GEM. In this case, I attempt to give a more thorough and detailed evaluation of the
code, assumptions and underlying physical models. The review is organized as a bulleted list,
and it does not follow a particular order, although I did try to arrange the comments by similar
subjects. I hope this review provides some useful feedback in the development and improvement
of the GEM compliance simulation tool.
GEM Executable and Output
This section provides general feedback on the GEM executable and its output.
The location of the "Vehicle Model Year" dropdown menu is not intuitive. This is one of the
most important parameters of the simulation and it is part of the inputs that affects the results,
but it has been grouped with the identification parameters. These should be separated as they
currently are, but somehow the "Vehicle Model Year" was left in the top section.
Having radial buttons with all of the vehicle configurations in the "Regulatory Class" section
is not necessary. It occupies space and reduces the GUI's flexibility to add other parameters
in the future. This type of list is probably better addressed through the use of a drop down
menu. It would reduce the profile of this parameter list, and it would show much more
clearly what vehicle type is being used. Currently, closer attention has to be paid to the GUI
to notice which radio button of the ten available is selected, whereas with the dropdown
menu it is only necessary to read what is displayed.
On the other hand, it is not clear why there should be a dropdown menu for the "Coefficient
of Aerodynamic Drag" parameter. Furthermore, the dropdown menu allows the values to be
overwritten by the user, so the dropdown menu has no real purpose. Typically dropdown
menus are used to provide the user with a set of fixed options, which are usually not
numerical values. A better approach would be to just provide a sample value in the
parameter name to give the user an idea of what would be an expected input in the box.
Page 95 of 104
-------
Basically, it should look something like the "Steer Tire RR" and "Drive Tire RR" input
boxes.
The same issue is seen in the "Speed Limiter" input box. There is no real reason why there
has to be another dropdown menu. If there are minimum and maximum values for the speed
limiter, this should simply be stated either in the GUI or in the documentation, and just allow
the user to input whatever integer value they require within these limits.
A similar observation can be made regarding the "Extended Idle Reduction" parameter.
According to the documentation, this parameter is an on/off option. Providing another
dropdown menu, which can also be overwritten, is simply confusing. This gives the
impression that any number of values, maybe between 0 and 5, can be used as inputs, which I
am not sure is the case here. A checkbox object should be used for this type of on/off
parameter.
It appears that some options are only available when a certain "Regulatory Class" is selected,
such as the "Vehicle Speed Limiter", "Vehicle Weight Reduction" and "Extended Idle
Reduction". But from the GUI, it is not clear which ones can be selected with the various
vehicle types. It is generally useful to gray-out the options that are not available in relation to
another parameter. For example, if one of the vocational vehicles is selected as the
Regulatory Class, the three options mentioned above should be grayed-out, letting the user
know unambiguously that these are not available with this vehicle class. Currently, it is not
clear whether the code is robust enough so that these options are not applied when a certain
vehicle is selected, or if you would just obtain incorrect results if these were to be selected
unknowingly.
One significant drawback I found relates to the output file naming scheme. First of all,
naming the files based on date and time is not very useful or descriptive. When multiple
simulations are performed, it becomes difficult to determine what file you should be looking
into, unless you actually open it. The file names should include at least some sort of
indication of what the simulation configuration was. The second problem I found was the
lack of flexibility to specify where these output files are saved. There should be an option
allowing the user to browse and select the main directory where these files are to be saved.
As a final comment on this, there is really no reason for each of these files to be saved to a
different folder if there is just a single output file. This simply adds an unnecessary layer to
the file structure. If multiple outputs were generated, then it would make some sense, but
currently, there is a single xml file within a folder with the exact same name.
When the simulations are run, a series of plots with the drive-cycle profiles are generated. It
is not very practical to have to close each of these in order for the next one to show. These
should either be generated in different windows or, preferably, in a single tabbed window
with all three plots. It should also not be necessary to close the plots for another simulation
Page 96 of 104
-------
to begin. This way, various simulations could be performed and the users could clearly see
how the parameters varied affect the vehicle behavior without having to create the plots
themselves. A small table with some drive-cycle output, such as the one given in the xml
files, would also be useful to see together with these plots.
Within the output file, there are three sheets with the drive-cycle traces. Plots should be
automatically generated because the explicit profile is not of much use unless it is plotted.
Miles per gallon (MPG) is generally assumed to be a measure of fuel economy, not fuel
consumption. Although this is a small detail, it might be worth revising to be consistent with
the industry standard. Adding the fuel consumption equivalent in gallons/ hp-hr (liters/hp-hr)
or liters/100 km might also be useful.
For compliance purposes, it would be good to see the actual target value next to the
simulation result, and probably some sort of percentage difference between these. It would
give the manufacturer/user an idea of how their product performs with respect to the
expected regulation standard.
Although the idea of the current program is to reduce complexity and provide only the
necessary information for compliance purposes, some additional results might be helpful for
manufacturers to determine if the simulation is representative of their vehicle. Because many
model parameters and vehicle operating strategies have been standardized using internal
assumptions and algorithms, the overall behavior of the vehicle in question could end up
being very different from what the vehicle manufacturer actually observes. This can result in
a significant over-estimation of fuel consumption and COz emissions, and possibly non-
compliance. For this reason, it is fair that the manufacturer be able to assess the validity of
the simulation without having to investigate the model in detail. This could be achieved by
providing a series of additional results, which could be related to the engine operation over
the drive-cycles, the shifting strategy, the electrical system, etc. Exactly what parameters
these should be might not be so simple to determine, but it could provide some confidence in
the simulation results.
Matlab/Simulink GEM Model
This section provides an evaluation of the GEM Matlab/Simulink model and simulation.
General comments on the GUI and Matlab simulation:
o The internal names (tags) for the objects (buttons, dropdown menus, etc.) in the
Matlab GUI script should be more explicitly named for clarity and understanding of
the GUI functionality.
o Default parameters should be assigned in the GUI opening function so the user has a
better idea of what to select or provide as input.
Page 97 of 104
-------
o The date is not automatically imported as in the executable, and the simulation
crashes if the user forgets to write it in.
o If the model is run with any of the Stateflow blocks open, it increases the simulation
time substantially.
In general, the Simulink model is well organized and intuitive. The use of the following
modeling techniques and Simulink components make the model particularly elegant and easy
to understand:
o Multiple "Bus" elements and collecting them into a "System Bus" to keep signals
clearly labeled and organized.
o "GoTo" tags to avoid excessive model clutter with connections between blocks.
o Stateflow instead of explicit Simulink logic blocks, which greatly simplifies
development and implementation of the various logic controllers.
o Signal-activated blocks to avoid additional logic blocks for signal generation and
routing.
Even though the use of various "mux" and "demux" components, as well as a series of
component Buses and an overall System Bus is a very elegant modeling approach, it seems
that many of the signals are also being output separately, which somehow defeats the purpose
of having a Bus. I am sure there is some reasoning behind this, but I would have expected
this Bus component to be used more widely, routing the signals directly to/from the Bus
everywhere they are required.
Some blocks go into deeper levels unnecessarily. Examples can be found in the electrical
system and in the driver models. Although the approach used in this model of grouping
models into blocks based on their physical components or functionality is fairly intuitive,
adding extra layers can also make the model more difficult to follow if done excessively.
Is it necessary to have an "Ambient Bus"? The Bus component is used to collect signals
calculated in the model, whereas the ambient parameters are all prescribed and fixed. They
could just as easily be called as a variable from the workspace wherever they are needed (this
seems to be what is typically done in the model anyway).
In the pre-processing file, where the parameters for the individual configurations are selected,
there appears to be a lot of repeated code and "if/else statements". Most of these parameters
can simply be collected in arrays, which can then be indexed using the "veh_type" variable.
This way they can be included in the original parameter files, as they are more explicit and
easier to read by a user than having to review a long pre-processing script with many
conditional statements. It would also take advantage of Matlab's array operations, which are
Page 98 of 104
-------
usually more efficient than "for loops" or "if statements", as well as removing a lot of the
repeated code. This will most likely result in improved code performance.
Control for most of the vehicle components seems to be achieved by fairly standard PID
controllers. Usually the gains for these controllers are tuned to a specific plant, but in this
case they remain fixed for all the vehicle configurations. Were these gains tuned for all the
plants individually and then somehow averaged to account for all of them, or were they
computed for a single vehicle? Although for the test cases do not show any major problems
with following the prescribed velocity profile, simulation of some vehicles or with a different
set of parameters could possibly suffer if the controller gains are not appropriate. For the
driver, for example, more elaborate, robust and reusable driver models exist, and it might
useful to investigate the possibility of incorporating one of these in order to avoid possible
issues with the simulations.
In a related comment, Simulink offers pre-developed PID blocks in which only the gains
must be prescribed. Is there any particular reason why the PID controllers have been
explicitly created? It might help reduce the profile of the individual models if these were to
be employed.
The engine speed appears to be calculated within the Accessories block, which is not very
intuitive when reviewing the model. I would expect this to be within the main Engine block
and then passed to the other engine-related blocks from there.
In both the Gear and Clutch blocks of the Transmission model, it is assumed that the gears
and clutch are either fully engaged, where they pass the total torque being input, or fully
disengaged, where they pass zero torque. Although this might be a fair simplification for the
given modeling purposes, there are simple models that can calculate the transmitted torque
based on the gear/clutch slip or speed differential. This will add a little more complexity to
the model, but it should result in more realistic vehicle behavior.
The electric components and EES seem to be fixed for all the vehicles in the simulation, but
in reality the electrical system is probably designed for a given application to account for the
particular load requirements. It is understandable that due to the complexity of acquiring
parameters such as these, the system model is standardized, but it could also result in
simulation inaccuracies. It might be more appropriate to provide at least some basic scaling
capability for the overall electrical system so that with one or two additional inputs, the
electrical components and EES are scaled to match the actual setup more closely.
A similar observation can be made regarding the starter and alternator models. Both appear
to be parameterized based on HD Class 8 components. Does this mean that these
components are oversized when used in smaller vehicle classes? If so, would they not impart
a larger load on the engine, or require a larger amount of electrical power to operate when
Page 99 of 104
-------
compared to a right-sized component? These are most likely not critical components of the
model due to the drive-cycles being used, but again, a scaling factor, even if it is an internal
value, should be applied to ensure these are representative of the actual system in the vehicle.
There seems to be an internal option for an acceleration test. Will this be made available to
manufacturer users? Acceleration tests are in general much simpler to perform than a full
transient drive-cycle, so providing this optional capability might give the manufacturers
another way of validating the model. If acceleration numbers are completely different, then
it would be hard to expect that a transient drive-cycle simulation would be at all
representative of the real vehicle.
One of the most important input data for a fuel economy drive-cycle simulation is the engine
mechanical load and fuel consumption maps. The mechanical load maps are usually simple
because only the WOT (or Diesel equivalent) values are required, but obtaining full range
fuel consumption values is much more difficult. Several engine maps appear to be available
for each vehicle class, but making these completely standard with a prescribed displacement
volume and operating range might be a limiting factor for some manufacturers. A more
flexible approach would be to have normalized load and fuel consumption maps, given in
BMEP and BSFC values. The current maps can be easily converted into BMEP and BSFC
with the data available. The user could then provide the engine displacement and possibly
another key parameter such as rated torque or power and the engine speed, and an algorithm
could automatically manipulate the normalized maps to obtain more representative absolute
values for the engine in question. Even though this compliance tool assumes that the engines
have already been certified, the fuel economy and CO2 values that the simulation predicts are
directly related to the maps given, and manufacturers might want to ensure the engines in
their vehicles are properly accounted for.
The closed throttle or motoring torque in all of the engine maps is -5 N-m, except at the idle
speed. This might be a reasonable simplifying assumption, but in general the motoring
torque increases with engine speed due to the rise in friction. It might be worth adding some
sort of speed dependence to ensure correct engine decelerating behavior during non-fueling
conditions.
The shifting strategy can also be considered a significant factor affecting vehicle behavior in
a drive-cycle simulation such as this. Moreover, they tend to be very specific to the
combined engine/vehicle configuration, making them hard to obtain from manufacturers or
extrapolate from ones currently available. The shifting strategy shown in the transmission
parameter file is only a function of vehicle speed, whereas shifting, in general, is load
dependent as well. When load is included as a dependency factor, the shifting strategy has to
be related directly to specific engine map. Most likely these shifting speeds are for WOT,
but at lower loads the strategy tends to be slightly different to maximize fuel economy. In
my experience, it is possible to develop reasonable shifting maps optimized for fuel economy
Page 100 of 104
-------
based on a given engine map. This is usually achieved by finding the most efficient engine
speed at various engine load intervals, thus creating an optimum engine operating line, which
can be related to the vehicle speed through the various gear ratios. The goal of the shifting
strategy is then to maintain the engine operating as close to this line as possible. This
approach works well for low/mid load operation. For high loads, where acceleration and
gradeability are more important, the shifting maps should be corrected, which can be done in
this case using the available WOT shifting speed numbers. Internally generated shifting
maps would also allow for engine map scaling as mentioned above, without requiring new
shifting strategy data.
As part of the simulation output and the suggestion for some additional data provided to the
user, it would be interesting if plots of the engine map and shifting strategy are included. A
simple assessment of these could give the user a good idea about the appropriateness of the
given modeling assumptions for their vehicle setup being evaluated. As an extension of this,
the various drive-cycle visitation points could be plotted on the engine map as well.
Some models, such as the electrical system, appear to be extremely complex and detailed for
this type of dedicated simulation. Unless there is a particular reason, such as future
extensions to GEM for hybrid-electric trucks or different drive-cycles, where such details are
necessary, then the electrical system model can probably be stripped down substantially
without sacrificing much fidelity in the simulation.
Similarly, the Stateflow engine logic controller contains some states, such as the ones with
the engine off, which are probably not seen in any of the simulations, except at the first time
step. There is no stop-start functionality or cold-start behavior, so it might not be necessary
to have a full starter model and the engine logic could be somewhat simplified. Another
related model simplification could be removing the idle controller. A saturation block in
Simulink could be used to limit the engine operation to a minimum idling speed without
having an additional controller that can end up slowing down the simulation and increasing
the complexity of the model.
There is a block in the engine model called "trans gear shift" whose output does not appear to
be actually used anywhere. This block also has a PID controller. It is not clear to me why
this block is needed, but if it is not, then it should definitely be removed to prevent the
controller from slowing down the simulation unnecessarily when the block is activated.
In general, the rest of the model looks good. I have looked into the various submodels, in
particular for the engine, transmission and vehicle, and they seem to follow the correct
approaches. Overall, the model is in great shape and should be a strong starting point for a
dedicated simulation oriented to compliance purposes.
Page 101 of 104
-------
EPA's Response to Peer
Reviewer Comments
Page 102 of 104
-------
EPA Response to Peer Review
Overall, the reviewers' comments toward the Greenhouse gas Emissions Model (GEM) are
positive and constructive, which can be summarized as follows:
Accuracy of systems (i.e., driver, electric, ambient, engine, vehicle and transmission)
Parameter sensitivity study related to coefficient of aerodynamic drag (Cd) and
coefficient of tire rolling resistance (Crr) to fuel economy or CC>2 emissions
More model validation
Model documentation
Equations, references, etc.
At a component level, it can be further summarized as follows
Driver model with better feed forward components and more configurable PID
(proportional-integral-derivative) gains
More consistent electric and accessory models
Environment model with more realistic air density and ambient temperature
Engine model improvements
Many changes have been made since GEM was first released to the public. One of the
key changes is the driver system model. The enhanced system uses a target vehicle driving
speed to estimate vehicle torque demand at any given time. Then, the power required to drive
the vehicle is derived to estimate the required accelerator and braking pedal positions. If the
driver misses the vehicle speed target, a PID controller applies speed correction logic that adjusts
accelerator and braking pedal positions for matching targeted vehicle speed at every simulation
time step. The driver system, with its feed-forward driver controls, more realistically models
driving behavior.
Electric system model is modified to use constant electrical power to simulate vehicle
electronics power consumption. The values for electronic and accessory power consumption are
modeled as constant over all classes of vehicles. The reason behind modeling power
consumption in such a manner is that the certification with use of GEM is done on a relative
basis by comparing the new vehicle model result with the pre-selected engine and vehicle result,
where all vehicle models use the same electrical and accessory power. The difference in
selecting electrical or accessory power consumption is not critical and has no impact on the final
certification results. Since GEM is not used to model absolute vehicle emissions, assigning
default parameters in the model achieves this objective, even if the absolute emissions may differ
from those predicted. In other system-level development, the value for ambient density of air
has been changed to represent more realistic conditions, in accordance with standard SAE
practices.
Page 103 of 104
-------
All bugs noted by the peer reviewers have been identified and fixed with the exception of
the implementation of an algorithm for "deceleration fuel cut-off during zero throttle
deceleration. The agencies recognize that different manufacturers have different fuel cut-off
control logics and it would be challenging to implement all control logics without manufacturers
providing the data for final model validation. Consequently, we are delaying implementation of
a fuel cut-off strategy until a future rulemaking.
In this phase of rulemaking, the agencies have decided to regulate engines and vehicles
separately, except for heavy-duty pickups and vans. We believe this separation is the most
appropriate way to achieve the near-term reductions without introducing substantial new testing
burden on heavy-duty vehicle manufacturers. In the future, though, it may be desirable to certify
vehicles and their engines to a complete vehicle standard using a complete chassis test procedure
or a more fully-integrated vehicle model to determine emissions levels for combination tractors
and some vocational vehicles. At that point, it would be necessary to use fuel maps specific to
the engines installed in the vehicles being certified. In this first phase of GHG emission
regulation, though, the GEM model uses fixed engine maps to prevent double counting emission
reductions from engine improvements (which are subject to compliance with engine standards)
and then again in the truck model (to comply with vehicle standards). Further, in direct response
to reviewer comments on the GEM engine system, the engine brake torque value at the closed
throttle position is no longer negative in the engine fuel map.
As described in the Regulatory Impact Analysis (RIA) Chapter 4, GEM has been
validated and benchmarked against test data as well as other well known vehicle simulation tools
since GEM was first released to the public. We extended the model validation to both Class 7
and Class 8 vehicles using test data. We also benchmarked GEM's model prediction against the
GT-Drive model which is commonly used in industry.
A sensitive analysis of coefficient of aerodynamic drag (Cd) and coefficient of rolling
resistance (Crr) was conducted following the reviewers' comments. The study shows that the
vehicle behavior follows an almost linear relationship between these input parameters and CO2
emissions. Charts in RIA Chapter 2 show the linear trend of the GEM inputs relative to the CC>2
emissions results. Nonlinearity is fairly weak in the range of variation of those input parameters.
Therefore, it is acceptable for the GEM to take as inputs a linear average of fleet Cds and a linear
average of tire rolling resistances, or Crr.
The agencies fully recognize the importance of the transmission to overall vehicle
performance. However, as noted in the peer review, GEM is not designed to model different
transmissions. Likewise, transmission improvements are not part of the technology package on
which the GHG emission standard for these vehicles is predicated. GEM's purpose is to quantify
the relative effectiveness of a limited suite of technologies and not to discern the absolute GHG
emissions or fuel consumption of whole trucks. As such, the agencies decided to model only
those parameters most easily associated with vehicle greenhouse gas emissions reductions. For
example, in a sleeper cab combination tractor, parameters identified to be the most significant
include Cd, Crr, weight reduction, governed vehicle speed, and extended idle reduction.
Transmission improvements could potentially be evaluated as an innovative credit and thus be
utilized for demonstrating compliance on that basis.
Page 104 of 104
------- |