External Peer Review of EPA's
OMEGA Model
Final Peer Review Summary Report
SEPA
United States
Environmental Protection
Agency
-------
External Peer Review of EPA's
Final Peer Review Summary Report
Assessment and Standards Division
Office of Transportation and Air Quality
U.S. Environmental Protection Agency
Prepared for EPA by
ERG, Eastern Research Group, Inc.
EPA Contract No. 68HE0C18C0001
Work Assignment 3-05/4-05
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.
OMEGA Model
NOTICE
4>EPA
United States
Environmental Protection
Agency
EPA-420-R-23-010
April 2023
-------
Peer Review Report
Work Assignment 3-05/4-05, Contract 68HE0C18C0001
CONTENTS
1.0 INTRODUCTION 1
2.0 BACKGROUND 1
3.0 PEER REVIEW PROCESS 1
4.0 SUMMARY OF REVIEWER COMMENTS 2
5.0 NARRATIVE COMMENT AND EPA RESPONSES 3
5.1 Comments Submitted by Mr. Sujit Das 3
5.1.1 General Comments 3
5.1.2 Specific Comments 4
5.2 Comments Submitted by Mr. John M. German 6
5.2.1 General Comments 6
5.2.2 Specific Comments 7
5.3 Comments Submitted by Dr. Jeremy J. Michalek 20
5.3.1 General Comments 20
5.3.2 Specific Comments 20
5.4 Comments Submitted by Mr. Richard A. Rykowski 41
5.4.1 General Comments 41
5.4.2 Specific Comments 42
APPENDIX A RESUMES OF SELECTED REVIEWERS A-l
Sujit Das A-3
John M. German A-ll
Jeremy J. Michalek A-17
Richard A. Rykowski A-59
APPENDIX B CHARGE TO REVIEWERS B-l
-------
Peer Review Report
Work Assignment 3-05/4-05, Contract 68HE0C18C0001
1.0 INTRODUCTION
This report documents the results of an independent external peer review of the U.S. Environmental
Protection Agency (EPA), Office of Transportation and Air Quality's (OTAQ), OMEGA version 2.0 model and
documentation.
ERG (a contractor to EPA) organized this review and developed this report. The report provides background
about the review (Section 2), describes the review process (Section 3), provides a high-level summary of
reviewers' comments (Section 4), and includes reviewers comments with EPA's responses (Section 5).
Appendix A provides resumes for the selected reviewers, and Appendix B provides the charge to reviewers.
Reviewer comments are presented exactly as submitted, without editing or correction of typographical
errors (if any).
2.0 BACKGROUND
EPA/OTAQ develops its programs to control C02 and other greenhouse gas (GHG) emission measurements in
onroad and nonroad vehicles and equipment, there is a continuing need to evaluate the costs and benefits of
any such regulations. As such, EPA has developed its Optimization Model for reducing Emissions of
Greenhouse Gases from Automobiles, or OMEGA model, to facilitate its analysis of the costs and benefits of
the control of GHG emissions from cars and trucks.
The EPA OMEGA model applies various technology packages to a defined set of vehicles to meet a specified
GHG emission target and then calculates the costs and benefits of doing so. The GHG target can be a flat
standard applicable to all vehicles within a vehicle class (e.g., cars, trucks, or both cars and trucks) or the
"target" can be in the form of a curve which varies the target as a function of a defined vehicle "fleet." GHG
emission targets are specified in terms of C02-equivalent emissions. They can simply be C02 emissions from
the tailpipe or can be a combination of tailpipe C02 and refrigerant emissions.
3.0 PEER REVIEW PROCESS
EPA tasked ERG with identifying four reviewers who had no conflict of interest (COI) in performing the review
and who, collectively, met the following selection criteria:
• vehicle operations and analysis, including the physical process of generating and controlling vehicle
emissions.
• linkages between mobile source emission modeling and transportation modeling and planning.
• application of current mobile source emissions models for analysis for regulatory purposes and/or
policy evaluation.
ERG initiated a search process, asking interested candidates to describe their qualifications and respond to a
series of "Conflict of Interest" (COI) analysis questions. ERG carefully screened submissions to identify a pool
of qualified, COI-free candidates. From this pool, ERG selected four experts (listed below) who collectively
best met the selection criteria. ERG contracted with the reviewers after EPA verified that they were
appropriately qualified.
• Sujit Das, MS, MBA; Senior Research Staff Member, Oak Ridge National laboratory
• John M. German, BS; Consultant, JG Automotive Consulting, LLC
• Jeremy J. Michalek, Ph.D.; Professor, Carnegie Mellon University
• Richard A. Rykowski, MCE; Independent Engineering Contractor, Environmental Defense Fund
ERG provided reviewers with access to a secure ERG FTP site to download all review materials and the
technical charge to reviewers (Appendix B). Prior to the start of the review, ERG organized and facilitated a
1
-------
Peer Review Report
Work Assignment 3-05/4-05, Contract 68HE0C18C0001
virtual meeting between reviewers and EPA to provide reviewers an opportunity to clarify their
responsibilities for the review. EPA provided background about the review materials and responded to
reviewers' clarifying questions. Reviewers then worked individually (i.e., without contact with other
reviewers or EPA) to prepare written comments. During this time, three reviewers sent a request for
technical clarifications and additional materials to ERG. ERG forwarded this request to EPA and provided
EPA's responses to all four reviewers. Reviewers completed their individual reviews and submitted their
written comments to ERG. ERG provided the individual reviewer comments to EPA. EPA then responded to
reviewers' comments and provided a response to comments document to ERG. ERG then prepared this
report, including the high-level summary of reviewers' comments (Section 4.0).
4.0 SUMMARY OF REVIEWER COMMENTS
This section provides a high-level summary of the comments submitted by the four peer reviewers (Mr. Sujit
Das, Mr. John German, Dr. Jeremy J. Michalek, and Mr. Richard Rykowski) on the OMEGA version 2.0 model
to analyze the effects of new GHG standards and policies for light-duty vehicles. EPA's charge to the
reviewers requested their expert opinions on the concepts and methodologies upon which the model relies
and whether the model correctly executes the associated algorithms. EPA's charge also asked them to
comment on specific aspects of the model's design, execution, outputs, and documentation.
High-Level Themes
All reviewers commented favorably that they appreciated the increased capability and complexity of
OMEGA2 over the previous OMEGA version. In general, the reviewers provided numerous specific detailed,
complex, and nuanced comments and recommendations that indicated a good understanding of the model's
design. The most common category of comments consisted of recommendations for improving the model's
documentation by adding further explanations or specifics to enhance the user's understanding.
The second most common category of reviewer comments concerned the model's overall approach,
including the functions of each module. Reviewers commented on specific details, recommended
improvements, and noted unexpected results where observed. In several cases, reviewers noted specific
unexpected results and recommended that EPA should either repair the model's respective function or
improve the documentation to explain that the function in question intentionally produces the unexpected
result. The vast majority of reviewer recommendations for new or additional functionality focused on specific
enhancements of the existing modules; reviewers did not recommend additions that deviated significantly
from the current model's scope. EPA will likely be able to address almost all recommendations using
currently available information and knowledge.
Most Common Specific Recommendations and Comments
Certain topics were raised by multiple reviewers. For example, all reviewers commented on some aspect of
the model's handling of greenhouse gas (GHG) credits, especially as it relates to manufacturers banking these
credits from one year to the next and, in some cases, how credit banking would interact with manufacturers'
multi-year model development cycle. Reviewers also indicated that the model did not seem to address the
level of technology penetration in the baseline vehicle fleet. They perceived that the model treats the
baseline technology penetration as zero and suggested that the true level of baseline technology penetration
should be greater than zero.
The interaction between technology, fuel price, and vehicle miles traveled (VMT) also received frequent
comment. Reviewers indicated that it was likely that wide-scale implementation of the technologies available
in OMEGA2 could cause a significant change to overall fuel prices that should be considered. Also, they
indicated the model would benefit from further consideration of VMT rebound due to increased vehicle fuel
2
-------
Peer Review Report
Work Assignment 3-05/4-05, Contract 68HE0C18C0001
economy. Reviewers also tended to disagree with the model's assumption that vehicle fuel economy and
cost only affect a manufacturer's market share of a particular segment; rather, they thought it likely that
overall vehicle sales rates can also be affected by the balance of vehicle cost and fuel economy.
Two details received notable attention from reviewers. First, reviewers expressed general confusion around
the hauling/non-hauling segments and how they related to vehicles being categorized as cars versus trucks.
This included unexpected GHG emission effects in the outputs of hauling and non-hauling vehicles. Second,
reviewers questioned the model converting a car to a truck by the addition of all-wheel drive (AWD) when
there was no other technology that, when added, shifted the vehicle to another segment.
Finally, all reviewers commented on some aspect of the model algorithm's treatment of iterative
convergence on a final result. Mr. German requested that the convergence criteria and justification be
provided in the documentation. Mr. Rykowski thought the documentation would benefit from explanations
and warnings to the user indicating the types of inputs that could cause lack of convergence. Dr. Michalek
and Mr. Rykowski were concerned about the model's actions when iteration does not converge, as well as
the specific algorithms' potential for convergence on suboptimal solutions.
In addition to the key themes and most common comments summarized here, reviewers provided numerous
other specific observations and recommendations for the OMEGA model in response to the technical charge
to reviewers (Appendix B).
5.0 NARRATIVE COMMENT AND iSPONSES
This section includes reviewers' comments as received by ERG along with EPA's responses to their comments.
General EPA Response on Model Revisions: Note that OMEGA Model version 2.1.0 is the latest version
of the OMEGA model that includes both the changes to the model as the result of this peer review and
the continued development of the model as the result of the rulemaking process. References below to
"current version" and "latest version" are referring to version 2.1.0. References below to "a future
version of the model" indicate revisions beyond 2.1.0.
5.1 Comments Submitted by Mr. Sujit Das
5.1.1 General Comments
A significantly enhanced EPA OMEGA 2.0 version model than the earlier OMEGA 1.0 version is an excellent
and unique producer-consumer vehicle choice model by minimizing the resulting effects of societal costs and
emissions subject to user-input emission policies. It is intended to find a solution which simultaneously
satisfies producer, consumer, and policy requirements while minimizing the producer generalized costs. A
draft model documentation 300+ pages including the development model version accessible also directly
from Github provides the necessary detailed model scope description and its modeling approach capabilities.
A well-qualified team with economics background has been involved towards the second generation of
OMEGA2 model as evidenced by the model documentation and thorough explanations/responses provided
to a wide range of reviewer set of questions. The interactive decision-making process between producer and
consumer modules to meet the final annual emission targets (Policy module) is quite complex based on
economic theory but reasonable and adequate, could potentially be limited to only experienced economist
model users. The logical decision-making process in OMEGA2 seems too complex for the necessary level of
review (consisting of a review of at least 90+ pages out of total 300+ pages containing some source code of
the model documentation besides the model running) due to the limited effort available to reviewers. Some
additional documentation would be useful in terms of logical explanation of various iterative producer-
3
-------
Peer Review Report
Work Assignment 3-05/4-05, Contract 68HE0C18C0001
consumer vehicle choice process (suited to a general non-economist audience) before the public model
release.
5,1,2 Specific Comments
A summary section in the model documentation would be useful to identify the strengths and limitations of
the estimation approach and by specific decision modules in the modeling framework. In addition, how
OMEGA 2.0 model distinguishes among other similar publicly available models, if available.
EPA's Response (SD01): Thank you for the suggestion. We will consider adding a background section to
the model documentation that provides a broader discussion about the model, and its potential
limitations.
The policy-oriented model seems too complex for a novice user. A professional economics background is
necessary for interpretation of model results and so it'd be useful if the model documentation could be made
more simpler for the general users.
EPA's Response (SD02): We do not agree that a professional economics background is needed to
interpret the model results. While some of the internal concepts are based on economic principles, the
model outputs are expressed in simple and broadly understandable terms. With that said, we will
continue to look for ways to make the documentation clearer and more useful for a broad range of
readers and users.
What's the basis of an array of test values for producer "cross subsidy" price multipliers as shown in Figure
4.16? A schematic process flow diagram would facilitate in a better understanding of these multipliers
towards the final vehicle sales share outcomes/results.
EPA's Response (SD03): Thank you for the suggestion. Figure 4.2 does show the feedback between the
producer and consumer modules in the model's iteration process. We will consider adding some overall
schematic to describe the iteration process in a future version of the documentation.
Would be helpful for the novice EPA model users to include a short section on the other EPA complementary
ALPHA model with differences between it and OMEGA2 model.
EPA's Response (SD04): As described in the documentation, ALPHA is an optional physics-based
simulation tool used to generate vehicle inputs for OMEGA. Alternatively, OMEGA users may use any
other approach (e.g., other simulation tools, or empirical data) to generate the inputs required by
OMEGA.
It'd be good to have a model calibration/validation, if not already done so.
EPA's Response (SD05): There are test inputs that can be used to ensure that the model runs, and the
users Python environment contains the necessary libraries. Additionally, the user can make use of EPA's
model runs and compare local results to EPA's results to ensure that the model generates equal values.
Calibration/validation of inputs unique to the user's analysis are the responsibility of the user.
4
-------
Peer Review Report
Work Assignment 3-05/4-05, Contract 68HE0C18C0001
It'd be helpful to update the model user documentation by taking into consideration many excellent detailed
insights/responses to reviewers' comments.
EPA's Response (SD06): We intend to do this as appropriate.
It'd be helpful to include explicitly the learning effect on decreasing BEV production cost with an increasing
vehicle production volume in its generalized cost equation representation.
EPA's Response (SD07): This has been implemented in the current version of the model with a cost
reduction curve based on cumulative GWh of batteries produced.
It'd be good to consider in future updates (as suggested in one of the EPA's responses) to evaluate how the
algorithm might explore strategies where producers are intentionally over/under target in the various years
so that they minimize producer generalized costs over the entire analysis period, as opposed to the current
version's use of a 0 Mg C02 credit target (+/- a tolerance) in each year.
EPA's Response (SD08): Thank you for the suggestion. The current version does account for the use of
historical credit banks, and the carry forward and carry back of credits to offset over target years.
However, we have not implemented the strategic overcompliance and carry forward of credits in order
to reduce total costs over multiple years. Please see our response to reviewer comment RR05.
Steps for running the model seem not very straight forward to a novice user using the existing model GUI. It
was hard to locate some of the listed files in the model documentation. Additional step-by-step flow
diagrams including the file listing with a clear indication of necessary user inputs for both necessary pre-
processing and post-processing would be useful to be included in User's manual. In addition, a definition
glossary of major variables including a data dictionary would be useful.
EPA's Response (SD09): The GUI was designed to provide users the ability to run model examples and
become familiar with the various input and output files along with the directory structure. Many updates
have been implemented in the documentation describing these details.
It was quite interesting to note that the consumer preference dominates the final production decision in case
the convergence doesn't occur between producer and consumer preferences.
EPA's Response (SD10): The current version of OMEGA accepts the consumer shares one market class at
a time, which gives the producer more opportunities to adjust their production decisions. In general, it's
the case that the consumer and producer have different preferences and there is some negotiation
between them in the form of the cross-subsidy iteration. The larger the cross-subsidy range, the more
likely the producer is to achieve their desired shares.
EPA should pursue its suggestion to ensure that its complementary ALPHA results reliably represent absolute
emissions levels (consistent with OMEGA 2.0) as a function of vehicle loads, performance capabilities, and
technology applications.
EPA's Response (SD11): Independently of the OMEGA model development, we have a team of people
that are completing this work, comparing ALPHA results with test data from certification. A revised, final
set of ALPHA runs will be completed soon.
5
-------
Peer Review Report
Work Assignment 3-05/4-05, Contract 68HE0C18C0001
Does the past decision made affect its future decision? If so, has the recursive decision-making approach
been used?
EPA's Response (SD12): Yes, in the current version of OMEGA, prior production decisions affect the
current year's production decision. For example, redesign intervals determine the technology on the
vehicles that aren't available for redesign. There is also now a rate limit that restricts the maximum rate
of change of shares. For example, the limit would restrict a wholesale changeover from ICE to BEV over a
user-definable time span. As a result, the available share range in the current year is partially a function
of the prior year and the ramp rate. We are also investigating other year-over-year production decision
limitations and effects. The decision-making process is not recursive at this time.
Is there any consideration of efficiency improvement of conventional ICEs and cost similar to EVs?
EPA's Response (SD13): The efficiency of technologies considered are defined by the user in the input
files. The input files for emissions rates and energy rates are tied to model year, so the user can specify
those values improving over time. The user can apply time-based learning curves to adjust costs over
time.
5.2 Comments Submitted by Mr. John M, German
5.2.1 General Comments
Overall approach, specific approaches for modeling individual modules, and methodologies chosen
I want to acknowledge the level of improvements to OMEGA 2 from OMEGA 1. The original OMEGA model (I
was one of the original reviewers) was rather crude, requiring massive amounts of work to both the inputs to
the model and in processing and interpreting the results. OMEGA 2 is not just updated, it is essentially a new
model that can handle feedback loops and interactions between different modules and inputs, as well as
adding internal checks, more inputs, more flexibility in inputs, and a wide range of stock "auto-generated
image" output files.
Note that I did not do a general review of Sections 6 (Developer Guide) or 7 (Code Details) of the
documentation, as these are beyond my expertise.
Also note that because I started reviewing OMEGA 2.0.0 documentation before I received 2.0.1
documentation (and because 2.0.1 did not include page numbers), all documentation references below are
to 2.0.0.
The six model updates (i.e. expanded model boundaries, independent policy module, modeling of multi-year
strategic producer decisions, addition of a consumer response component, addition of feedback loops for
producer decisions, and use of absolute vehicle costs and emission rates) are all welcome additions.
The four modules, Producer, Consumer, Policy, and Effects appear to be structured well to allow changes in
one module to interact with the other modules while minimizing code changes throughout the model.
Flexibility in vehicle resolution (page 34) and market class resolution (page 35), as well as measures to
address consumer heterogeneity (page 36), are also welcome.
Calibrating to projected elements (page 41) is essential - glad to see this.
6
-------
Peer Review Report
Work Assignment 3-05/4-05, Contract 68HE0C18C0001
Sample input files, appropriateness, and completeness
The input files appear to cover a wide range of the potential influences on emission compliance and the
impacts of the standards.
Conceptual algorithms and equations: accuracy and appropriateness for technology application, market
inputs, and compliance calculations
Based on reviewing the documentation, the conceptual algorithms and equations appear to be accurate,
although this was not a primary focus of my comments.
Documentation: clarity, completeness, and accuracy
Description of how to run OMEGA, Sections 3.1 to 3.2, is good. Description of output files in Section 3.3 is
also good, although it could be slightly improved.
Section 4 description is a good overview and is mostly clear without being too long. Fig. 4.2 showing OMEGA
process flow is very helpful.
Policy Module (Section 4.2), Consumer Model (Section 4.4), and Physical Effects Calculations (Section 4.6)
descriptions very clear.
Congruence between conceptual methodologies and program execution
The complexity of the simulated_vehicles.csv file, especially the multiple rows for seemingly identical
vehicles (see comments in Section 4 and 7), precluded me from making any changes to the technology input
files (technology being my primary focus) and examining changes to the results.
Visualization output: clarity, completeness, and accuracy of technology display
The variety of stock "auto-generated image" output files "intended to provide a high-level view of the key
results" is impressive. Variety and depth of .csv output files is also impressive.
Note that I also included comments on the .csv output files in this section.
Recommendations for functionalities beyond "future work"
Given the relatively small number of references to "future work" in the documentation, I simply placed my
recommendations for future new functionalities in my comments on the previous sections.
5.2.2 Specific Comments
4.3 and 2nd set of EPA responses to peer reviewer questions pages 7-8
Producer Module, re discussion in the 2nd set of EPA responses to peer reviewer questions on pages 7-8 on
converting a car to a light truck (likely by adding AWD/4WD to a 2WD CUV). The primary advantage for a
producer to make this switch is to gain a less stringent emission standard. Perhaps I misunderstood EPA's
response, but it did not seem to directly address the technology cost savings from certifying to a less
stringent C02 standard. Bottom line is that OMEGA should model the producer decision as the sum of the
technology cost savings from certifying to a less stringent C02 standard plus the (heterogeneity) consumer
value of AWD/4WD minus the incremental cost of adding AWD/4WD. If OMEGA does not already perform
the calculation this way, it should be revised.
7
-------
Peer Review Report
Work Assignment 3-05/4-05, Contract 68HE0C18C0001
EPA Response (JG01): EPA agrees that manufactures may consider changing vehicle attributes as part of
a compliance plan under an attribute-based standard. Under the current light-duty GHG standards,
footprint, ground clearance, AWD, and an open cargo bed are some of the key consumer-facing
attributes that determine the GHG targets and might therefore be included as potential vehicle changes
in response to a new GHG policy. We have updated the model to include a user-definable consumer
valuation of footprint as part of the producer generalized cost value used in the cost-minimization
function for producer decisions. We will consider incorporating other attributes with explicit consumer
valuation parameters in a future version of the model. However, the model does not currently
accommodate shifting of reg classes by changing vehicle attributes as a producer decision.
4.3, 4.4, especially pages 48-49
The sidebar on pages 48-49 states that "OMEGA allows the user to define the consumer decisions from the
producer's perspective, which may be different from (or the same as) the representation within the
Consumer Module." Allowing the OMEGA user to define separate consumer decisions from the producer's
and consumer's perspective is nonsense. This implicitly assumes that the user knows better than producers
what the consumer response will be. Given the manufacturers' very existence is predicated on the best
possible understanding of their consumers and the vast amount of confidential data available to producers
but not the public, OMEGA should not allow the user to second-guess the producer's perspective. The
Producer and Consumer Modules should use the same definitions of consumer decisions.
• Note that this is not inconsistent with allowing the user to set parameters that affect the Consumer
Module output, only that the same parameters must be used for the Producer Module.
EPA's Response (JG02): We are currently using the GCAM model's equation forms for determining the
ICE-BEV consumer demand share. In those equations, consumer generalized cost is represented as a cost
per mile value. The user can select input parameter values for the producer generalized cost equation,
expressed in dollars, that are consistent with the cost per mile values for consumer generalized cost.
Although we did not adopt the reviewer's recommendation to use the exact same formulation for
consumer demand in the producer and consumer generalized cost for this version of the model, we will
continue to consider that change in a future version of the model.
4.3.3, Table 4.5, page 50
"epa_size_class" is defined as "Standard" SUV and pickups, while Fig. 4.3 on page page 42 uses "small" and
"large" definitions of size class. Need to be consistent.
EPA's Response (JG03): This various vehicle classifications used in the model can be defined by the user
in the model input files. We will consider revising the model documentation in a future release to clean
up any inconsistencies in class names used in the examples.
4.4 - A concern with the Consumer Module is that it can change the proportion of sales attributes to different
market classes. This affects compliance with the standards. Are you assuming that OEMs will make these
same sales adjustments when they plan on how much technology they need to comply with their targets? If
yes, need to clearly state this in the documentation. If not, then this portion of the Consumer Model will
cause OEMs to under or over comply.
• Note that the other parts of the Consumer Model are not affected by this concern (e.g. new vehicle
sales if sales attributed to different market classes are not affected, total vehicle stock, and total
vehicle use.)
8
-------
Peer Review Report
Work Assignment 3-05/4-05, Contract 68HE0C18C0001
EPA's Response (JG04): We agree that it is important that the model maintain consistency between the
GHG targets for a given policy definition and the vehicle attributes of the produced vehicles. In the
model's producer compliance search, each candidate solution is represented by its producer generalized
cost and compliance credits. The compliance credits are calculated relative to a candidate solution's
specific GHG targets, which will depend on the vehicle attributes in that solution. Similarly, the GHG
certification values are specific to the vehicles in that solution, which ensures that each solution
maintains consistency between GHG targets, GHG certification values, and therefore GHG credits.
There appears to be no way to compare incremental technology penetration, cost, and GHG reduction
versus the baseline vehicles in OMEGA 2, as the baseline vehicle input file (vehicle.csv) does not contain
information on technology penetration in the baseline fleet. This functionality should be added. Or, if it does
already exist, the documentation needs to explain how it is done.
EPA's Response (JG05): The model operates based on absolute GHG emissions and cost, rather than the
incremental effectiveness and costs. The model outputs are also reported as absolute values, from which
the user can determine the incremental effects of any sessions. Conversely, if we only provided
incremental values, it would not be possible to determine absolute values.
4.1.1. page 36
For 'Producer Resolution', the only options are perfect trading or no trading. Neither of these options are
realistic. Future versions of the model should include an option for individual entities with imperfect
trading between entities.
EPA's Response (JG06): Thank you for the comment. We have incorporated a 'credit market efficiency'
parameter that can be set by the user. A value of '1' represents a perfect credit market; a value of '0'
represents no credit trading provisions. Values between 0 and 1 represent imperfect trading, where
manufacturers with cost-minimizing compliance pathways that involve the purchase of credits will apply
more technology than they would under perfect trading. And as a result, some of the industries credits
will expire, unused.
4.1.4 page 41
Among the "purely exogenous" items listed is "the state of available technology". I understand this is how
virtually every technology analysis has always been done (with the notable exception of the 2013 NAS study
on Transitions to Alternative Vehicles and Fuels). However, new technology beyond the "state of available
technology" has continuous been developed. There is every reason to believe that this will continue to occur
in the future (e.g. 2021 Roush technology reports). It is important for projections out to 2050 to include
some way to account for future technology, beyond that just contained in currently "available."
EPA's Response (JG07): The user can supply the model inputs so that represents their own definition of
"available technology", including potential technologies that might be available in the future. We will
consider revising the model documentation to clarify that "available technology" does not need to be
limited to technologies that are in current mass production.
4.1.4 page 42
Size class descriptions. Fig. 4.3 shows that the default size classes divide pickup truck, van, utility, and
crossover classes into only two sizes each (small and large), compared with 6 different sizes for cars. Given
the shift to pickups, utility, and crossover vehicles, especially the growth in mid-size LDTs, it is time for EPA to
develop default size classes for at least small, midsize, and large crossover, utility, and pickup vehicles.
9
-------
Peer Review Report
Work Assignment 3-05/4-05, Contract 68HE0C18C0001
EPA's Response (JG08): The reviewer is referring specifically to the context size class definition that is
used for projecting fleet mix in future years. The demo example uses AEO projections, and therefore
maintains the AEO size class definitions and names. If a user has an alternate data source with greater
resolution for projecting changes fleet mix for types of vehicles, they can specify those class names and
associated projected volumes for the analysis context session in the new_vehcile_market.csv input file.
4.2 page 46 and .cvs input files
One of the inputs to the Policy Module is "offcycle_credits". This gives g/mi credits for listed off-cycle
technologies, which are turned off and on in simulated_vehicles.csv.
• The Off-cycle credits are given in fixed gC02/mi. However, off-cycle technology related to efficiency
(e.g. AC efficiency) are not fixed, but are proportional to the vehicle gC02/mi emissions. This should
be fixed in future versions of OMEGA, to avoid giving excess off-cycle credits as vehicle efficiency
improves.
EPA's Response (JG09a): Thank you for the suggestion. The model has been revised to allow the user to
specify AC efficiency and off-cycle credits based on any discrete vehicle attribute. For example, fueling
class, reg class, or a combination of the two in a combined attribute. For future revisions, we will
consider this suggestion for make the credit values a function of a continuous vehicle attribute, such as
efficiency.
• The SAFE rule and EPA's and NHTSA's revision of the standards through 2026 treat off-cycle and on-
cycle technologies separately, with off-cycle technologies being applied first regardless of their cost-
effectiveness. However, manufacturers freely choose between on-cycle and off-cycle technologies
and select the most cost-effective ones when planning for compliance. OMEGA needs to do the
same and compare cost-effectiveness of all technologies, including both on-cycle and off-cycle C02
reductions, simultaneously. As the off-cycle credits are turned off and on in the same file as the on-
cycle credits, it appears that OMEGA 2 might finally be doing this.
o If OMEGA 2 is treating on and off cycle technologies the same, this is important and should
be stated as one of the updates to the OMEGA model in section 1.
o If OMEGA still has some differences in how it implements on and off cycle technologies,
this should be fixed in future versions.
EPA's Response (JG09b): Thank you for the suggestion. The model does treat off-cycle technologies in
the same way was other technologies. Specifically, the model applies off-cycle technologies when they
are cost effective, and not automatically as was the case with off-cycle credits in modeling used by EPA
previously. We will consider providing a specific description of this approach in a future version of the
model documentation.
4.4.3 page 66
While it is a good first step to address heterogeneity, the method implemented in OMEGA appears to be
simplistic and require the user to define every item. For example, page 46 states:
"If users identify heterogenous consumer groups with separate sales share responses, the analysis
context must include the appropriate inputs. For example, the proportion of the vehicle buying
population in urban and rural areas for each year being analyzed within the model." This is confirmed
by the inputs in sales_share_Params-contxt_a.csv.
10
-------
Peer Review Report
Work Assignment 3-05/4-05, Contract 68HE0C18C0001
Future versions of OMEGA could consider separately modeling different types of customers (e.g. early
adopters, mainstream, late majority) as has been done by David Greene. This could both improve accuracy
and reduce the burden on the user to define the inputs.
EPA's Response (JG10): The default ICE-BEV share consumer demand model is based on the Pacific
Northwest National Laboratory's GCAM model. The share weight parameter is specified as a function of
calendar year, in the form of a logistic ('S' shaped) curve, which is a common way to represent the
progression over time in the acceptance of a new technology in a heterogeneous consumer market.
Although the GCAM model represents the market heterogeneity using a continuous logistic function, if a
user has identified an alternative consumer choice model with discrete consumer segments, it would be
possible to implement it into OMEGA with a new consumer module. We will continue to consider
whether any other consumer choice models may be appropriate substitutes for the GCAM model, and
possibly include in a future version of the model.
4.4.4 pages 69-71
Vehicle stock. Vehicle life has continuously increased for decades. Thus, it would be appropriate to at least
include an option to increase the reregistration rate as time goes by, if not make it the default for OMEGA.
EPA's Response (JG11): We agree that this is useful functionality. We have added a 'start_year'
parameter to the reregistration rate input file. The user can now specify unique reregistration rates by
calendar and model year.
4.4.4 p 71
States, "The demo analysis does not currently implement VMT rebound estimations." I assume that the
official release of OMEGA 2 will include VMT rebound estimations?
EPA's Response (JG12): We have implemented VMT rebound in the OMEGA effects module. The user can
specify VMT rebound values for BEV and ICE fueling classes separately.
4.3.2. page 49 & 1st set of questions from OMEGA Peer Reviewers page 13
Documentation states (page 49), "At this time, OMEGA assumes that producers aim to meet the GHG target
in each year, with any banked credits used only to make up differences between the certification and target
values."
The EPA Response states, "For future updates, we are evaluating how the algorithm might explore strategies
where producers are intentionally over/under target in the various years so that they minimize producer
generalized costs over the entire analysis period, as opposed to the current version's use of a 0 Mg C02
credit target (+/- a tolerance) in each year." And asked for reviewer responses.
It is extremely important to allow manufacturers to use up banked credits by intentionally exceeding
C02/mi target levels in various years; by modelling credit carry-forward strategies. This can be seen in the
industry's deliberate undercompliance in every model year from 2016 to 2020, when the industry's C02/mi
emissions exceeded the performance requirements each year (EPA 2021 FE Trends report). As explained in
an ICCT blog (https://theicct.org/blog/staff/us-fuel-economv-trend-reflects-business-strategy-not-tech-
challenge) this is just good planning on the part of manufacturers, as using banked credits reduces the overall
cost of compliance. This strategy must be included in EPA's modeling.
It is less clear to me the value of allowing manufacturers to intentionally over/under comply in various years
and to model credit carry-forward and carry-back strategies. This would be needed if EPA were modeling 5-
11
-------
Peer Review Report
Work Assignment 3-05/4-05, Contract 68HE0C18C0001
year design cycles, but in the absence of design cycles the additional complexity of modeling a general
over/under compliance strategy may not be worth the effort.
EPA Response (JG13): In the current version, the default behavior is to allow overcompliance only when
a producer needs to make up for a year that was over target, or the consumer demands more emissions
reductions than are required by the policy. We have implemented logic for the producer's compliance
strategy that allows under-target performance when it minimizes producer generalized costs for that
model year. This feature is still under development and can be enabled or disabled using the
'voluntary_overcomliance' setting. Similarly, we have revised the logic so that the producer will use
historical banked credits before they expire, even if it results in being over target for the early analysis
years. We have not implemented the change that allows the manufacturer to either strategically
generate credits based on a prediction of preferring to use those credits in later years, or strategically run
a deficit based on a prediction that credits can be more cost effectively generated in later years. We will
continue to consider whether it is possible in a future version of the model to evaluate various under-
and over-target pathways as part of a compliance trajectory that minimizes costs over all analysis years in
total, rather than for each analysis year separately as is the current logic.
4.4.1 page 58
Documentation states, "When a (typo?) modifies a submodule, they must ensure that the submodule retains
consistency with the other submodules within the Consumer Module, as well as with the rest of OMEGA. For
example, if the market class submodule is changed from the demo analysis version, the sales share
submodule must be updated as well since sales shares are determined by market class."
In future versions of OMEGA, might investigate if the model can make appropriate adjustments to the
other submodules when the user modifies a submodule.
EPA Response (JG14): Thank you for the suggestion. We will continue to consider how to make any
future versions of the model more adaptable and user friendly.
.CSV Files - It would be helpful, at least for the more complicated files, to include a sheet with a brief
description of the variables in the file.
EPA Response (JG15): Thank you for the suggestion. We plan to update the description of variables in the
documentation in a future release. We note that adding descriptions directly to the csv files would
introduce the potential for errors in the formatting and data reading. Furthermore, some files have
variable names repeated in multiple rows and/or columns and could not be easily tagged with field
definitions.
Simulated vehicles.csv
It appears that FTP results are reported by bag (T:W and Z:AC), but without reporting overall C02 or kWh/
mi. Does OMEGA do this calculation?
• It might be helpful if the documentation said that FTP inputs have to be reported for each bag (and
that OMEGA calculates overall FTP results?).
EPA Response (JG16): We have implemented the capability for the user to define drive cycles using any
combination of weighting. For example, the FTP drive cycle can be specified by individual bags, or as a
single weighted cycle. We will consider updating the documentation to clarify this flexibility in a future
release.
12
-------
Peer Review Report
Work Assignment 3-05/4-05, Contract 68HE0C18C0001
There are multiple rows for each combination of year, cost curve class, technology, and curb weight. For
example, rows 27477-27487 are all 2026 LPW_LRL HEVs with AC leakage, AC efficiency, GFI, and 2981 curb
weight. Yet each of the 11 rows has a different new vehicle cost and FTP emissions.
• The multiple rows appear to be linked to an undefined 'vehicle_id'. However, I was not able to find
any definition of the vehicle attributes associated with this vehicle ID, either in the documentation or
in the input files. Documentation should explain how vehicle IDs are generated and there needs to
be an input file added with the attributes associated with each vehicle ID (e.g. technology, vehicle
class).
• I checked Section 7.2.1.3.2 on the cost_cloud module in Python but it was not clear to me what the
"cost cloud frontiers" discussed in this section referred to.
EPA's Response (JG17): The vehicle IDs are automatically generated in sequence within the model, with
no meaning apart from the function of uniquely identifying individual vehicles both within and across
years. We will consider adding this clarification to in a future version of the model documentation.
The technology choices appear to be quite limited, with just HEV, PHEV, BEV, weight reduction, 2 types of
DEAC, CEGR, atk2, GDI, and 2 types of turbos (turbll and turbl2), plus 4 types of off-cycle credits (AC
leakage and efficiency, high efficiency alternator, stop/start). This is quite limiting, especially beyond 2025
when technologies such as e-boost, variable geometry turbos, and lower bore to stroke ratios should become
common.
• Is this just because the "demo" files are not intended to be complete? If so, should state clearly
when discussing the demo examples in Section 3 that the technology files are just examples and are
not complete.
• Is it possible for users to add technologies to the simulated vehicles .csv file, for example by adding
new columns? If so, should state this someplace in the documentation and give some idea how to do
this.
• I also checked Section 7.2.1.3.2 on the cost_cloud moduel in Python, but this does not even mention
technology options, much less how to add them.
EPA's Response (JG18): We provided the demo for the peer review for the purpose of illustrating the key
model functions using a simplified set of inputs. The user can add any number of possible technologies to
the simulated_vehicles.csv input files, as long any unique technology flags are represented in the
powertrain_costs.py module, and the associated costs defined in the powertrain_costs.csv input file.
vehicles.csv
This file of baseline (currently 2019) vehicles does not include a list of technologies already implemented in
the baseline fleet. How can you calculate incremental costs and C02 reduction without knowing the
baseline vehicle technologies?
Also, costs in the baseline 'vehicles.csv' file are MSRP, while costs for future vehicles in 'simulated
vehicles.csv' are manufacturing costs. How does OMEGA calculate incremental costs? Should at least be
explained in the documentation somewhere.
EPA's Response (JG19): The vehicles.csv base year input file now contains fields for 'cost_curve_class'
and 'body_structure_material' that describe the technology packages in the base year fleet. This change
was made to accommodate the implementation redesign cycle constraints in the model, and the
requirement that the powertrain package and body structure be carried over from one year to the next
until the next redesign opportunity. This includes the years following the base year, and the constraint
13
-------
Peer Review Report
Work Assignment 3-05/4-05, Contract 68HE0C18C0001
that base year technologies be carried over for as long as a full design cycle span depending on when the
prior redesign year was.
4.4.4 p 71
Reiterate that the official release of OMEGA 2 should include VMT rebound estimations.
It appears to be virtually impossible to modify the technology input file due to it's complexity, for example to
evaluate the impact of changing the cost or effectiveness of a single technology. It would be helpful to
implement a way to do this in future versions of OMEGA.
Or, as an alternative, the OMEGA documentation could discuss how to link OMEGA to ALPHA and provide
some guidance on how to modify technology inputs to ALPHA in order to generate new OMEGA input files.
EPA's Response (JG20): We provided the demo for the peer review for the purpose of illustrating the key
model functions using a simplified set of inputs. The user can add any number of possible technologies to
the simulated_vehicles.csv input files, as long any unique technology flags are represented in the
powertrain_costs.py module, and the associated costs defined in the powertrain_costs.csv input file.
1.1 page 2; 4.1.1 page 33
Fig. 1.1 diagrams how technology is input into OMEGA 1, but does not for OMEGA 2. Would be helpful to
diagram here (and in Fig 4.1) and discuss in Section 1.2 how the ALPHA model fits in with the OMEGA
model.
EPA's Response (JG21): This is outside of the scope of the OMEGA model and documentation. The user
can define the technology energy consumption rates and C02 emissions rates in the
simulated_vehicles.csv input files using whatever data sources are deemed appropriate. Our use of
ALPHA vehicle simulations, or any other simulation tool would be described separately in future
rulemaking documentation.
1.2 page 4
Again, as EPA uses the ALPHA model to generate technology inputs, an overview of how ALPHA fits with
OMEGA would be helpful in Section 1.2
2.3 page 8
No mention of the executable file for Mac computers, only win.exe
3.1 page 10
Might help to provide some examples of Input Batch Files for Element 2 - is there just one OMEGA input file?
If there are more than one, what might they be?
• I note that Section 3.3. on output files contains a great deal of discussion of the various output files,
but there is nothing for input files.
• I also see that Section 4.3.3 Table 4.5 contains a definition of the input files. If you don't add a section
on input files here, similar to Section 3.3 for output files, Section 3.1 should at least reference Tables
4.2, 4.3, 4.4 and 4.5.
3.2.1 & 3.2.2 page 11-12
A new term, 'Configuration file' is introduced. Should explain what is contained in the Configuration file (e.g. I
don't understand how this differs from the Input Batch File).
14
-------
Peer Review Report
Work Assignment 3-05/4-05, Contract 68HE0C18C0001
3.3 page 16
Where is the "out" folder located?
There is no description of the "Results" tab in the documentation.
3.3.1 page 17
• Again, should state where the results files can be found.
• Demo example is "reading the manufacturer compliance plot", but this is not one of the files
described in Table 3.1
• There are three output .png files that are not listed in Table 3.1 on page 17: Sales v Year and V Cost
Mkt Cat/Cls.
3.3.1 page 18
Figure 3.7:
• Might be helpful to output a description with the .png for Figure 3.7, as I found it difficult to follow
Figure 3.7 even with the description.
• What is calendar year used for? (didn't see this in description)
3.3.1 page 19
What does "hauling" and "non-hauling" mean?
Why does "hauling" have more stringent C02 targets than "non-hauling" in upper panels (the reverse of the
lower panels)?
3.3.1 page 24
Caption for Fig. 3.11 should state these are changes in stock parameters and that the results do not include
rebound effects.
3.3.1 page 24-25
For costs, BEV cost parity is reached far later than most analyses. Would help to describe what the vehicle
mix assumption is. BEV costs are much more strongly affected by the mix of cars, CUVs, SUVs, and pickups
than ICEs, so assuming a small car share (for example) would delay cost parity.
3.3.2 page 26
Again, should state where the output files can be found.
3.3.2 page 27
The 'physical_effects.csv' output file has about 50,000 rows. It might help to give the user some idea how to
handle these rows. For example:
• To generate Figure 3.16, were the total GHG emissions in each row simply totaled by year and BEV-
upstream, ICE-tailpipe, and ICE-upstream?
• Should also discuss how to conduct more detailed analyses, which appear to depend on how to
interpret 'vehicleJD' (and 'reg_class_id'?), as there are multiple rows with no other distinction. Is
there a file somewhere that defines what they mean and the attributes of each index? (I did a search
for vehiclejd, but came up with nothing helpful.) Should state where to find the definition for each
vehicleJD and reg_class-id.
Note that if they are not defined anywhere, it makes it impossible to do more detailed analyses of the
results.
15
-------
Peer Review Report
Work Assignment 3-05/4-05, Contract 68HE0C18C0001
Note that this also applies to many other files, such as the technology tracking output file and Figure 3.17.
3.3.2 page 31
Figure 3.18 has a "fuel_pretax_cost", but there is no electricity cost. Electricity is not normally defined as a
"fuel", so if this parameter includes both, should change the name.
• On page 80, 'FuelConsumptionkwh' is used for Electricity consumption, confirming the problem in
Figure 3.18. Should use 'energy' cost and consumption instead of 'fuel'.
Also, what is "refueling_cost" and why is it different from fuel cost?
4.1.3. p 39
Some of the elements in Table 4.2 have values instead of 'input file names', i.e. New Vehicle Price Elasticity of
Demand and Max and Min Producer Cross Subsidy Multiplier. The last sentence on page 38 before Table 4.2
says, "Table 4.2 shows the complete set of input files and settings for Context A as defined in the
'demo_batch-context_a.csv' file." However, it would be helpful to:
• Start by briefly discussing the role of the batch-context files in defining the input files that OMEGA
looks for.
• More clearly state that the New Vehicle Price Elasticity of Demand and Max and Min Producer Cross
Subsidy Multipliers can be found and changed in the batch-context input file (e.g. demo_batch-
context_a.csv).
4.1.3; 4.2; 4.3.3;
Tables 4.2, 4.3, 4.4, 4.5: Right side is cropped off in pdf version (fine in .html).
EPA's, 2nd response to peer reviewer comments (page 1) states the online version of the docs is the preferred
method of viewing the docs. If you want this to be the preferred method, you need to:
• Find some way to help the user reference specific information, by adding page numbers or more
levels to the Table of Contents. For example, Section 4 is over 50 pages long and Section 7 is over 200
pages long, making it extremely difficult to provide references to specific information.
• Some of the descriptions in Table 4.2, for example, are extremely long, plus the user has to scroll
back and forth for every row in the .html. This makes it very difficult to understand what is in each
file. It would be much better to wrap the descriptions in a smaller box, so that users do not have to
continuously scroll back and forth.
• The biggest issue is being able to conduct word searches, which works fine with pdf files but the
searchfunction for the .html file did not work for me (maybe a Mac problem?)
4.3.3. Table 4.5. page 50
• "epa_size_class" is defined as "Standard" SUV and pickups in Table 4.5, while Fig. 4.3 on page 42 uses
"small" and "large" definitions of size class. Need to be consistent.
• More importantly, Table 4.5 does not appear to include any technology information for the baseline
vehicles in 'vehcles.csv', which I confirmed when I opened vehicles.csv. Some explanation of how
incremental technology benefits and costs over the baseline vehicle are handled should be included.
o Note that Section 4.3.4 states that technology and cost information is included in
'simulated_vehicles.csv', but no comparable discussion is included in the discussion of baseline
vehicles in 4.3.3.
16
-------
Peer Review Report
Work Assignment 3-05/4-05, Contract 68HE0C18C0001
4.3.5.
I found the discussion of selecting the minimum cost package available to be confusing. Might help to
explain:
• What each of the orange and blue points in Fig. 4.7 represent - from the documentation, I had little
idea why there are so many points, no idea if each point represented a single technology or
technology aggregations, etc.
• Why it is appropriate to use the cost-minimizing frontier of the cost cloud.
4.4.3 page 68-69
Top of page 69 states the amortization rate in the demo example is set at 10 years, but
'price_amortization_period' in Table 4.6 is set at 5 years.
4.4.4 p 70
The demo example uses a fixed schedule based on vehicle age for reregistration. It would be helpful to
identify the source for the default reregistration schedule in the OMEGA model, as was done for BEV share
parameters on page 68.
4.5 p 73
What do the red stars mean on Fig. 4.15? (Fig. 4.14 says "the selected compliance option from the initial
Producer-Policy compliance search is shown by a single red star.", but Fig. 4.15 has multiple red stars)
4.5 p 74
Discussion of Fig. 4.16 says nothing about the lower right pane.
4.5
Convergence criteria: Section 4.5 discusses the convergence criteria in many places and page 75 mentions
the iteration set limit. However, the documentation fails to discuss what these criteria are or the basis for
setting the limits, which should be added.
4.6.1.1.. page 80
Transmission Efficiency for both fuel and electricity consumption are stated to be "as set by the user". This
implies there are no default OMEGA assumptions.
5.. p 84-86
The Batch Input Files are all .py files. This implies that the files used when running with Python are different
than the files used when running using the GUI.
• Section 5 (User Guide) should make clear upfront that Section 5 is for Python users, not GUI users.
• Some explanation as to why different files are used with Python would be helpful.
• Still, it is nice to have all the input files listed in one place. Something similar for the earlier .csv input
files (in Section 3?) might be helpful.
3.3.1 page 18
Why are both calendar year and model year cert C02e plotted? Should explain in the documentation what
calendar year and model year are used for and why each is needed.
3.3.1 page 19
Figure 3.8:
17
-------
Peer Review Report
Work Assignment 3-05/4-05, Contract 68HE0C18C0001
• Why does "hauling" have more stringent C02 targets than "non-hauling" in upper panels (the reverse
of the lower panels)?
• Why are there no BEVs in the upper panels
3.3.1 page 24
Caption for Fig. 3.11 should state these are changes in stock parameters and that the results do not include
rebound effects.
3.3.1 page 26
Figure 3.13 has an orange "sales" line in the legend, but not in the graph.
3.3.2 page 31
Figure 3.18 has a "fuel_pretax_cost", but there is no electricity cost. Electricity is not normally defined as a
"fuel", so if this parameter includes both, should change the name - perhaps "energy_pretax_cost".
Also, what is "refueling_cost" and why is it different from fuel cost?
3.3.3 page 31
The run log is mixed in with all of the image and .csv output files. Might be better to put it someplace by
itself - preferably a new link on the GUI.
GUI - results tab
The Results Tab has not yet been developed. Due to the large number of predefined .png and .csv output
files, I would recommend:
• Using descriptive names in the Results tab, such as "Cert and compliance versus year - no action" for
the first .png file. As the output files are predesigned, this should be relatively easy to set up and link.
• It would also help to have separate lists for the .png and .csv files.
• For .csv files, the batch-context summary results file should be highlighted somehow.
• Similarly, if there is an overall summary .png file, this should also be highlighted somehow - or
perhaps better yet, add a default .png summary figure based on the batch-context summary results
file?
2021_ll_17_ll_31_31_demo_batch-context_a_NoActionPolicy RC Shares.png - I have no idea what a, b,
and c regulatory classes are. Unless I missed this, a description should added to the documentation
2021_ll_17_ll_31_31_demo_batch-context_a_NoActionPolicy Stock C02 Mg.pgn - Title is "Vehicle stock
C02 mg", but clearly this figure is not total vehicle stock C02. Need a more accurate title.
2021_ll_17_ll_31_31_demo_batch-context_a_NoActionPolicy Stock Gas Gallons.png - Title is "Vehicle
stock registered count", but clearly vehicle stock won't grow from 10 million in 2020 to 240 million in 2050.
Need a more accurate title.
2021_ll_17_ll_31_31_demo_batch-context_a_NoActionPolicy Stock kWh
• Should change title from "Fuel" to "Energy".
• Vehicle stock fuel consumption is about 1.4 kWh in 2050? This isn't even accurate for per-vehicle
consumption. Y-axis label problem?
2021_ll_17_ll_31_31_demo_batch-context_a_NoActionPolicy Stock VMT
18
-------
Peer Review Report
Work Assignment 3-05/4-05, Contract 68HE0C18C0001
Title is "Vehicle stock miles traveled", but clearly VMT won't grow from 0.1 mile in 2020 to 2.8 miles in 2050.
Need a more accurate title - and y-axis title?
2021_ll_17_ll_31_31_demo_batch-context_a_NoActionPolicy V Cert C02e gpmi Mkt Cat & CIs
Is there a reason why ICE emissions go up after 2028 - especially since the target values do not? (in
2021_ll_17_ll_31_31_demo_batch-context_a_NoActionPolicy V Tgt C02e gpmi Mkt Cat)
New .png - 2021_ll_17_ll_31_31_demo_batch-context_a_NoActionPolicy V Cost Mkt Cat & CIs - Average
vehicle cost is fine, but would also appreciate an automatic .png file for incremental cost versus 2019
baseline, at least for ICE
2021_ll_17_ll_31_31_demo_batch-context_a_NoActionPolicy V Tgt C02e gpmi Mkt Cat - The target
values for hauling are much lower than the other categories - mislabeled lines?
2021_ll_17_ll_31_31_demo_batch-context_a_NoActionPolicy V Tgt C02e gpmi Mkt CIs
• Target values for hauling much lower than non-hauling - mislabeled?
• There are no lines for hauling or non-hauling BEV (listed in legend)
New .png -1 would appreciate an automatic graph of ICE technology penetration, at least major tech
categories (e.g. MHEV, SHEV, HCR2, Turbo 11, Turbo 12)
¦ CSV
2021_ll_17_ll_31_31_demo_batch-context_a_NoActionPolicy_cost_effects.csv
2021_ll_17_ll_31_31_demo_batch-context_a_NoActionPolicy_physical_effects
2021_ll_17_ll_31_31_demo_batch-context_a_NoActionPolicy_tech_tracking
• There are multiple rows with no distinction beyond 'vehicleJD' numbers (and 'reg_class_id'?).
o Is there a file somewhere defining the attributes of each vehicle ID (and reg class)?
o It would be helpful if the attributes associated with vehicle ID and reg class could be added to
each output file that generates separate rows based on vehicle ID, to make it easier to interpret
the results.
• In these files, the reg classes appear to be car and truck. How does that compare with "hauling" and
"non-hauling" used in the documentation and in the .png files? And why is it different?
• In the policy cost effects file, there are 4 different costs presented for each cost variable (e.g. CH4
and C02). What do the 25 cost, 395 cost, 3 cost, and 5 cost columns mean, why do they differ, and
why are they all needed?
New .csv - 2021_ll_17_ll_31_31_demo_batch-context_a_NoActionPolicy_new_vehicle_prices
Average vehicle cost is fine, but would also appreciate an automatic .csv file for incremental cost versus
2019 baseline, at least for ICE
2021_ll_17_ll_31_31_demo_batch-context_a_NoActionPolicy_vehicle_annual_data
This file has index numbers, which I have not seen in any other file.
• How do the index numbers relate to other files when no other file has them?
o Note that there are 53,856 index numbers, which does not agree with the number of rows in
previous files with multiple rows for each configuration.
2021_ll_17_ll_31_31_demo_batch-context_a_NoActionPolicy_vehicles
19
-------
Peer Review Report
Work Assignment 3-05/4-05, Contract 68HE0C18C0001
There are no entries for technology on baseline 2019 vehicles, confirming that there is no technology
penetration entered for the baseline fleet. How can incremental technology benefits and costs be calculated
without knowledge of baseline vehicle technology?
EPA's Response (JG22): Thank you for the suggested clarifications and updates to the documentation.
We will consider revising a future version of the documentation.
There appears to be no way to compare incremental technology penetration, cost, and GHG reduction
versus the baseline vehicles. This functionality should be added. Or, if it does already exist, the
documentation needs to explain how it is done.
EPA's Response: Already addressed earlier.
5.3 Comments Submitted by Dr. Jeremy J, Miclialek
5.3.1 General Comments
Thank you for the opportunity to review EPA's OMEGA2 model and provide comments.
In providing feedback on OMEGA2, I first offer feedback on the mathematical problem that OMEGA2 aims to
solve; then I offer feedback on the algorithm approach to solving the problem; and finally I offer feedback on
a list of additional items. OMEGA2 is a major expansion of scope beyond OMEGA1, and the efforts to make
OMEGA flexible to a variety of user modeling choices is excellent and will be useful in informing policy.
I will focus my review on critiques and recommendations that I hope will be most helpful in considering how
OMEGA2 or a future version of OMEGA might be generalized, expanded or improved.
5.3.2 Specific Comments
1. Comments on the Mathematical Problem OMEGA2 Aims to Solve
In my understanding, the optimization problem that OMEGA2 aims to solve is to minimize producer
generalized cost with respect to the attributes, price, and production volume shares of each (composite)
vehicle in the producer's fleet subject to policy constraints, consumer demand constraints, and a constraint
that sales-weighted average sale price must equal sales-weighted average baseline price, where baseline
price of each vehicle is production cost times a fixed margin.
I think it is helpful to represent the core mathematical problem that OMEGA2 aims to solve on one page. The
way I would write this optimization problem in standard form for a single year (using matrix calculus notation
where all vectors are in bold and all scalars are in italics) is
minimize c(x) minimize producer generalized cost
a
with respect to x = p
v
with respect to a vector of decision variables x that includes the
attributes a, prices p, and production volume shares vfor each of
the producer's vehicles,
subject to
subject to the following constraints:
20
-------
Peer Review Report
Work Assignment 3-05/4-05, Contract 68HE0C18C0001
g{a,v) < 0
q(a, p) - v = 0
vTp — vTb(a) = 0
a 6 R"xm
p 6 Mn
v 6 Mn
where
b(a) = mk(a)
• the fleet must comply with GHG standards;
g(a, v) is the policy module, which takes attributes and shares
as inputs and returns Mg GHG credits short of the standard
• production volume shares must equal demand shares;
q(a, p) is the consumer demand model, which takes attributes
and prices as inputs and returns a vector of shares, one for
each vehicle
• sales-weighted average price = sales-weighted average baseline
price
• a is a vector with m attributes for each of the producer's n
(composite) vehicles. One of these m attributes is GHG
emissions per mile.
• p is a vector with one price value for each of the producer's n
(composite) vehicles
• v is a vector with one production volume share value for each
of the producer's n (composite) vehicles
where
• baseline price is production cost k(x) times a fixed markup
(m = 1.5 in OMEGA)
Eq.(l)
Eq(l) is nonconvex nonlinear optimization problem (NLP). Eq(l) could be a nonconvex mixed-integer
optimization problem (MINLP) if some of the attributes in the vector a are discrete, but discrete technologies
are represented by a convex hull among a set of discrete points in OMEGA2, which can be represented as a
linearly-constrained continuous slack variable, eliminating the discrete variables, and reducing the problem
to an NLP. Figure 1 shows how the symbols in Eq(l) map to the diagrams in the OMEGA documentation.
Policy Alternatives
Policy
Module
to/from Producer Module:
Vehicle GHG
cert and
target values
Quantities,
Attributes of
Candidate
Vehicles
fl( a,v)
a,v
Analysis Context Inputs
to/from Policy Module:
Analysis Context Inputs
Consumer Module
to/from Producer Module:
Vehicle Demanded q(a.p)
to/from Consumer Module:
Vehicle Demanded
Vehicle Prices, Attributes
of Candidate Vehicles
Prices, Attributes of
a,p
Candidate Vehicles
Vehicle
Stock
Vehicle
Use
Attributes of
Produced Vehicles
Figure 1: Relationship between Eq(l) and the OMEGA2 documentation
21
-------
Peer Review Report
Work Assignment 3-05/4-05, Contract 68HE0C18C0001
The documentation emphasizes iteration between the producer module and the policy / consumer modules,
and the solution algorithm and code are set up this way, but in terms of the mathematical problem that
OMEGA2 is fundamentally solving, structurally the policy and consumer modules are just functions that
return certification compliance values and a vector of demand values, respectively, as in Eq.(l).
OMEGA2 a major expansion of the prior OMEGA model in scope, and it is designed to be flexible in
accommodating different user modeling choices for policy design and consumer demand. These are welcome
changes, as they allow modelers to avoid some of the restrictive assumptions of OMEGA1 and potentially
account for additional important factors. For example, Whitefoot et al. (2017) find that "The ability of
automakers to improve the fuel economy of vehicles using engineering design modifications that
compromise other performance attributes, such as acceleration, is not currently considered [in OMEGA1]
when setting fuel economy and greenhouse-gas emission standards for passenger cars and light trucks" and
"our analysis suggests that acceleration trade-offs play a role in automaker compliance strategies with
potentially large implications for both compliance costs and emissions." OMEGA2 provides flexibility for
modelers to potentially account for factors like this.
The model also has flexibility for modeling heterogeneous consumers: "users can define sales shares to
explicitly consider consumer heterogeneity by defining separate sales share estimates for different consumer
groups." This is a particularly useful flexibility, given the difference not only in consumer preferences across
the population but also in situational factors that affect willingness to adopt. For example, renters and home
owners without off-street parking may be much less likely to adopt electric vehicles than home owners with
off-street parking who can install private charging infrastructure, and modeling flexibility can allow users to
capture effects like these.
The structure of OMEGA2, however, still does have limitations that restrict its ability to model some
important aspects of policy compliance in practice. The most important of these limitations that I identified
are all related to restrictions in OMEGA2's ability to model firm profit-seeking compliance behavior.
EPA's Response (JM01): Thank you for the suggestion. We agree that the minimization of generalized
cost is not the same as profit maximization. We have incorporated willingness to pay for changes in
footprint into the generalized cost definition, so at least in that respect the model now uses an approach
for vehicle size changes that is consistent with the objective of maximizing profit.
The modeled average prices of vehicles may change based on the changes in the vehicle technologies
applied and associated production costs. But as the reviewer notes, price cross-subsidization between
BEV and ICE vehicles is applied in a way that maintains average prices. So, while the modeled profit
margins for BEV and ICE vehicles are allowed to vary within OMEGA, for this version of the model we did
not attempt to implement free pricing of vehicles. At this time, we have not identified a model
specification for the relationships between vehicle prices, attributes, and quantities that could be used to
provide a computationally tractable solution within the iterative search structure of the model.
We will continue to consider the possibility of including a free-pricing optimization algorithm in a future
version of the model.
1. Firm Pricing Strategy and Profit Maximizing Behavior
Recommendation 1: Implement a generalization of OMEGA2's structure that will allow users to model
profit-maximizing behavior by including free pricing of the vehicles during the compliance search,
rather than fixed margins.
Rationale: The model structure requires that the sales-weighted average vehicle price is equal to the
sales-weighted average baseline price, which is production cost times a fixed markup for each vehicle.
22
-------
Peer Review Report
Work Assignment 3-05/4-05, Contract 68HE0C18C0001
This restriction prohibits producers from preferring technologies that sell at higher margins (unless it is
impossible to satisfy regulation without them).
• The document emphasizes that the "assumption that producers will attempt to minimize their
generalized costs is consistent with a producer goal of profit maximization/' but I believe this is
only true under restrictive conditions. In the appendix I provide a simplified Matlab code to solve
for (1) minimum production cost, (2) minimum producer generalized cost, and (3) maximum
profit subject to (a) average price constraints or (b) no price constraints using a multinomial logit
model of demand. The solutions to (la), (lb), (2a), (2b), (3a) and (3b) are all different, suggesting
two things: (i) minimizing producer generalized cost is not necessarily consistent with profit
maximization and (ii) fixing average price, using a fixed margin, restricts solutions, leading to less-
profitable solutions.
• When consumer willingness to pay (WTP) for some products is higher than production cost and
WTP for other products is comparable to or lower than production costs, minimizing producer
generalized cost (which will prefer the lower cost products) may not be consistent with profit
maximization behavior (which will prefer the higher margin products). In order to model profit-
maximization behavior, it is necessary that (1) prices are treated as free variables without
constraining margins to a pre-determined average, and (2) negative revenue is included in
producer generalized cost. The latter is possible in OMEGA2's current structure, depending on
user-specified functions, but the former is not.
Recommendation 2: Advise users that when looking to model profit-maximizing behavior, producer
generalized cost should include total production cost minus total revenue for the fleet, rather than
willingness-to-pay.
EPA's Response (JM02): The OMEGA model's use of WTP values in the generalized cost function is
limited to the fuel costs considered over a specified number of years after vehicle purchase, and vehicle
footprint. We agree that cases where the other vehicle attributes vary dramatically between the
alternatives, that the willingness to pay values for fuel costs and footprint may also vary. In the OMEGA
model, the other vehicle attributes are generally held constant, except for vehicle price. Even considering
the difference between ICE and BEV powertrain types, the other vehicle attributes related to
performance and capability are assuming to be similar. Under these conditions, we believe it is
appropriate to include an absolute WTP value for footprint and fuel cost.
Rationale: The OMEGA2 documentation states that WTP can be embedded in the producer
generalized cost function:
"While the producer's modeled objective function is cost minimization, the term 'cost' is used
generically here, as it is not necessarily the case that the lowest production cost option is the best
option for the producer. Consumer willingness to pay for a particular vehicle attribute can result in
another choice if the producer expects that the additional production costs can be more than
offset by a higher price. Here, the term producer generalized costs is defined as the net of vehicle
production costs and the producer's view of consumer's willingness to pay for that vehicle."
However, in (random utility) consumer choice models, WTP is a relative measure. There is no such
thing as "WTP for product A" because demand for product A depends on the attributes and prices of
competitor products as well. What these models estimate is WTP for a change from attribute level 1 to
attribute level 2. Thus, WTP for a product is always relative to a reference point (e.g.: WTP $x more for
product A than product B, given their attributes). Further, WTP is an average value - it really measures
the iso-demand price equivalency of a change in attributes. For example, if WTP for the attributes of
product A is $x more than WTP for the attributes of product B, and if product A is priced at $x more
23
-------
Peer Review Report
Work Assignment 3-05/4-05, Contract 68HE0C18C0001
than product B, then share for products A and B will be equal. If product A is priced at $(x+l), then
share for product A will be less than product B. This all matters because (at least using the discrete
choice model paradigm) there is no way to specify a single "WTP" value for a single vehicle in the
producer generalized cost function independent of all other vehicles.
2. Firm Sales Volume Strategy
Recommendation 3: Implement a generalization of OMEGA2's structure that allows users to run profit-
maximizing cases that consider strategic pricing and overall sales volume in producer decisions.
EPA's Response (JM03): We have not included strategic vehicle pricing by manufacturers to mitigate the
effect on sales or maximize overall profits. We agree that the minimization of generalized cost is not the
same as profit maximization. We have incorporated willingness to pay for changes in footprint into the
generalized cost definition, so at least in that respect the model now uses an approach for vehicle size
changes that is consistent with the objective of maximizing profit.
Rationale: The model structure allows the producer to consider only the effect of market share on its
generalized costs and does not allow the producer to consider the effect of overall sales volumes nor to
strategically adjust average prices. This prohibits producers from adjusting to changes that may affect
overall volume, the way producers seeking to maximize profit would in practice.
• For example, a more stringent policy may increase vehicle costs overall and reduce overall
demand for automobiles. A producer observing this may choose to lower prices to mitigate the
effect on overall sales volume, if it is profitable to do so, but OMEGA2 does not allow this.
• It is understandable why EPA may choose to make this assumption, since models that do allow
firms to strategically seek to change average prices and overall sales volume can potentially
produce solutions far from today's observed sales patterns and extrapolated far from data on
which demand models have been built (e.g.: firms setting prices at infinity to generate zero
sales when generalized costs are positive; monopolistic producers pricing much higher than
oligopolistic producers, etc.). However, there may be important differences between the way
firms respond to regulation when they can consider effects of their choices, including average
pricing, on overall sales volume. OMEGA2's assumption represents a potentially significant gap
between the problem that OMEGA2 solves and the behavior of profit-maximizing firms.
3. Accounting for Competition:
Recommendation 4: Clarify how OMEGA2 handles multiple producers competing for market share, if at
all, and should consider implementing functionality that would allow users to model multiple firms in
equilibrium.
EPA's Response (JM04): Thank you for the suggestion. In the latest version of the OMEGA model, we
have implemented a "credit market efficiency" (CME) parameter, which allows the user to specify a
degree of market inefficiency in the credit trading assumptions. Specifically, this involves a 2-pass
modeling approach, where in the first pass, manufacturers are considered to take full advantage of credit
trading to minimize generalized costs. The achieved C02 values from the first pass are used in a second
modeling pass, where manufactures are run individually. Manufacturers that have C02 levels above the
C02 standard (i.e., target) in the first pass will attempt to achieve C02 levels closer to the C02 standard
in the second pass. While manufacturers that are at or below the C02 standard in the first pass will retain
that level in the second pass. As a result, some credits for the overall industry will go unused. The
amount of unused credits will depend on the specified credit market efficiency value. We did not
implement profit-seeking behavior.
24
-------
Peer Review Report
Work Assignment 3-05/4-05, Contract 68HE0C18C0001
Rationale: The demo uses a single monopolist producer with a fleet of products. In practice, multiple
producers compete with one another for market share, and competition drives prices down. The
simplicity of a single producer model can be helpful; however, the optimal way for a monopolist to
respond to regulation may differ substantially from the equilibrium solution of an oligopoly responding
to regulation. The documentation states that "The user can choose to model the producers either as an
aggregate entity with the assumption that compliance credits are available in an unrestricted market
(i.e. 'perfect trading'), or as individual entities with no trading between firms," but it is not clear how
equilibrium among competing firms is handled. There is no reference to equilibrium among firms in the
documentation. If OMEGA2 is generalized in a way that allows it to model profit-seeking behavior, it
may be important to simultaneously allow it to model competition among producers.
4. Modeling Multi-Year Producer Strategy:
Recommendation 5: Pursue a revision of OMEGA2 in which firms can make strategic decisions
considering multiple years at a time and allowing for users to model dynamic factors such as
compliance strategy, volume-based production cost, and volume-influenced consumer demand.
Rationale: The OMEGA2 model structure solves Eq(l) for each time step (year) separately. This has a
number of implications.
• Compliance: In OMEGA2, firms attempt to achieve compliance in every time step, prohibiting
firms from strategically banking credits in some years to use in future years. This may result in
overestimating the cost to producers of a policy because it limits the options available to
producers to comply with the policy. EPA states that "In a future revision, we plan to consider
incorporating producer decisions that are intentionally under- or over-target based on the
assumption that producers make strategic decisions looking beyond the immediate present to
minimize generalized costs over a longer time horizon." An effort to integrate this option would
be worthwhile.
EPA's Response (JM05): Thank you for the suggestion. We have implemented logic in the latest version
to prioritize historical banked credits, so that, if possible, those credits will be used before expiring. We
have also implemented logic so that if a manufacturer is unable to meet the C02 target each year, that
the manufacturer will aim to be under the C02 target in subsequent years, until that deficit is offset using
credit carry-back. We have not implemented logic that would allow a manufacturer to reduce total costs
over a multi-year timeframe by using strategic over- or under- target performance in particular years.
• Production Cost and Learning: OMEGA2 permits technology costs to decline over time
exogenously, as defined in input files, but the structure does not allow for experience-based
learning, where costs decline as a function of cumulative production volume. Experience-based
learning could be an important factor in firm compliance strategy, where firms may choose to
produce unprofitable EVs today in order to reduce costs so that EVs are more profitable
tomorrow. Arthur Yip's Carnegie Mellon Dissertation (Ch4) provides an example of strategic firm
equilibrium under GHG regulation and experience-based learning where firms solve for multiple
years of decisions simultaneously. When expanding OMEGA2 to address multi-year decision
strategy in (a), EPA should consider whether the model can adopt user-provided endogenous
technology costs that depend on volume. Such flexibility could accommodate volume-based cost
reductions from economies of scale as well as learning and other factors.
EPA's Response (JM06): Thank you for the suggestion. For most technologies within OMEGA, while
learning is applied on a year-over-year basis, it is meant to reflect the learning-by-doing phenomenon or,
in effect, volume-based learning. This should not be confused with the concept, reflected in the term
25
-------
Peer Review Report
Work Assignment 3-05/4-05, Contract 68HE0C18C0001
"time-based learning/' that EPA introduced in a past analysis but subsequently rejected as an
inappropriate characterization of the learning-by-doing phenomenon. The one exception is battery
costs, where in the latest version of OMEGA we have implemented learning-by-doing endogenously as
the reviewer suggested.
• Consumer Demand and the Neighbor Effect: Consumer demand for new technologies can
increase when they observe these technologies being adopted by others (Mau et al.. 2008). If
OMEGA2 were flexible enough so that the consumer demand function could consider cumulative
prior sales of each technology, current share of the fleet, or other such factors, a multi-year
producer strategy may consider how investment in new technologies today (like EVs) could
change future consumer demand.
If EPA were to generalize OMEGA2 to address the recommendations above, the problem that OMEGA2
solves for a single producer might look more like Eq(2), with changes in blue (credit banking and
trading could be added as additional variables, not shown):
EPA's Response (JM07): Thank you for the suggestion. In the current version of OMEGA, the share-
weight 'S' curves for BEV and ICE demand are exogenous. We will consider making the 'S' curve a
function of BEV volume in a future version.
maximize n(x1,x2, ...,xr)
with respect to xt =
Pt
L vt
vt
maximize discounted expected future profits (which could be
reduced to minimizing producer generalized costs if the user
sets revenues appropriately (e.g.: to zero))
with respect to decision variables x that include the attributes a,
prices p, and production volume v of the producer's vehicles
across all time steps t,
subject to
g(at,vt Vt) < 0
qt(at'Pt) - vt = 0 Vt
a 6 Mnxm
m 6 Mn
V G Mn
subject to the following constraints:
• the fleet must comply with GHG standards;
g(a, v) is the policy module, which takes attributes and
shares as inputs and returns a vector of Mg GHG credits in
excess or short of the standard across all years t
• every year t production volume shares must equal demand
shares;
q(a, p) is the consumer demand model, which takes
attributes and margins (price multipliers) as inputs and
returns a vector of shares sales volumes, one for each
vehicle
• a is a vector with m attributes for each of the producer's n
(composite) vehicles. One of these m attributes is GHG
emissions per mile.
• m is a vector with one markup value (price multiplier) for
each of the producer's n (composite) vehicles
• v is a vector with one production volume share value for
each of the producer's n (composite) vehicles
Eq.(2)
26
-------
Peer Review Report Work Assignment 3-05/4-05, Contract 68HE0C18C0001
A model with only a single producer, as above, would be a monopolist, which may tend produce optimal
solutions with substantially higher prices than today's competitive auto market. An oligopoly could be
modeled using multiple producers to find a Nash equilibrium. Nash equilibria could be solved numerically
either (1) by sequentially iterating across the producers until convergence or (2) by solving for the first order
(KKT) conditions of Eq.(2) simultaneously across all producers. For example, see Figure 5-6 of Shiau and
Michalek (2007), provided below.
<^~Starting poirt~^>
r
Figure 5: Flowchart of the sequential optimization
algorithm
Equilibrium solution^)
Figure 6: Flowchart ol Lagrangian FOC algorithm
Figure 2: Figures from Shiau and Michalek (2007) showing two approaches for finding Nash equilibria among competing producers.
An important risk in considering expanding scope of OMEGA2 to model equilibrium behavior is that
equilibria may not exist or may be nonunique for some problems.
2. Comments on OMEGA2's Solution Approach
The approach to solving Eq(l) in OMEGA2, as I understand it, is to discretize the space of the decision
variables using user-defined resolution (which is iteratively refined in the consumer module iteration) and
test all combinations using full factorials in the a, v subspace and in the p subspace. Specifically,
1. Solve minimize c(x) with respect to a. v subject to g(a. v) < 0 (given p = mkfx')).
First a full factorial sweep over aG and v is executed, where aG is a vector that contains all of the
elements of a corresponding to GHG emission levels of the vehicles (there may be other attributes in a
that affect demand but do not affect the policy module). vTk(x) and g(a, v) are calculated for each
point, and among all the results the point selected is the one that minimizes vTk(x) subject to
g(a,v) < 0 (within tolerance).
27
-------
Peer Review Report
Work Assignment 3-05/4-05, Contract 68HE0C18C0001
2. Solve qfa. p) = v and vTpfx) = mvTkfx) with respect to p for fixed a and v.
Next, a full factorial sweep over cross-subsidy
multipliers is executed (with iteratively tighter
resolution and iteratively smaller domain bounds,
q(a, p) is calculated for each point using the
consumer demand model, and among the results
a. If any of the points solves q(a, p) = v
and vTp(x) = mvTk(x) within
tolerance, go to step 4.
b. If not, take the point Qbest with the
smallest error (||q(a, p) — v || and
|| vTp(x) — mvTk(x) ||) and go to Step 3.
3. Solve
minimize c(x) with respect to a subject to g(a. v) < 0 for fixed v (p = mkfx')').
If Step 2 failed to find a solution, then Step 1 is repeated with v = qgEST found in Step 2: A full factorial
over a is executed with fixed value for v, and among all the results the point selected is the one that
minimizes k(x) subject to g(v, a) < 0 (within tolerance). Return to Step 2 and iterate until a solution is
found or an iteration limit is reached.
4. The composite vehicles used in the iterations are decomposed into individual vehicles.
My primary comment about this solution approach is (1) the potential for it to produce suboptimal solutions
and (2) the implications of suboptimal solutions for OMEGA's purpose of comparing policies.
Suboptimality Concern #1
It is not clear to me under what conditions this algorithm can find the solution to Eq(l). Decomposing the
problem with variables a, p, v into (1) a subproblem in the a, v space and (2) a subproblem in the p space is
possible, but decomposition methods typically require certain types of coordination strategies to ensure
convergence to a system optimum.
In OMEGA2's approach, if the solution to Step 1 produces values for a, v such that a vector p exists that
solves Step 2, the algorithm will accept this as the solution, even if the combination a, p, v together do not
minimize the producer's objective. Such a point will only be known to be a system optimum (solving Eq(l)), in
general, if the producer's objective does not change across all permitted values of vehicle prices. I believe
OMEGA2 aims to impose exactly this restriction by requiring that sales-weighted average prices are equal to
baseline sales-weighted average prices at a 1.5 X margin. EPA describes this as "forcing any cross-
subsidization strategies to be revenue neutral"; however, this requirement is restrictive, as discussed
previously, and any future generalizations of OMEGA2 permitting producers to adjust prices in more flexible
profit-seeking ways would violate this condition, leading the OMEGA2 algorithm to produce suboptimal
solutions.
EPA's Response (JM08): Please see our responses to reviewer comment JM01 about the modeling of
profit maximizing behavior.
Suboptimality Concern #2
The iteratively refined grid search approach is not guaranteed to find a solution if one exists, except as the
resolution approaches infinity. Consider the hypothetical example below. The search grid here (for producer
cross-subsidy multipliers) is used to search for points that satisfy two constraints within tolerance: (1) market
2025_0 Producer Cross Subsidy Multipliers
+ hauling
o non hauling
OOf®
© 0OO®
Figure 3: Full factorial sweep grid search
implemented in OMEGA2
28
-------
Peer Review Report
Work Assignment 3-05/4-05, Contract 68HE0C18C0001
share proposed by the producer matches market share resulting from the consumer model and (2) sales-
weighted average price matches the reference average price. In the figure, I create one hypothetical curve
representing the set of points that satisfies each of these conditions. The shaded area surrounding each
curve represents the tolerance. The intersection of these curves represents the point that satisfies both
conditions. The iteratively refining grid search strategy will miss the solution in this case because among the
initial points tested, the point at [1.05,1] is closest to satisfying both conditions, so that will be the center
point of the refined grid, but the actual solution is closer to [0.97,1], which will be missed by the refined grid
with reduced bounds in this case. This issue of iteratively-refined grid search missing solutions is exacerbated
in higher dimensions.
EPA's Response (JM09): Thank you for the comment. We have made revisions to search algorithm to
implement the cross-subsidy search by market class (body style, currently), rather than attempting to search
multiple market classes simultaneously as was the case in the demo. Separating the searches simplifies the
search space and improves the search results. Our testing has shown expected results.
1.06
1.04
1.02
20?5_0 Producer Cross Subsidy Multipliers
O 1
0.98
0.96
0.94
0.94 0.
Hypothetical locus of
points that satisfy
market share condition
(match producer shares
within tolerance)
Point from initial search
grid (dark blue) closest
to satisfying both
constraints
Hypothetical locus of
points that satisfy sales-
weighted average price
condition (maintain
sales-weighted average
price)
Figure 4: Hypothetical example where the iteratively-refined grid search approach will concentrate on the right-hand side of the
search space because the grid point there is closest to satisfying the two equality constraints simultaneously within tolerance
(represented by the shaded area), even though the solution is actually on the left-hand side of the search space.
Suboptimality Concern #3
A third concern is that the OMEGA2 search algorithm uses iteration limits as termination criteria, and it
carries forward results from prior years when the algorithm fails to converge within iteration limits. This is
another reason why the OMEGA2 search algorithm could return suboptimal solutions.
EPA's Response (JM10): The algorithm has been revised so that the model will always converge between the
producer and consumer. Please see our response to reviewer comment JM17.
Implications of Suboptimal Solutions
For all of the above reasons, my concern is that if the OMEGA2 algorithm for solving Eq(l) produces
suboptimal solutions, the solutions may be more suboptimal for some policy scenarios than others, making
comparisons of policies unreliable and potentially biased in systematic ways. For example, if the OMEGA2
algorithm is systematically better at solving problems with policies of type A than it is at solving problems
29
-------
Peer Review Report Work Assignment 3-05/4-05, Contract 68HE0C18C0001
with policies of type B, then users may obtain results suggesting that policy type A produces better outcomes
than policy type B, even if policy B actually has a superior true system optimum (see figure below). This
possibility could lead users to make conclusions about which policies are best based on how good the
algorithm is at solving each policy case, rather than based on which case actually has better outcomes.
o
O
©
CD
G
a>
O
S-c
o>
o
2
o
Policy A has lower producer generalized
cost at the (suboptimal) solutions found
by the search algorithm
Policy
lower
gene]
cost a1
globa
B has
producer
ized
t the true
solution
3il:
Policy A Policy B
¦ Global minimum ¦ Suboptimal solution found
Figure 5: Hypothetical case where Policy B has lower producer generalized cost at the true system global optima, but the search
algorithm finds a better suboptimal solution for Policy A than for Policy B.
Alternative Search Algorithm Options
Alternative approaches to solving Eq(l) for global system optimality that may mitigate the risk of suboptimal
solutions include
1. Conduct search in the entire x space simultaneously, rather than decomposing the problem as
OMEGA2 does. Options include:
o Deterministic Global Optimization: If the functions (c, g, and q) can be restricted to algebraic
forms, a modern convexificiation-based branch-and-bound algorithm, such as BARON, could
be used to guarantee global solutions to nonconvex problems within a specified tolerance
(the algorithm requires the functions to be available in algebraic form in order to generate a
series of valid convex relaxations of nonconvex functions). This approach has the key
advantage that global optimality can be guaranteed, avoiding concerns about biased
comparisons due to algorithm suboptimality. The key disadvantages are (1) functions must
be algebraic and avoid trigonometric functions and (2) the computational efficiency of
algorithms like BARON can depend on the functional form,
o Multistart Local Optimization: If the functions (c, g, and q) cannot be restricted to algebraic
forms but the optimization variables are all continuous, efficient NLP local solution
algorithms, such as SQP with BFGS, could be used to solve the problem for a local solution,
and an array of SQP-BFGS starting points (such as the initial search grid sweep in the current
approach) could be used to search for the best local minimum. The key advantage of this
approach is that it can typically efficiently produce local solutions within tight tolerances, and
with a sufficient array of starting points it will typically find the global solution. The key
disadvantage of this approach is that it does not guarantee global optimality, and NLP
30
-------
Peer Review Report
Work Assignment 3-05/4-05, Contract 68HE0C18C0001
algorithms are not as robust as LP algorithms, depending on the problem formulation and
scaling, so there may be cases where they could stall or fail to produce local minima,
o Grid Search: EPA could extend its current full-factorial grid search approach on discretized
decision variables but conduct the grid search simultaneously in the full x space rather than
decomposing the space into an a, v search and a separate p search. This approach would
approach system optimality as the resolution approaches zero, but it would still leave open
the possibility of suboptimal solutions in practice due to Suboptimality Concern #1 and #2,
and scalability will be limited as dimensionality increases,
o Other Gradient-Free Methods could be used to search the full x space, such as response
surface methods, pattern search, evolutionary algorithms, or other methods. These methods
generally do not guarantee system optimality (except as iterations approach infinity), so the
disadvantages to these approaches are similar to the current OMEGA2 approach of
iteratively refining grid search.
2. Use formal decomposition algorithms: The problem can be decomposed similarly to OMEGA2 but
using a method guaranteed to produce system optima. The literature on multidisciplinary design
optimization covers a variety of decomposition approaches and algorithms. Khajavirad and Michalek
(2009) provide one relevant approach for finding system optima while decomposing nonconvex
MINLPs into quasi-separable subproblems using BARON to globally solve the subproblems in order to
improve efficiency. A similar approach could be used for OMEGA2 to guarantee global optima while
decomposing the problem into compliance and market subproblems if the producer generalized cost,
consumer demand, and policy compliance functions can all be written algebraically.
Recommendations
1. Recommendation 6: Flag suboptimal results prominently.
When comparing across policy scenarios, the graphical outputs do not indicate whether differences
in results across the scenarios are due to differences in the producer solutions across scenarios or
differences in the algorithm's performance across scenarios. I suggest that the results should
prominently differentiate between results where the OMEGA2 algorithm can be confident that the
result presented is a system optimum vs. results that may be suboptimal (such as when iteration
limits are reached).
EPA's Response (JM11): The algorithm has been revised so that the model will always converge between
the producer and consumer. Please see our response to reviewer comment JM17.
2. Recommendation 7: Consider an alternative solution approach.
I recommend EPA assess whether an alternative solution approach can be used that will consistently
return system optima and that can be flexible to variations in modeling assumptions recommended
in Section 1 of this review.
EPA's Response (JM12): Thank you for the comment. We have considered this approach but determined that
it is to numerically intensive to simultaneously search compliance and pricing strategies as a full cross-subsidy
search would have to occur at every point in the search. As a compromise, in the latest version of the model
we have added additional dimensions to the pricing search, by body style, and producer-consumer shares
converge one market class at a time, rather than all at once as in the provided demo. This provides a more
incremental approach to taking consumer shares and allows the producer to adjust their production
decisions as the process proceeds. In the limit, for a large number of market classes, this would approximate
a simultaneous search.
31
-------
Peer Review Report
Work Assignment 3-05/4-05, Contract 68HE0C18C0001
a. Deterministic Global Optimization: The most promising approach, in my estimation, may be
to use BARON (which has a Python interface), and this is what I recommend pursuing. The
applicability of BARON will depend on the ability to represent functions algebraically (or
logically with disjunctions), and the practical performance of BARON will depend on the form
of the nonconvex functions, so these must be considered.
b. Deterministic Global Optimization with Decomposition: I recommend that EPA also consider
continuing to decompose the problem into compliance and market subproblems for global
bounding using BARON to potentially improve the efficiency of global search, as in Khajavirad
and Michalek (2009).
c. Deterministic Local Optimization with Multistart: Alternatively, if the functions cannot be
written algebraically or BARON is otherwise not a good fit, I recommend the gradient-based
local NLP approach (SQP with BFGS) using multistart to search for global minima. Such an
approach will have much tighter tolerances than the current grid search approach, may be
more likely to consistently find the global minimum, and may reduce algorithm artifacts in
the results (such as credits earned and spent as a pure artifact of algorithm search
tolerance).
d. Expanded Grid Search: Performing the current OMEGA2 grid search approach in the full x
space, rather than in separate decomposed spaces, is another alternative to address
Suboptimality Concern #1, though it may not address concerns #2-3, and computational cost
may be prohibitive as dimensionality grows.
Other Recommendations
1. Producer Generalized Cost
Recommendation 8: Update documentation to clarify and ensure consistency in definitions of
producer generalized cost and clarify for the user what it can and cannot capture. Clarify that users
need to include fleet production cost less fleet revenue (not WTP) in order for the objective to be
aligned with (price constrained) profit maximization.
EPA's Response (JM13): Thank you for the suggestion. We will consider updating the documentation in a
future version.
Rationale: Producer generalized cost is described a varied of ways in the document that seem
imprecise or inconsistent - or at least could be clarified.
Producer generalized cost as profit
o "The term producer generalized costs is defined as the net of vehicle production costs and the
producer's view of consumer's willingness to pay for that vehicle." This definition implies that
the value consumers pay for the vehicle is included, suggesting that producer generalized
cost will be negative for a profitable product (sale price > production cost). But this is not
how the demo example works.
o "The total of the monetary expenses of bringing a product to the consumer, and the value the
producer expects can be recovered from consumers at the time of purchase." This statement
has the same apparent implication as the prior.
o "This assumption that producers will attempt to minimize their generalized costs is consistent
with a producer goal of profit maximization." This statement again seems to sound like the
prior two.
32
-------
Peer Review Report
Work Assignment 3-05/4-05, Contract 68HE0C18C0001
Producer generalized cost as production cost plus fuel cost
o "In the demo example, producer generalized costs represent the marked-up production costs,
plus a price modification (i.e. an external subsidy incentive, if applicable), plus the portion of
fuel costs that the producer assumes to be valued by the consumer at the time of purchase."
This statement appears to match how producer generalized cost is handled in the demo. In
contrast to the "value consumers pay" described above, in this statement it is more like
production cost plus a relative negative value to consumers of future fuel costs,
o "producer generalized costs ... include 5 years of fuel costs at 15,000 miles per year that the
producer assumes are valued by consumers at the time of purchase (as defined in the analysis
context input file 'producer_generalized_costs.csv'.) Note that the producer generalized costs
are higher than the production costs, and also form a cloud with a different shape than the
blue production cost cloud." This statement is a specific case of the prior statement.
2. Convex hull cost curve:
I have three observations / recommendations about the way the cost curves are handled in
OMEGA2:
Convex hull approximations
Recommendation 9: Review the state of the art of convex hull algorithms and determine whether
OMEGA2 can generate cost curves that are precise convex hulls, rather than approximate affinity
solutions.
EPA's Response (JM14): Thank you for the suggestion. We are using the 'affinity factor' to define a
reasonable representation of the frontier, without requiring a large number of points to define the
frontier, and thus increase the computational requirements. The structure of the model makes it possible
to implement alternate approaches for approximating the frontier without necessarily requiring major
modifications to the code. We will continue to consider whether any alternate approaches can improve
the current model's identification of the frontier, and at the same time maintain or improve the model's
computational efficiency. Conceptually, the frontier is meant to be a plausible frontier, more than a
literal frontier. In reality, every point in the cloud has some uncertainty associated with it, both in terms
of cost and C02 performance. A "frontier" that cuts through the cloud of points can be considered to be
within the reasonable range of values supported by those points.
Rationale: The documentation shows examples of approximate convex hulls that are not accurate
convex hulls. For example, the figure below shows a cost curve that leaves out at least six inferior
points. The documentation describes an "affinity" parameter that the user can use to adjust this
curve, and I did not have time within the review period to understand this approach fully, but the
affinity approach appears to be able to generate nonconvex curves, depending on the affinity
parameter value, whereas the convex hull is well defined and deterministic - it is not subject to
parametric variation. I am not very familiar with convex hull construction algorithms, but I know
there are a variety of efficient convex hull identifying algorithms dating back to the 1970s, and
because OMEGA2 has a finite point cloud, it seems like it should be computationally fast to identify
the convex hull in 2D, which will simply be a polygon (the epigraph). If OMEGA2 uses an approach
that generates an approximate cost curve that is not the convex hull, how can EPA be sure that the
approximation is "good enough" or even "convex enough"? Might there be systematic biases where
the curve is better at representing some areas of the space than others?
33
-------
Peer Review Report
Work Assignment 3-05/4-05, Contract 68HE0C18C0001
Pareto set vs. convex hull
Recommendation 10: Use the convex hull of the Pareto points as the cost curve rather than the
convex hull of the full set of the points.
EPA's Response (JM15): Thank you for the suggestion. In the latest version of the model, we have added
a user option which will truncate the non-optimal points that are more costly than other points with
lower emissions. This truncation is the default behavior.
Rationale: It is not clear to me why OMEGA2 needs the non-Pareto portion of the convex hull. The
left three points of the convex hull represent the Pareto frontier, whereas the four points to the right
(circled in red) are all dominated and will never be optimal solutions (except if there are additional
attributes not shown - see next bullet).
Vehicle attributes
Recommendation 11: Revise documentation to make clear to users the requirements for which
factors must be included in their producer generalized cost estimates in order for the convex hull
cost curve approach to be a valid way to generate optimal solutions.
EPA's Response (JM16): The current default producer generalized cost includes vehicle production cost,
and the fuel cost portion and valuation of footprint that producers assume consumers consider in their
purchase decision. A user could define any other elements of producer generalized cost. The search
algorithm will attempt to minimize the generalized cost with whatever definition is used.
Rationale: Use of convex hull cost curves can be a computation-saving representation, since the
search algorithm need not search the entire point cloud. However, if consumers have willingness to
pay for attributes of the technologies in the point clouds other than C02e, then these convex hull or
Pareto curves may not identify optimal technologies. For example, suppose the technology package
circled in teal, which is inferior to other technologies on producer generalized cost and C02e metrics
alone, has other attributes that are highly valued by consumers (e.g.: rapid acceleration). The curve
can then not necessarily identify which technology packages will be optimal unless the producer
generalized cost captures consumer willingness to pay for such attributes.
• Production Cost
• Generalized Cost
—Cost Curve
veh 62 ICE Large Van truck minivan 4WD
200 250 300 350 400 450
C02e [g/mi]
Figure 6: Illustration of three issues with the cost curves: (1) the cost curves used by OMEGA2 (black) can be inferior to the convex
hull (purple) and leave out superior solutions, highlighted in green; (2) the convex hull includes points that are non-Pareto and will
34
-------
Peer Review Report
Work Assignment 3-05/4-05, Contract 68HEOC18COOO!
never be part of an optimal solution (red circles) unless (3) the optimal solution may not lie on the convex hull if technology packages
affect other attributes besides C02e that influence consumer willingness to pay (e.g.: light blue circle).
Cost versus cert_co2_grams_per_mile ice_Truck
allow upslope=True. frontier affinity factor=10
46000 — ~ "
jwrmric ^ 4?
Fig. 7.2 Cost cloud and frontier, 02 . optior)^Po5'*g&irve_frontier_affiTtity_factor=2 ii
aLLow_up5Lope=rrue Higher affinity factor follows cloud poinls more closely
Figure 7: Figure from the documentation showing that the affinity approach can generate nonconvex results that are not the convex
hull. The convex hull is shown in green.
3. Communication of credit banking results
Recommendation 12: Prominently identify results that are due to algorithm tolerance or non-
convergence separately from those that are meaningful results. In particular, use a different linetype
or color for credit transfers that are due to algorithm tolerance alone, or provide the option to
suppress spurious results like this from the graphical output.
EPA's Response (JM17): In the latest version of the model, we have continued to refine the search
algorithm so that a convergence between the producer and consumer is always achieved. Specifically, as
a last resort, if the producer is not able to find a product mix, pricing, and application of technology that
consumer will accept while meeting the producers desired Mg C02 compliance target, then the producer
reverts to the closest possible solution that is acceptable to the consumer, while relaxing the Mg C02
constraint. As a result, instead of considering whether the producer and consumer have converged or
not, the model outputs should be viewed in terms of whether the compliance targets are achieved or
not.
Rationale: It would be useful if the presentation of results would differentiate between credit
carryover that is due to algorithm tolerance vs. credit carryover that is due to inability to comply (or,
in a future version, strategic dynamic compliance). Much of the green curve credit carryover in the
demo results is just an artifact of the search algorithm and the resolution/tolerance used, not
representative of how automakers would be expected to respond to the policy.
If EPA follows my recommendation of using deterministic global search algorithms or local WLP
algorithms, this issue may be a moot point because tolerances can be set much tighter than with grid
search.
35
-------
Peer Review Report
Work Assignment 3-05/4-05, Contract 68HEOC18COOO!
2015 2020 2025 2030 2035 2040 2045 2050
tear
consolidated_OEM 2021_10_21_ll_50_22_demoj5atch-context_a_dev_ActionAlternative
Cert and Compliance Versus Year
ieg Total Cost $12202.37 Billion
target_co2e_Mg
caIendar_year_cert_co2e_Mg
mod el_y ea r_cert_co2e_M g
Figure 8: Example OMEGA2 figure with green arrows showing credit year-to-year carryover that are primarily artifacts of algorithm
tolerance, rather than realistic representations of firm compliance strategy,
4. Composite vehicles
Recommendation 13: Alert users in the documentation that the use of composite vehicles can affect
choice model predictions in systematic ways that could bias results.
EPA's Response (JM18): Thank you. We will consider adding that to a future documentation release.
Rationale: Providing flexibility to model vehicles at the individual level or at the composite level may
be useful. The handling of individual vehicles within the composite is restrictive, but the user appears
to have the flexibility to model all vehicles at the individual level if s/he prefers. EPA should be aware,
however, that the use of composite vehicles in vehicle choice models has implications for choice
model predictions (Yip et al (2018)). and these implications can affect which vehicle solutions are
optimal (Yip et al (2021)). Further, while the attributes of individual vehicles can be observed, the
attributes of composite vehicles are modeler decisions, and Yip et al (2018) discusses which choices
of composite vehicle attributes can produce choice shares equivalent to the shares that would be
produced if all individual vehicles were modeled.
Figure 9: Figure from Yip et al. (2018) showing known attributes of individual vehicles and the question of how to determine
attributes of composite vehicles to represent the individuals in choice models.
GMC
Sierra
2500HD
$34,800
19 mpg
36
-------
Peer Review Report
Work Assignment 3-05/4-05, Contract 68HE0C18C0001
5. Exogeneous fuel prices and the rebound effect
Recommendation 14: Consider expanding the scope of OMEGA2 to be capable of modeling fuel
market rebound for users that wish to model it or, if this is not computationally feasible, include a
prominent disclaimer that fuel price endogeneity is structurally excluded and the fuel market
rebound effect is ignored in this model.
EPA's Response (JM19): Thank you for the suggestion. We will consider updating that clarification in a
future documentation release.
Rationale: Because the policies that users may model with OMEGA2 could have dramatic
implications for fuel consumption (e.g.: vehicle electrification), the policies may also substantially
affect fuel prices, which will affect fuel use and emissions globally. OMEGA does not allow users to
model the effect of policy on fuel prices because they are structurally treated as exogeneous. EPA
explained that this is due to computational challenges as well as a belief that fuel price response is
likely small. But literature suggests that the "fuel market rebound effect can reduce or reverse
climate benefit of low-carbon fuels" (Hill et al. 2016) and the rebound effect is generally non-
negligible (Gillingham et al. (2016)), so this effect could potentially be substantial or even dominant,
depending on the policies being analyzed.
References
1. Gillingham, K., D. Rapson and G. Wagner (2016) "The rebound effect and energy efficiency policy,"
Review of Environmental Economics and Policy vlO nl.
2. Hill, J., L. Tajibaeva and S. Polasky (2016) "Climate consequences of low-carbon fuels: the United
States renewable fuel standard," Energy Policy v97 p351-353.
3. Khajavirad, A. and J. Michalek (2009) "A deterministic Lagrangian-based global optimization approach
for quasiseparable nonconvex mixed-integer nonlinear programs," ASME Journal of Mechanical
Design vl31 p051009 1-8.
4. Mau, P., J. Eyzaguirre, M. Jaccard, C. Collins-Dodd and K. Tiedemann (2008) "The 'neighbor effect':
simulating dynamics in consumer preferences for new vehicle technologies," Ecological Economics
v68 nl-2 p504-516.
5. Shiau, C.S., and J. Michalek (2007) "A game-theoretic approach to finding market equilibria for
automotive design under environmental regulation," Proceedings of the ASME 2007 International
Design Engineering Technical Conferences & Computers and Information in Engineering Conference,
Sept 4-7, 2007, Las Vegas, NV, USA.
6. Whitefoot, K., M. Fowlie and S. Skerlos (2017) "Compliance by design: influence of acceleration
trade-offs on C02 emissions and costs of fuel economy and greenhouse gas regulations,"
Environmental Science & Technology v51 nl9 pl0307-10315.
7. Yip, A. (2020) Integrated modelling of consumer choice, producer decisions, and policy design in the
automotive market, PhD. Dissertation, Carnegie Mellon University, Department of Engineering and
Public Policy, Pittsburgh, PA.
8. Yip, A., J.J. Michalek and K. Whitefoot (2018) "On the implications of using composite vehicles in
choice model prediction," Transportation Research Part B: Methodological, vll6 pl63-188.
9. Yip, A., K. Whitefoot and J.J. Michalek (2021) "Implications of competitor representation for profit-
maximizing design," ASME Journal of Mechanical Design vl44 nl p011705.
37
-------
Peer Review Report
Work Assignment 3-05/4-05, Contract 68HE0C18C0001
Appendix
Simplified Matlab code demonstrating (1) that minimizing producer generalized cost (including willingness to
pay) is not consistent with profit maximization and (2) that constraining average prices to be equal to
average prices under fixed margins restricts profitability and changes technology adoption decisions.
function run()
% Reports solution to two problems:
% 1) minimize average unit producer cost and
% 2) maximize average unit profit
% with respect to x === [pricel, price2, productioncostl, procluctioncostZ]*
% for a two-product producer (production cost is a proxy for technology)
% subject to sales-weighted average price constraints from OMEGA2
% given a fixed baseline margin m=1.5
% as well as without sales-weighted average price constraints "Free"
% with a logit demand function
% Set up the problem.
xO = rand(4,l); % Random starting point
LB = [0;0;0.1;0.1]; % Lower bounds
UB = [inf;inf; 1; 1]; % Upper bounds
beta = [1 -2]; % Utility function coefficients for [k;p]
options = optimoptionsffminconyMaxFunctionEvaluations'Jnf,...
'Maxlter'Jnf,'Algorithm1,'sqp'/TolXVle-M/StepTolerance^le-14);
% Minimize production cost w/ constrained prices
[x,fval,exitflag] = fmincon(@(x)cost(x,beta),xO, [],[],[],[], LB, UB,...
@(x)nonlcon(x, beta),options);
MinCost = round([x; q(x,beta); cost(x,beta); gencost(x,beta);...
-negprofit(x,beta); exitflag],2);
% Minimize producer generalized cost w/constrained prices
[x,fval,exitflag] = fmincon(@(x)gencost(x,beta),xO,[],[],[],[],LB,UB,...
@(x)nonlcon(x,beta),options);
MinGenCost = round([x; q(x,beta); cost(x,beta); gencost(x,beta);...
-negprofit(x,beta); exitflag],2);
% Maximize profit w/ constrained prices
38
-------
Peer Review Report
Work Assignment 3-05/4-05, Contract 68HE0C18C0001
[x,fval,exitflag] = fmincon(@(x)negprofit(x,beta),xO,[],[],[],[],LB,UB,...
@(x)nonlcon(x, beta), options);
MaxProfit = round([x; q(x,beta); cost(x,beta); gencost(x,beta);...
-negprofit(x,beta); exitflag],2);
% Minimize production cost w/ free prices
[x,fval,exitflag] = fmincon(@ (x)cost(x, beta), xO, [],[], [],[], LB, UB,[], options);
MinCostFree = round([x; q(x,beta); cost(x,beta); gencost(x,beta);...
-negprofit(x,beta); exitflag],2);
% Minimize producer generalized cost w/ free prices
[x,fval,exitflag] = fmincon(@(x)gencost(x, beta), xO, [],[],[],[], LB, UB,[], options);
MinGenCostFree = round([x; q(x,beta); cost(x,beta); gencost(x,beta);...
-negprofit(x,beta); exitflag],2);
% Maximize profit w/ free prices
[x,fval,exitflag] = fmincon(@(x)negprofit(x,beta),xO,[],[],[],[],LB,UB,[],options);
MaxProfitFree = round([x; q(x,beta); cost(x,beta); gencost(x,beta);...
-negprofit(x,beta); exitflag],2);
% Report results
T = table(MinCost,MinGenCost, MaxProfit, MinCostFree, MinGenCostFree, MaxProfitFree,...
'RowNames1,{'price 1','price 2','cost l','cost 2','demand 1','demand 2',...
'total cost','gen cost','profit','exitflag'});
disp(T);
end
%% Average price constraint
function [g,h] = nonlcon(x,beta)
m = 1.5; % Fixed baseline markup
[p,k] = nameVariables(x);
h = [q(x,beta)'*p - q(x,beta)'*(m*k)]; % Equality constraint
g = []; % No inequality constraints
end
39
-------
Peer Review Report
Work Assignment 3-05/4-05, Contract 68HE0C18C0001
%% Demand function
function [demand] = q(x,beta)
[p,k] = nameVariables(x);
v = beta(l)*k + beta(2)*p; % Utility function
demand = exp(v)/(sum(exp(v))+l); % Log it demand
end
%% Production cost
function [c] = cost(x,beta)
[p,k] = nameVariables(x);
c = q(x,beta)'*k; % Share-weighted production cost
end
%% Producer generated cost
function [cG] = gencost(x,beta)
[p,k] = nameVariables(x);
WTP = -(beta(l)/beta(2))*(k);
% Relative WTP based on coefficients of the utility function
% (note, WTP here is relative to the outside good, which has k=p=0)
cG = q(x,beta)'*(k - WTP); % Producer generalized cost using logit WTP
end
%% Profit
function [P] = negprofit(x,beta)
[p,k] = nameVariables(x);
P = -q(x,beta)'*(p-k); % Average unit profit
end
%% Split variables
function [p,k] = nameVariables(x)
p = x(l:2); % Prices
k = x(3:4); % Production cost
end
40
-------
Peer Review Report
Work Assignment 3-05/4-05, Contract 68HE0C18C0001
Sample Results:
MinCost MinGenCost MaxProfit MinCostFree MinGenCostFree MaxProfitFree
price 1
7
. 35
7
.4
1
.16
6 . 94
6 . 03
0
.82
price 2
0
.15
0 .
15
1
.03
10 . 84
7 . 98
0
.82
cost 1
0
.46
0 .
45
0
.86
0.1
0.1
0.1
cost 2
0 .1
0
.1
0.6
0 . 81
0 . 57
0.1
demand 1
0
0
0
.16
0
0
0
.15
demand 2
0
.45
0 .
45
0
.16
0
0
0
.15
total cost
0
. 05
0 .
05
0
.23
0
0
0
.03
gen cost
0
. 02
0 .
02
0
.12
0
0
0
.02
profit
0
. 02
0 .
02
0
.12
0
0
0
.22
exit flag
1
1
1
1
1
1
Comparing the first three columns, in this example, with this particular logit utility function parameters, the
minimum production cost solution and minimum producer generalized cost solution (which includes WTP
from the logit model) are nearly identical. Different utility function coefficients ("beta") produce different
results, and the minimum production cost solution can differ from the minimum producer generalized cost
solution. This example illustrates that including discrete choice model-based WTP in producer generalized
cost is not generally consistent with profit maximization. In contrast, including negative total producer
revenue in producer generalized cost does make it equivalent to profit maximization.
Comparing the first three columns to the last three columns, the optimal solutions when price is treated as a
free variable differ from the optimal solutions when share-weighted average price is constrained to match
share-weighted average prices under a fixed 1.5 X margin. When prices are not constrained, the cost-
minimizing producer sets prices high to generate zero sales. The solution minimizing producer generalized
cost is the same in this case, though it can change depending on the utility function coefficients, suggesting
again that WTP is not the right way to include revenue in the producer generalized cost function. Finally, the
profit maximizing solution with free prices produces higher profit than the profit maximizing solution with
average price constraints, suggesting that the average price constraints in OMEGA2 are restricting its ability
to model automaker profit maximizing behavior.
5.4 Comments Submitted by Mr. Richard A. Rykowski
5.4.1 General Comments
Re: Comments on OMEGA 2.0
Dear Ms. Waite:
I appreciate the opportunity to comment on the current version of the OMEGA2 model via the peer review
process. My comments are attached.
I've tried to organize my comments by topic through the use of headings. I apologize for the wordiness of
some of the comments. I concentrated my time on addressing as many aspects of the model as time allowed,
and not on producing a tightly worded finished product.
41
-------
Peer Review Report
Work Assignment 3-05/4-05, Contract 68HE0C18C0001
5,4,2 Specific Comments
General Model Design
Output Only Applying to One Case
The absence of figures and tables which compare the results of two or more cases is a significant weakness.
Going through the figures provided in the demo batch output, I constantly have to compare two separate
figures, except for the one showing total vehicle sales by year. In that case, the figure for the Action case of
demo batch a showed the context sales, as well as sales under the Action case. In the analogous figure for
the NoAction case, the "three" curves showing context sales, sales under the NoAction policy and
consolidated OEM sales all appear to be the same. Thus, I wonder if the figure for the Action case would have
showed the results of the NoAction case had these been different than the context sales. Thus, this one
figure might not be an exception.
EPA's ICBT tool was designed to provide comparisons of two regulatory strategies. OMEGA2 already has
access to every value of every table output for each regulatory case. It would seem to me very easy to simply
perform a comparison of cases which the user requests much as was done by ICBT. The absolute results of
any one regulatory scenario are rarely sufficient for any analysis. Thus, OMEGA2's current output for any
single case has limited usefulness on its own (e.g., estimates of technology penetration levels). The issue is
almost always how the impacts of one regulatory approach differ from those of another.
EPA's Response (RR01): The calculation of differences between sessions will typically require that results
are aggregated at some level that is of interest to the user, and values weighted according to the number
of vehicles in the aggregated group. We cannot anticipate in advance what aggregation will be of most
interest to the user (e.g., aggregation by various combinations of regulatory class, body style,
manufacturer, fuel type, etc). Instead, the model outputs, comprised of absolute values, have been
organized in a format to facilitate pivot tables and pivot charts that can flexibly provide deltas.
Odd Results
From the 2021_ll_17_ll_31_31_demo_batch-context_a_NoActionPolicy V Cost Mkt Cls.png figure, I see
that the cost of both hauling and non-hauling ICEVs decreases substantially over time, except for the bump in
hauling ICEV cost when upstream emissions begin to be included in BEV certification. I understand the effect
of learning on GHG control technology. However, the decrease in hauling ICEV cost between 2020 and 2034
is more than $2000 per vehicle. This is likely greater than the cost of the GHG technology on the vehicle in
2034, so it can't be due to learning. The cause of this decrease should be noted and explained.
EPA's Response (RR02): The demo example included with the peer review package was intended for
illustrative purposes only and was not meant to represent actual vehicle technology costs. The user can
specify technology costs in the model inputs.
Tracking of Producer GHG Credits
I understand that the treatment of GHG credits is not yet complete in the version of OMEGA2 that I
reviewed. However, I have a comment on this topic that is likely to carryover to future versions of the model
which I believe could benefit from modification.
OMEGA2 currently only calculates a producer's total GHG credits generated or used each model year. The
rationale is the unlimited level of credits which can be traded between passenger car credits and light truck
credits.
42
-------
Peer Review Report
Work Assignment 3-05/4-05, Contract 68HE0C18C0001
These credits are indeed fungible. As 0MEGA2 currently does not consider the use of banked credits in
facilitating compliance to the degree reflected in the real world, the issue is moot. However, should EPA
develop methods in OMEGA to allow its simulations of producer responses to GHG regulations to include the
use of banked or purchased credits, it is paramount that OMEGA2 track GHG credits from passenger cars and
light trucks separately.
The reason is that credits can generally only be carried forward a fixed number of years. A producer of both
cars and trucks can generate positive car credits and negative truck credits, for example, in the same year.
This producer does not need to use its current car credits to mitigate or resolve its current negative truck
credits. It can use its previously banked credits to resolve its negative truck credit balance and bank its
current car credits. This effectively extends the life of a portion of its credit bank. OMEGA2's current
approach negates this possibility, which is explicitly allowed and even encouraged by EPA's current GHG
regulations. OMEGA2 calculates car credits and truck credits separately internally. Thus, I see no reason why
OMEGA2 could not simply output each credit balance separately, as well as a combined balance. Until EPA
adds a methodology for the use of banked credits, compliance in the current model year could still assume
the immediate use of current car or truck credits to compensate for a negative credit level from the other
vehicle class. But at least the model would be set up for the eventual more realistic modeling of banked
credit use and credit trading in the future.
EPA's Response (RR03): Thank you for the suggestion. The latest version of the model incorporates credit
banking. Please see the response to comment JG13. New output files track credit balances and credit
transactions from year to year, by manufacturer. The model does not apply any restrictions that
distinguish between car and truck credits for averaging, banking, or trading. Therefore, the credit output
files do not report credits by regulatory class. However, that information can be aggregated by the user
from the detailed vehicle outputs.
Use of GHG Credits
I make two simple suggestions for simulating the use of credits in OMEGA2 as EPA develops methods to do
so in the future. First, if the model finds that credits are expiring unused, it repeats its analysis of the past 5
years and allocates the credits that would have expired evenly over each of the 5 years. This is clearly
simplistic, but clearly better than letting credits expire. Spreading their use out over 5 years also maintains a
gradual increase in the application of technology and avoids situations where the model might be tempted to
reduce the application of technology, which, while possible in certain cases, is generally problematic from a
regulatory point of view.
EPA's Response (RR04): Thank you for the recommendation. In implementing the credit banking logic,
banked credits from the base year inputs (historical credits) are used to reduce the compliance burden in
the early analysis years, up until the most recent credits expire. Credits earned during the analysis years
are used only to fill compliance gaps, if any, and are not used to strategically under-comply. Debits
accrued during analysis years are paid off, if necessary, by subsequent strategic overcompliance, with the
amount of overcompliance increasing over time if it takes more than one analysis year to repay the debit.
The strategic use of credits is an open area of study, and many compliance strategies are possible. In
general, we don't observe that the use of credits greatly affects the cost of compliance, so we have
implemented a rather straightforward strategy that has effects that are easily explained and understood.
When OMEGA2 finds a situation where compliance is not possible, I again suggest that the model repeat its
analysis of the previous 5 years increasing the stringency of the GHG standard by one fifth of the level of non-
compliance. If the model cannot converge at these more stringent standards, at least some level of credits
would hopefully be generated. If fewer than all of the five years are unable to reach these tighter standards,
43
-------
Peer Review Report
Work Assignment 3-05/4-05, Contract 68HE0C18C0001
0MEGA2 could push further in the years which were able to comply to determine if even more stringent
standards were feasible and could generate the needed credits.
EPA's Response (RR05): Thank you for the suggestion. While the latest version of the model does
incorporate the use of historical banked credits, it does not have the capability to strategically over
comply in earlier years in order to deliberately build a credit bank for later compliance, or to minimize
overall costs. Credits earned through voluntary overcompliance, however, can be used in later years to
make up a compliance shortfall, as the reviewer recommends. We will consider the possibility of
implementing strategic over/under compliance over a multi-year time horizon in a future version of the
model.
Context
Fuel Costs (p. 41 User Guide, Section 4.1.4)
Fuel costs in OMEGA2 are part of the "context" and are assumed to be unchanged. Analyses utilizing ElA's
NEMS modeling system have shown that changes to CAFE standards (and thus, GHG standards) have a
significant impact on gasoline price.1 EPA has traditionally taken its projections of future gasoline from EIA
analyses based on NEMS. Thus, it is actually inconsistent to not use this same model for the impact of
gasoline demand on gasoline price. The relative impact of CAFE and GHG standards on U.S. gasoline demand
generally far exceeds their impact on vehicle prices in percentage terms. I strongly recommend that EPA add
the capability to account for the impact of GHG standards on the price of gasoline to OMEGA2.
A question was posed on this topic by one of the peer reviewers: "Exogenous Fuel Prices: Is it necessary in
the model structure to assume that fuel prices are not affected by (potentially large) changes in fuel
consumption?"
EPA's Response was: "Given our current understanding of the relatively small magnitude of a fuel price
response, we are not currently planning to pursue an investigation or adoption of endogenously modeling
fuel prices. The challenge with endogenously modeling fuel prices is that it requires an additional feedback
loop where consumer and producer decisions influence fuel prices, which in turn influence consumer and
producer decisions. While that approach is technically feasible, it is also more difficult to model and has
possible tradeoffs with run time and tractability. If you are aware of any evidence that the fuel price
response would be significant enough to merit further consideration within the model, please feel free to
elaborate in your review."
I believe that I have done that. The change in gasoline prices found using NEMS was very significant in my
view.
EPA's Response (RR06): Thank you for the suggestion. This version of the model does not include an
endogenous fuel price response. We will consider adding that capability in a future version of the model.
Rebound Rate
The current version of OMEGA2 only allows the specification of a single rebound rate. I suggest that EPA add
the capability to specify two rebound rates. One would apply to fossil-fueled vehicles, the other would apply
to battery electric vehicles for two reasons. One, fuel cost per mile for battery electric vehicles will be much
lower than those for fossil-fueled vehicles. This will produce significant increases in electric vehicle miles
1 One of several potential references on this issue: Petition for Reconsideration of EPA's Final Rule—The Safer
Affordable Fuel-Efficient (SAFE) Vehicles Rule for Model Years 2021-2026 Passenger Cars and Light Trucks, Docket No.
EPA-HQ-OAR-2018-0283, June 29, 2020
44
-------
Peer Review Report
Work Assignment 3-05/4-05, Contract 68HE0C18C0001
relative to fossil fueled vehicles. Two, electric vehicles face challenges for fast recharging that fossil fueled
vehicles do not.
EPA's Response (RR07): Thank you for the suggestion. We have implemented the capability for the user
to assign separate rebound rates for BEV and ICE vehicles.
Price Multipliers
The current version of OMEGA2 only allows the specification of a single technology price multiplier. Past EPA
analyses (TAR) utilized indirect cost multipliers which varied by the type of technology being addressed. I
suggest that EPA address this issue in the User's Guide. Indirect cost multipliers could be embedded in the
total technology package costs input via the vehicle_simulation file and the price multiplier set to 1.0. This
possibility could be discussed in the User's Guide.
EPA's Response (RR08): Thank you for the suggestion. In the current version of the model, the user can
specify different markups by powertrain type in the input files.
Producer Module (Section 4.3)
Description of the Cost and Effectiveness ofGHG Control Technology: Vehicle_simulation file
The estimates of the cost and effectiveness of GHG control technology in OMEGA2 are contained in the
vehicle_simulation file. The vehicle types addressed in the demo file line up with the very basic vehicle types
used in much of the technology effectiveness calculations made on the Effectiveness worksheet of the
Machine from the OMEGA1 process. I assume that these 6 vehicle types were chosen because they represent
the majority of the very specific vehicle simulation results produced using ALPHA. What seems to be missing
is a more specific application of these general projections to individual vehicles, which was also performed by
the Machine.
EPA seems to be emphasizing the description of GHG control effectiveness through the use of absolute C02e
emission levels. However, individual vehicles (thousands, not the 51 included in the demo vehicles file) vary
by their absolute emission level. I see vehicle characteristics which could be used to modify both the relative
and absolute effectiveness of GHG control technology in the vehicles file. However, I did not see any
reference to the use of these attributes to do so. This seems to be a fundamental requirement in the
projection of adding control technology to the baseline fleet or determining the technology required to
enable compliance. Using the absolute GHG emissions for six vehicles seems to be a very crude method to
reflect the baseline emissions of the fleet. It is not clear from the discussion of the Producer module in the
User's Guide whether the baseline emissions listed in the vehicles file are used to adjust the emission levels
of the exemplar vehicles listed in the vehicle_simulation file or not. As discussed further below, it appears
that these emission levels from the vehicles file are not being used, at least in the generation of cost clouds
and cost curves. I see this as a fundamental issue with projecting future producer compliance with GHG
standards.
In a response to a question in this area, EPA stated: "EPA recommends the generation of RSEs (response
surface equations) to derive particular cost clouds unique to each vehicle."
It is not clear to me whether these RSEs describe GHG performance, technology cost, or both. Nor was it
clear whether these RSEs would replace the extensive lists of technologies in the vehicle_simulation file or
would be used to adjust the effectiveness of these technologies for the individual vehicles listed in the
vehicles file. I recommend the latter.
Considering both the cost and effectiveness of GHG control technology, EPA seems to be assuming that the
approach to accurately estimating the GHG effectiveness of technology needs to be the same as that used to
45
-------
Peer Review Report
Work Assignment 3-05/4-05, Contract 68HE0C18C0001
accurately estimate the cost of this technology. I do not believe that this is true. The interactions of
technology on GHG performance are very complex and overlapping. Two plus two in this space does not
equal four. That is the reason EPA developed the very sophisticated, ALPHA model. Synergies and dis-
synergies dominate this task. This is also why EPA lists about 100,000 technology packages in the
vehicle_simulation file in the demo package for each of six very basic vehicle types.
The costing of technology is also very complex, but at a completely different level. The complexity of
technology costing is developing an accurate estimate of the cost of a single technology (e.g., accounting for
the cost of the added and subtracted parts, the cost of modifications to existing parts, the difference in the
cost of assembling these different sets of parts, even changes to warranty costs, etc.). Usually, there is no
synergy or dis-synergy is the costing of two technologies (e.g., gasoline direct injection and an 8-speed
automatic transmission). When there are, such as the costing of hybrid or battery electric powertrains with
various degrees of mass reduction, these can be addressed through a thorough accounting of how the cost of
the one technology is affected by the other. In this example, the lighter the vehicle, the lower the cost of
hybridization. Equations or RSEs can be easily developed to describe these limited interactions. (See the
OMEGA_TechCosts files.)
Thus, while EPA's vehicle_simulation file lists 100,000 distinct technology packages for each of six ICEV
classes, EPA's OMEGA_TechCost file for OMEGA1 is able to provide achieve greater precision in the costing of
technology packages using a total of less than 12,000 lines for the entire vehicle fleet. By providing the cost
and GHG performance of technology to OMEGA2 in exactly the same way, EPA has forced a compromise
where one is not needed, in this case, technology costs.
The synergies or dis-synergies involved in GHG performance require the separate treatment of enormous
numbers of technology packages. At the same time, the adjustment of these final C02e emission levels to
represent individual vehicles which only differ by starting emission level, total mass or aerodynamic drag is
fairly straightforward and could be accomplished using information already being provided via the vehicles
file. This does not appear that this is being done, since neither the vehicle_simulation file nor the vehicles file
provide coefficients which would describe such adjustments. EPA's emphasis on the desirability to describe
GHG performance in terms of absolute C02 emissions seems to confirm this, since any such adjustments
would require converting these absolute emission levels into percentage changes. Also, when discussing
Figure 4-9, EPA states that the minimum cost curves for several individual vehicles overlap. The wide
variability in the vehicles described in this figure indicates that EPA is applying the absolute C02 emissions
described in the vehicle_simulation file to individual vehicles and no adjustment is being made. If correct, this
introduces significant errors in the results.
When I raised this question to EPA staff, they responded: "The reviewer is also observing that the technology
present on an existing base year vehicle is not considered when the Producer module considers the design of
a vehicle in the future. This is true in the demo example. It might be helpful to consider the producer decision
in OMEGA2 as follows: In a future year, "how should a vehicle be designed which has certain characteristics
(utility, size, performance) using available technology packages?' Note that this is a different framing of the
question than is posed in OMEGA1, which could be described as "how should the technology components of
an existing vehicle be modified without changing characteristics of utility, size, performance?" We think that
the advantages of how we are framing the question in OMEGA2 include 1) it ties emissions rates more closely
with the simulation results (mentioned above), 2) it eliminates the high model sensitivity of assigning discrete
technologies manually to the historical fleet (mentioned above), and 3) it enables the model to consider
producer changes in vehicle attributes as a policy response. We acknowledge that while there are benefits to
our OMEGA2 approach, it requires additional effort in ensuring that ALPHA results reliably represent
absolute emissions levels as a function of vehicle loads, performance capabilities, and technology
applications."
46
-------
Peer Review Report
Work Assignment 3-05/4-05, Contract 68HE0C18C0001
I think that regardless of which question is being asked, the answer must eventually converted into an impact
on vehicle cost. This can only be done accurately if one knows the starting point. In all past regulatory
analyses, EPA has used the cost of a vehicle fleet from some recent model year. These vehicle designs are not
consistent with the most basic vehicle uses in describing its ALPHA-based emission values and technology
costs listed in the vehicle_simulation file. Actual vehicle fleets display much greater variability. Some of this
variability is reflected in the six vehicle types used in the vehicle_simulation file, but the granularity there is
sorely lacking.
The approach taken in OMEGA1 basically starts with this highly varied actual vehicle fleet and adds
technology which is cost effective from EPA's point of view. This retains most of the variability in vehicle
design reflected in the baseline fleet and recognizes the cost of the solutions which producers have already
implemented. EPA may be shifting to a different approach where it projects the design of a vehicle fleet
which meets the No Action policy as well as the Action policy. While this affects the inputs to OMEGA2,
which is not a part of this peer review, I believe that the current design of the Producer would make it very
difficult for a user to incorporate the type of regulatory analysis which has traditionally been conducted by
both EPA and NHTSA. Given the amount of effort required to develop a baseline fleet at the level of detail
used by EPA and NHTSA in the past, it may be unlikely that an entity or individual would develop one on their
own. The one exception could be one or more producers desiring to demonstrate the effect of this dramatic
change in baseline fleet design on projected costs of compliance. If the direct use of a very simplified baseline
fleet is EPA's intent, I suggest that it be made explicit in the User's Guide and used as a reason for describing
the effectiveness and cost of technology packages in exactly the same way.
EPA's Response (RR09): This comment appears to be specific to the demo example provided as part of
the peer review package. The level of resolution of the base year fleet can be specified by the user in
defining the model inputs. It was not our intention to restrict the model to only a simplified base year
fleet.
Separate Treatment of Hybrid Technology
In the demo, EPA generally lumps vehicles into hauling and non-hauling categories and into ICE and BE
categories. However, the demo OMEGA runs included in the review package designate some ICEVs as HEVs.
For example, the vehicle_simulation file includes a set of ~20,000 technology packages for six basic HEV
types. HEV as also a designator for some of the vehicles in the vehicles file. HEV is the only technology shown
in the vehicles file which does not place the vehicle into a car/truck class or ICEV/BEV class.2 For example,
Figure 4-9 shows a least generalized cost curve for a hybrid, subcompact vehicle. OMEGA2 seems to be
treating hybridization as an ICE technology in a unique way by only applying technology packages to hybrids
which include hybridization when it develops the cost cloud for full hybrid vehicles. This seems to run
counter to EPA's express desire to avoid manually describing the technology present in the baseline fleet.
This unique treatment of hybrids is not discussed in the User's guide. While this is a significant improvement
over ignoring the cost and effectiveness of hybridization already present in the baseline fleet, it serves as an
example of the compromise being made with regard to non-hybrid ICE technology present in the baseline
fleet. EPA appears to be ignoring the fact that significant swathes of the current fleet already reflect
advanced engines, advanced transmissions, etc. It is also not clear how the Producer module considers the
HEV packages listed in the vehicle_simulation file when treating non-hybrid ICEVs. Presumably, it considers
the application of both non-hybrid and hybrid packages. But how the model knows to do this is unclear.
2 Fuel cell technology is also listed, but there are no FCEVs listed in the demo vehicles file and it doesn't appear that any
fuel cell technology packages are listed in the vehicle_simulation file.
47
-------
Peer Review Report
Work Assignment 3-05/4-05, Contract 68HE0C18C0001
As an aside, ICCT has shown that mild hybridization coupled with advanced engine technology can achieve
GHG performance comparable to full hybrids at much lower cost.3 These vehicles would apparently be
treated as non-hybrid vehicles in OMEGA2. This approach would likely be fine for non-hybrid vehicles in the
baseline fleet. Even if advanced mild hybrid packages reached the effectiveness of full hybrid packages, their
lower cost would lead to them being selected prior to full hybrid packages in EPA's cost cloud. If full hybrid
packages were slightly more effective, they would still appear at the left side of the cloud as the slope of the
least-cost line would become very steep at this point (large increase in generalized cost for a very small
reduction in GHG emissions. However, this would not account for the possible conversion of current hybrids
to mild hybrid designs with considerable cost savings in the future.
Conceptually, EPA's special treatment of hybridization as an ICEV technology could be extended to other
technologies. However, extending this "path" approach has been shown to be problematic in numerous
comments made to NHTSA GHG rules. EPA's approach of comparing all of the available ICEV packages at the
same time is preferable. The problem is the consolidation of the baseline fleet, not in the evaluation of the
best technology packages for a selected group of vehicles.
That said, in OMEGA1, EPA developed least-cost paths for roughly 40 vehicle types versus the six used in the
GHG effectiveness modeling and the six to twelve used in the demo package for OMEGA2. Time did not allow
me to compare the least-cost paths for these 40 vehicle types to determine is groups of them were very
similar with similar effectiveness. If the paths for vehicles types falling into the six basic vehicle types all had
the same least-cost path, then evaluating only these six would be satisfactory. If they did not, I suggest that
EPA consider a slightly expanded set of vehicle groups evaluated in the vehicle_simulation file. (This last
comment could be considered to be a comment on an input file and not on the OMEGA2 model itself.)
EPA's Response (RR10): In the latest version of the model, we have implemented the capability for
response surface representations of the available technology packages to be assigned to specific vehicles
in the base year fleet. The user can define inputs for response surfaces for any number of technology
packages, including differentiating between hybrid and non-hybrid packages, and various types of hybrid
technologies.
Vehicle Cost Clouds and Curves
Figure 4-7 shows a cloud of vehicle cost versus C02 emissions and generalized cost versus emissions for a
single vehicle, #62. I could not find a vehicle #62 in any of the input files. Nor could I find a vehicle which
matched the group of descriptors for #62 (ICE Large Van truck minivan 4WD).4 Therefore, I could not tie back
either the cost or emission values shown in the figure to any of those in the input files. Specifically, I could
not determine if the starting emission level shown in Figure 4-7 (the rightmost point in the cloud) came from
the vehicle_simulation file or the vehicles file. The former contains base emission values for roughly 18
vehicles, while the latter contains base emission values for 51 vehicles, reflecting more vehicle specificity.
This ties back to my earlier comment that the base emission level for the individual vehicle being evaluated
should be based on the actual base emission level of an actual vehicle, or at minimum a group of vehicles
3 White Paper, EFFICIENCY TECHNOLOGY AND COST ASSESSMENT FOR U.S. 2025-2030 LIGHT-DUTY VEHICLES, Nic
Lutsey, Dan Meszler, Aaron Isenstadt, John German, Josh Miller, ICCT, March 2017
4 I was unable to find this vehicle in the "vehicle.csv", "simulated_vehicles.csv" or "context_new_vehicle_market.csv"
files. While I was able to find three "Large Vans" in the context_size_class column of the vehicle.csv file, the other
descriptors of these vehicles did not include truck minivan 4wd. All three Large Vans fell into the cost class,
ice_MPW_HRL. I found 98,952 lines of data referring to ICE_MPW_HRL in the simulated_ vehicles.csv file. I presume
that each line describes a technology package. I found twenty technology packages for this vehicle class which showed
"0" in all the technology tracking columns. Of this list package ICE1027 appears to be the baseline vehicle, as it shows
the lowest vehicle cost, though it is listed second in the list (package ICE1025 is shown first).
48
-------
Peer Review Report
Work Assignment 3-05/4-05, Contract 68HE0C18C0001
which shared the same platform (e.g., engine type and no. of cylinders, transmission type and no. of gears,
degree of hybridization, etc.). Also, the rightmost emission value should reflect the technology package
present on the vehicle, as OMEGA2 should not be adding technology which is already present in the fleet.
EPA does not describe the procedure used to generate the least cost curve. EPA does refer to it as defining
the "frontier" of the cloud and that OMEGA2 automatically generates it. The location of the least-cost curve
vis-a-vis the cloud implies that the method EPA has programed into OMEGA2 applies the same or very similar
logic to that used in the Tech_Ranking_Generator which was constituted a step in the OMEGA1 process. If so,
I would commend EPA's approach. If not, it would be helpful to know what logic is used.
EPA's Response (RR11): In the latest version of the model, with the implementation of redesign cycle
constraints, base year vehicle emissions levels are carried forward until the first redesign opportunity.
Base year vehicles are aggregated into shared platforms according to powertrain and driveline type,
manufacturer, size class, etc. We have implemented response surface equations which represent
absolute vehicle emissions and complete technology packages. As a result, at each redesign opportunity
there is no risk of adding an individual technology to a vehicle that already has it.The cloud of points is
now generated internally to the model, as is the associated cost curve 'frontier'. We will consider
providing an enhanced description of the cost cloud and cost curve generation approach in a future
version of the model documentation.
EPA also does not discuss the situation where the user might desire to restrict the application of a technology
to less than 100% of a vehicle's sales. The ranking process used in OMEGA1 allowed this. I recommend that
this capability be available in OMEGA2 and discussed. I see some input files with references to BEV
penetrations, but none which refer to specific technologies. The use of limits should be discussed in the
User's Guide.
One important issue is whether this sales penetration cap apply at the vehicle level or to some group of
vehicles, which could even be industry wide. This should be considered, as well.
EPA's Response (RR12): The reviewer is referring to the application of technology penetration caps in the
previous version of OMEGA, which could restrict the share of specific technology applications. Now that
the modeling incorporates constraints on redesign cycles, BEV ramp rates, and battery production
constraints, we do not think that there is a justification or need for additional technology-specific
constraints.
OMEGA2 then aggregates the least-cost curves for all(?) of the vehicles which fall into a sub-grouping whose
relative sales mix is not affected by the Consumer module. This allows this cost curve to be used throughout
a set of iterations between the Producer and Consumer modules. Further, as one of common uses of
OMEGA2 will be to evaluate a set of GHG emission control options with no changes to the vehicle fleet or
technology availability, this same cost curve could be used across the set of OMEGA2 runs. I recommend that
EPA install an option (controlled through the GUI) which would generate the lowest generalized cost versus
GHG emission relationship for each such vehicle grouping and save the results in the input folder. The user
can then refer to this file for any OMEGA analyses which utilize the same vehicle fleet and set of available
technologies. These subsequent analyses can vary the type and stringency of GHG controls and vary
consumer response to vehicle prices, as long as these responses to not affect any vehicle types which were
originally "grouped" in the development of the least generalized cost versus GHG emission curves. This will
clearly reduce processing time or allow a much more detailed evaluation of technology cost and
effectiveness on baseline vehicles with no increase in processing time.
EPA's Response (RR13): Thank you for the suggestion. We are interested in finding ways to improve the
computational efficiency of each OMEGA run, and re-using common inputs is a reasonable strategy.
49
-------
Peer Review Report
Work Assignment 3-05/4-05, Contract 68HE0C18C0001
There are a couple of additional consideration for the most recent version of the model that would
complicate the re-use of cost curves across policy sessions. The first is that redesign constraints will
impact how producer decisions in earlier years might affect the cost curves in later years. If the use of
redesign cycles causes differences in cost curves between policy sessions (due to a different prior
application of technology) then it may be difficult to re-use cost curves across policy sessions. Second,
the model allows the user to specify different cost inputs between sessions, which would result in
different cost curves. We will continue to consider other ideas for improving computational efficiency for
a future version of the model.
Returning to the aggregation process, EPA does not describe the methodology used to aggregate the least-
cost curves for each discrete vehicle. EPA does state that the "... processes of vehicle aggregation (also
referred to as composition or the creation of "composite vehicles") and decomposition are critical for the
solution search process. First, aggregation allows the model to efficiently search for a solution without a
complete enumeration of all possible choice combinations. Second, decomposition allows the model to draw
upon the key vehicle attribute details that have been retained and are required for calculating the
compliance emissions values and estimating the consumer response."
This aggregation process must use a methodology that maintains the minimization of generalized cost to
achieve a specified GHG standard. One, this requires sales-weighting. The individual least cost curves could
actually be sales-weighted prior to combination. Two, this requires evaluation of the slope of line segments
constituting each least cost curve (classic calculus). These line segments can then be ranked according to
their slopes, starting with the one with the most negative slope and moving upwards from there. This is the
process used in OMEGA1 to select the order in which to apply technology across all of a manufacturer's
vehicles. I believe that this would be consistent with everything else said in the User's Guide. It is also very
easy to accomplish mathematically. The curves shown in Figure 4.9 do not reflect this approach for several
reasons.5 However, the figure might be figurative and not indicative of the way that OMEGA2 combines the
individual cost curves.
EPA's response to a question about the decomposition of the selected point off of the composite curve
shown Figure 4.9 indicates that it is not following the approach described above. EPA refers to the selected
point being between two points on the composite curve, which is then connected to points on each
individual curve which are also being these two points. This leads me to believe that EPA is combining the
individual curves by sales weighting the rightmost and leftmost points (and those in between?). This
approaches loses all of the value contained in each individual curve and will not lead to a true least cost
solution.
EPA also refers above to a decomposition process which "... allows the model to draw upon the key vehicle
attribute details that have been retained and are required for calculating the compliance emissions values
and estimating the consumer response." If the line segments of the cost curves are ranked by their slope, as
recommended above, the exact technology packages and the resultant GHG emissions, fuel consumption per
mile, etc. for specific vehicles can be easily be determined at any point in the process.
However, as neither the aggregation nor decomposition process were described, I can't be sure of the
accuracy of the following claim made on page 56: "Because the composite vehicle is made up of individual
5 The greatest reason is the location and shape of the hybrid vehicle curve. It starts at a low emission rate, which is
consistent with hybridization. However, it also shows the potential for a 100 g/mi C02 reduction with consistently
negative generalized cost, which is should not be feasible for a hybrid. I also note that the curve for the Large Utility ICE-
HEV does not appear in the figure. This must be because it overlays one of the curves which is shown. Either the
emissions and generalized costs for this hybrid are the same as those for non-hybrids, or they are the same as those for
a subcompact ICE-HEV. Neither of which is accurate.
50
-------
Peer Review Report
Work Assignment 3-05/4-05, Contract 68HE0C18C0001
vehicles of fixed sales shares (at least relative to the other vehicles in the same market class, reg class
combination), there is one-and only-one solution for individual vehicle costs and emissions rates that will
result in the selected option for the composite vehicle's cost and emissions rate." Anytime one is combining
values off of several "curves", regardless of whether the curves are combined using fixed weighting factors or
not, one can easily move "up" one curve and down another at will and still maintain the same average cost.
For the statement on page 56 to be correct, there must have been a specific procedure for producing and
decomposing the aggregate curves which could only lead to one decomposed solution. I recommend that it
be the process that I suggest above.
As an aside, EPA further states that: "Because a producer can offer a range of different vehicles, each with
distinct costs associated with applying technology packages, it is not likely that the lowest cost compliance
solution will be a uniform application of technology to all vehicles. Nor will selecting the lowest cost option
for each vehicle likely result in producer compliance, except in cases where a policy is non-binding." This
sounds like OMEGA2 adds technology with a negative generalized cost slope whether or not the GHG
standard requires it. If this is true, I suggest that EPA make this clear and refer to the study that demonstrates
that producers have actually done so within a short time of a new technology coming to market.
EPA's Response (RR14): To the first point, the model retains all the information associated with each
point throughout the composition and decomposition processes. So, there is no ambiguity about how to
arrive at a given point on the cost curve.
To the second point, yes, we have added a feature (producer_voluntary_overcompliance) in the latest
version of the model does allow the producer to over comply with GHG standards, as long as that
approach reduces the producer's generalized cost. To provide a hypothetical example, if emissions
reducing technology were projected to become less costly than the incumbent (e.g., BEVs less costly than
ICE) and/or the fuel costs assumed to be considered in the producer decisions are sufficiently different, a
producer would reasonably expect to apply those technologies, even if it results in overcompliance with
the GHG standards. The actual result of overcompliance will be dependent about the user's inputs for,
e.g., technology costs, GHG standard stringency, fuel costs considered, etc.
Description of the Baseline Fleet
[There is some overlap between this comment and those made on the cost clouds and cost curves.]
In EPA's response to a question about the very simple description of the baseline fleet in the demo vehicles
file, EPA mentions a desire to avoid the burden of manually identifying the discrete GHG control technology
already present in the historical fleet. I acknowledge that this is a burden. I also acknowledge that the
classification of specific technology being applied by a manufacturer is sometimes difficult and subjective.
Still, the degree of this burden does not change the importance of accurately estimating the cost and
effectiveness of GHG control technology already being deployed in the baseline fleet. Take a crude example
where a producer's vehicles are 100% turbocharged. Converting these vehicles to any lower GHG technology
that doesn't involve turbocharging will reduce the cost of this technology relative to converting a non-
turbocharged vehicle. Without a precise accounting of this fact, any projections made regarding the potential
and cost of additional GHG control would appear to have limited usefulness.
The baseline emissions resulting from the application of specific technology already on the vehicle are also
very important. Some manufacturers produce vehicles with greater mass per footprint and higher
horsepower per the higher mass than others. They may be employing relatively advanced technology, yet
achieving average GHG performance. Of course, this manufacturer can choose to reduce the mass per
footprint or power per mass, but the need to do this is only discovered by identifying the GHG emissions of
these vehicles after the incremental application of technology.
51
-------
Peer Review Report
Work Assignment 3-05/4-05, Contract 68HE0C18C0001
An EPA response to a question from the peer review group distinguishes between the fundamental questions
being asked in OMEGA1 and OMEGA2 when reducing GHG emissions. The question being asked in OMEGA1
was: "how should the technology components of an existing vehicle be modified without changing
characteristics of utility, size, performance?". The question being asked in OMEGA2 is: "how should a vehicle
be designed which has certain characteristics (utility, size, performance) using available technology
packages?"
Per my understanding of the OMEGA1 process, both questions were asked and answered when developing
the sequence of technology additions described in the market and technology files. The second question,
which EPA says guides the OMEGA2 approach, was addressed in OMEGA1 through the process of identifying
the technology packages available for each vehicle type and then evaluating the cost and efficiency of each
package. Aspects of the baseline vehicles were only retained if they were part of more effective technology
packages not yet applied. In the case of very stringent GHG standards, the final technology projections were
very similar to those identified in the least-cost process.
Taken in the extreme, the answer to the OMEGA2 question applies to both the baseline and policy action
fleet. In other words, the evaluation of a policy scenario would be done using two idealized fleets. There are
rationales for this approach. However, they differ so significantly from previous EPA analyses that I believe
they need to be explicitly presented in the User's Guide. If this is not EPA's intent, I suggest that changes be
made to the emission and cost estimation processes discussed above.
Finally, all of this is not to say that the relative value of each of the attributes of the baseline fleet have the
same value. Differences in engine design are probably very important. Degrees of aerodynamic drag
reduction can be significant, but probably less important than engine design. Numbers of gear in
transmissions should be relatively easy to quantify. Thus, judgment can inform this develop of a baseline
fleet. But some consideration of what has already been done seems very important.
EPA's Response (RR15): Thank you for the suggestion. As described in our response to RR11, the
response surface equations provide absolute emissions rates, so there is no risk of adding incremental
technologies to vehicles that already have that technology present.
In the most recent version of the model, we have implemented redesign cycle constraints, which involves
the assignment of technologies to individual vehicles in the base year. As a result, the level of potential
emissions reductions and associated cost will be dependent on the technologies assigned to a base year
vehicle. The format for representing technology packages, and the associated emissions rates, is by
continuous response surfaces. I.e., each distinct technology package has its own response surface. This
approach also allows the model to distinguish between various performance characteristics and other
vehicle attributes in the base year (e.g., aerodynamics, power-to-weight ratio, mass) and to maintain or
modify those attributes in future years.
Vehicle Redesign Cycles
I see no mention of vehicle redesign cycles in the description of the producer module. I do not recommend
the approach taken in the Volpe CAFE model to predict each vehicle's redesign year and then extend these
changes to other vehicles whose design may depend on the initial vehicle's redesign, at least not without
careful study that demonstrates that producers consistently follow this approach. Vehicle redesign is also
much too fluid and within a producer's control for such predictions to be accurate at the vehicle model level.
At the same time, producers cannot redesign all of their vehicles to a significant degree in a single model
year. With OMEGA1, the average vehicle redesign cycle was typically assumed to be 5-6 years, implying that
16-20% of a producer's vehicles could be redesigned each year. Modeling the impact of a more stringent
GHG standard at least 5 years out was a general requirement if all of a producer's vehicles were allowed to
52
-------
Peer Review Report
Work Assignment 3-05/4-05, Contract 68HE0C18C0001
receive significant technology improvements, such as a major new engine or hybridization. The user should
be warned of this requirement, since 0MEGA2 projects costs one model year after the year of the baseline
fleet. This first year of projection may actually be the current year of production or in the past due to the fact
that the baseline fleet almost always represents an historical fleet. This presumes the use of a detailed
baseline fleet, which may not be EPA's approach with OMEGA2, as discussed elsewhere.
I noticed that the production_constraints-cntxt_a input file contains minimum and maximum levels for BEVs,
which is not really a technology in OMEGA2. Do these limits apply to each producer, to the fleet, to an
individual vehicle listed in the vehicles file and produced by an individual manufacturer? Is there a way to
limit the application of technology in the current OMEGA2 model? I suggest that EPA consider ways to limit
major changes in technology across a manufacturer's fleet in the first few years of a run.
EPA's Response (RR16): In the most recent version of the model, we have implemented several changes
that are relevant to the reviewers' suggestions. First, the user can choose to specify an initial analysis
year that is not the year immediately following the base year used to describe existing vehicles and their
attributes. The model will carry forward those base year vehicle attributes as unchanged until the first
analysis year. Historical banked credits, for which data may be more recent than the base year vehicles,
would be provided up until the year prior to the first analysis year, (for example, MY2019 base year
would carryforward vehicle attributes unchanged to MY2021, with a MY2022 initial analysis year).
Second, we have implemented redesign constraints, for individual vehicles with greater detail about the
applied technology packages in the base year as described in our response to comment JG19. Also, we
have implemented a ramp rate limit, which provides in upper bound on the portion of a base year vehicle
model's production that can be converted between ICE and BEV powertrains.
Fleetwide Vehicle Sales in the First Iteration of the Producer Module Following Year One
The starting point of vehicle sales by vehicle segment in each successive model year being evaluated during
an OMEGA2 run was not clear to me. On the one hand, it could be the vehicle sales reflected in the
context_new_vehicle_market.csv file. Or it could be the sales for each vehicle segment settled upon for the
previous model year. Both have value. One potential problem with using the level of vehicle sales in the
context file is that the assumptions made by whatever entity which made the projections may differ and
overlap with those implicit in the OMEGA2 run. For example, if future sales were taken from an Annual
Energy Outlook projection, that projection would necessarily include some projection of CAFE standards into
the future, as well as technology costs. Either sets of inputs will likely differ from those being modeled using
OMEGA2.
I lack the expertise to know if the converged levels of various market segment sales depend on the starting
sales levels going into the convergence process. If they do (for the types of consumer preference
relationships being envisioned), this is a significant problem. If they do not, then it is only a matter of
computing time. Given the likely divergence between assumptions involved in projecting the levels of vehicle
sales reflected in the context file and those being modeled in OMEGA2, it would seem reasonable to use
those in the context file for a couple of model years and then begin using the sales projected in OMEGA2 for
the previous model year. For example, an OMEGA2 run projecting the cost of increasing stringent GHG
standards could be expected to require significant levels of vehicle electrification (e.g., BEVs). The source of
future vehicle sales reflected in the context file may have assumed no increase in the stringency of GHG
standards and thus, reflect very low levels of vehicle electrification. It wouldn't make much sense to project
10% BEV sales in the previous model year and start out the next model year with 2% BEVs.
EPA's Response (RR17): The model is designed to use an external sales projection, which is associated
with its own assumptions about the policy. This is called the "analysis context", and the user is required
53
-------
Peer Review Report
Work Assignment 3-05/4-05, Contract 68HE0C18C0001
to set up the policy definition in the first session of a batch to match the policy definition in the external
sales projection. The model will then automatically calibrate the sales output from that session to match
the external projection. And subsequent policy sessions within the same batch will estimate the vehicle
price and fuel consumption changes, and the resulting sales changes, relative to the first session. For
example, if an AEO 2021 projection is used for the analysis context, then the first session of the batch file
should be set up to use the 2020 FRM for light-duty GHG standards, and no IRA provisions, consistent
with the assumptions used in AEO 2021.
Consumer Module (Section 4.4)
Modeling Consumer Preferences for BEVs
The demo package includes six BEV models, which vary by size class and car-truck class. Realistic projections
of BEV penetration will likely involve BEVs which differ by range, as well. The current framework for including
BEVs in both the Producer and Consumer modules appear to be sufficiently flexible to handle treatment of
BEVs of differing range. If not, I suggest that EPA consider this likely need and modify OMEGA2 to be able to
address it.
EPA's Response (RR18): Thank you for the suggestion. The model can accommodate different BEV ranges
with the definition of separate market categories. For example, 200 mile and 400 mile range BEV
categories. The user would need to establish the appropriate sales share response inputs to capture any
differences in consumer acceptance between the two ranges.
Future Vehicle Sales by Market Share
EPA's response to a question posed to the staff clarified my understanding of the way that OMEGA2 adjusts
individual producer sales to conform with future fleetwide sales and market class sales. The spreadsheet
which EPA provided6 helped me understand the mathematical relationships involved in making allocating a
change in future total vehicle sales and change in vehicle segment mix to individual manufacturers. I suggest
that this figure be included in future User's Guides.
The approach to this issue taken in OMEGA2 basically assumes that producers maintain their shares within
market segments (e.g., pickups, sedans, CUVs) (i.e., they do not compete across these segments). It also
assumes that manufacturers do not respond to shifts in the fleetwide sales mix over time by adjusting their
own sales mix. For market segments which differ dramatically, like large pickups and subcompacts, this
assumption seems reasonable. However, in the demo files, two of the market segments listed are midsized
sedan and CUV. The auto market in the U.S. has undergone a very dramatic shift over the past 5-10 years
from sedans and station wagons to CUVs and unibody SUVs. Ford, for example, has reduced its offering of
midsized sedans to one model: the Mustang. Even this sole model is hardly the traditional family sedan. In
constrast, Ford offers six CUVs. (I've excluded the Mustang MachE and Expedition). Thus, Ford, at least, has
dramatically shifted its product offerings to adjust to consumers' preference for CUVs over sedans.
The methodology embedded in OMEGA2 is likely to produce increases in total sales for some producers and
reductions for others. This could be controversial, particularly since I am unaware of studies showing that the
assumptions being made by EPA have actually played out in the market over time. Again, for subcompacts
versus large pickup trucks, I believe that the data supporting this assumption exist. I have my doubts that this
applies to the same extent for more easily interchanged segments such as sedans and CUVs. A more intricate
projection of the effect of a change in sales by market segment may be necessary to accurately reflect
producer responses.
6 OMEGA peer review_Example context projection_20211123 (l).xlsx
54
-------
Peer Review Report
Work Assignment 3-05/4-05, Contract 68HE0C18C0001
EPA's Response (RR19): The demo example that the reviewer is referencing provides analysis context
projections by size class, but not by manufacturer. As a result, it is necessary to make some assumption
about how changes in the sales mix of size classes will impact the relative sales of individual
manufactures. The assumption that we use in OMEGA is that each manufacturer retains its relative sales
mix position, compared to other manufactures. For example, if one manufacturer makes twice as many
pickups as another manufacturer, that relationship will hold in future years even as the overall share of
pickups rises or falls.
We could consider another approach where the model used projections of both vehicle sales by size
class, and changes of relative size class mix for different manufacturers. However, we are not aware of
any such projections, at least for the timeframes required for a typical OMEGA model run.
Role of Consumer Module on Technology Addition, Vehicle Cost and Compliance: Revenue Neutrality in
Evaluating Price Cross-Subsidization
OMEGA2 constrains the set of price cross-subsidization (PCS) options to those which maintain the average
vehicle price resulting from the current iteration of the Producer module. I believe that this constraint is
inconsistent with the constraints placed on the producer in the Producer module. In the Producer module, a
wide range of technology and market shares (now understood as only referring to the split of ICEVs and BEVs
within each market class) are evaluated using generalized cost and resultant level of GHG credits generated.
It is generally assumed that technology-market share combinations with negative generalized costs will not
reduce vehicle sales. In fact, they probably increase sales, and increase producer profits. However,
acceptable technology-market share combinations are not restricted to those with negative levels of
generalized cost. The goal is compliance. Combinations with positive generalized costs are acceptable since
compliance with the GHG standards is the primary goal in the running of OMEGA2. Importantly, the
acceptance of positive technology-market share combinations means that some loss in sales may result,
which in general reduces revenues and profits.
The consumer module takes the two technology-market share combinations with the lowest level of
generalized cost which also comply with the GHG standards being considered. The Consumer module then
evaluates whether these technology-market share combinations are acceptable, which I take to mean that
the price impacts from the Producer module lead the Consumer module to project (within some tolerance)
the same levels of total sales and market shares as used by the Producer module.
I suggest that the User's Guide provide a more detailed description of what kinds of conditions can produce a
failure to find convergence. I suggest the metric which tends to fail to converge also be discussed. Finally, I
suggest that in the case of non-converge, OMEGA2 produce a report (file) which describes the final two
estimated solutions (one from the Producer module and the other from the Consumer module. This would
allow the user to better understand what OMEGA2 was unable to accomplish.
More broadly, I wonder why a PCS solution which didn't recover all costs should be unacceptable (note the
double negative). The addition of technology is clearly going to increase production costs. Technological
solutions with positive generalized costs are expected to increase vehicle costs from a consumer's
perspective and reduce sales. Reduced sales could or would likely hurt profits. It would seem that PCS
solutions which were not revenue neutral would simply reduce profits further. But if compliance is the goal,
why should this further reduction in profits be unacceptable? Revenue neutrality seems fine when a solution
is possible. However, when it is not, I fail to understand why revenue neutrality should be a show stopper.
One of the demo simulations shows a manufacturer being unable to comply with the GHG standards in the
mid 2040's. Why do this occur, since a producer's ability to convert ICEVs to BEVs seems unlimited in this
timeframe? If this non-compliance is coming from the Consumer module, this isn't a problem of non-
55
-------
Peer Review Report
Work Assignment 3-05/4-05, Contract 68HE0C18C0001
convergence, but a problem of non-compliance due to the requirement of revenue neutrality. If so, this is
another kind of situation where revenue-losing PCS solutions should be allowed.
A peer reviewer posed a related question to EPA on this topic.
Exogenous Average Vehicle Price: "Is it necessary in the models structure to assume that "the model
automatically only searches for solutions that maintain the overall average vehicle price, thus forcing any
cross-subsidization strategies to be revenue neutral"."
EPA Response: "Note that as the producer decisions about vehicle design and technology application change
in over multiple producer-policy compliance search iterations, the overall average price will change. Still, as
the reviewer notes, the overall average price is maintained from the initial producer compliance search
within an iteration. We chose to apply this constraint in order to simply the solution search (see the
discussion in Q1 on convergence.) While it would be theoretically possible to relax the constraint that
average overall price be maintained within an iteration, doing so would require other model revisions,
including a change in the objective function, so that instead of minimizing generalized costs, the producer
would, for example, maximize the delta between revenue and costs (in order to avoid the issue raised by the
commenter of reducing quantities to zero.) EPA is interested in the reviewers' thoughts on this topic,
especially regarding the need to maintain the internal consistency of the modeling, and the possible
tradeoffs that might be involved with such a revision."
I believe that my suggestion is sort of a "middle road" solution. As long as the model converges, I see no need
to change the objective function. My concern is with cases which do not converge. In these cases, I believe
that it is likely that the average vehicle price is too high to maintain sales, so sales must be reduced. Assigning
a cost to this may be difficult. (EPA refers to estimating a difference between revenue and cost, which I do
not believe is performed in the current version of OMEGA2.) Or, the model could include a decrease in sales
in this particular model year for this producer as one of the impacts of the scenario.
EPA's Response (RR20): In the most recent version of OMEGA, we have revised the search algorithm so
that convergence is always achieved. Please see our response to reviewer comment JM17. If a solution
cannot be found with all of the normal constraints applied during the iteration process, the model will
converge on the solution that is closest to meeting the compliance target. In other words, the
compliance target constraint is relaxed.
The model does provide a log of the producer-consumer iteration steps, so a user can review the metrics
from the producer and consumer perspectives that eventually lead to convergence.
Impact of Classifying FWD CUVs as Cars and AWD/4WD as Trucks
Now that OMEGA is incorporating consumer preference into its modeling, I believe that the issue of the
regulatory treatment of CUVs should also be addressed. Current GHG regulations define smaller CUVs as
passenger cars and these same CUVs with AWD as light trucks. The GHG standards for trucks are less
stringent than those for cars at the same footprint, which is the case here. Therefore, manufacturers have a
potentially strong incentive to encourage the sale of AWD CUVs in lieu of the sale of FWD CUVs.
I posed the following question to EPA staff: Can a technology include a change in vehicle segment? I'm
thinking here of adding AWD or 4WD to a 2WD CUV in order to convert from a passenger car to a light truck.
There are other possibilities, of course.
EPA Response: "It might be helpful here for us to be specific about how the model handles different types of
producer decisions:
1) Changes in non-consumer-facing technology
56
-------
Peer Review Report
Work Assignment 3-05/4-05, Contract 68HE0C18C0001
These are considered as different points in the vehicle cost cloud. The points differ only in terms
of generalized cost and emissions rates. The generalized costs can reflect changes in fuel
operating costs, but not other consumer valuations specific to the technology
2) Changes in consumer-facing attributes, within a market class
As with #1, different design choices are represented by different points in the vehicle cloud,
except there can be some assumed consumer valuation of the consumer-facing attribute
included in the producer generalized cost.
3) Shifts in market class shares
Instead of choosing between different points within the cost cloud (or more precisely, which
point on the composite vehicle frontier curve), the producer is choosing the relative shares of
two or more market classes, each with its own composite vehicle frontier curve (in addition to
choosing the point from each curve.)
Clearly, the 2WD vs AWD/4WD example given by the reviewer would be a consumer-facing attribute. This
could be implemented using #2 above, where AWD is represented as additional points with higher emissions,
and with higher production costs which are offset to some extent by some assumed consumer valuation in
the generalized costs. Alternatively, AWD could be represented using #3 above as a choice in a different
market class, in the same way that the EV vs. ICE decision is handled in the demo example. This would
require a revised consumer response in the Consumer Module to estimate the demand for AWD as a
function of price and possibly other vehicle attributes. Both options would require some characterization of
the consumer's response to AWD. Option #2 would likely be simpler to implement. Option #3 would have the
advantage of producing a response that is consistent with heterogeneous consumer demand. E.g. Not all
consumers want AWD vehicles, even at zero price premium. "
I appreciate this response, as it indicates that it may be possible to accomplish this goal using the current
version of OMEGA2. I see two difficulties with approach #2 in this situation. First, EPA mentions including
both types of CUVs in a cost cloud. I thought that vehicles could only be combined in a single cost cloud when
their market shares were constant, which is explicitly not true in this case.
Second, I wonder how enabling this conversion of FWD CUVs which are cars to AWD CUVs which are trucks
would affect the development of cost clouds and curves in the Producer module. The vehicle_simulation file
describes technology that changes a vehicle's emissions, but doesn't describe a situation where the vehicle
emissions change and the vehicle's market class also changes. In the demo package, the market share of
hauling vehicles was fixed, as was the market share of non-hauling vehicles. In the CUV situation, I think that
it would be most reasonable to assume that the share of smaller CUVs was fixed, but the split between FWD
and 4WD was variable. How would this work if the shares of hauling and non-hauling vehicles were both
fixed? Or would this simply require 3 market classes: 1) smaller CUVs, 2) hauling vehicles which were not
smaller CUVs and 3) non-hauling vehicles which were not smaller CUVs?
EPA's Response (RR21): Please see our response to JG01. The model does not currently accommodate
shifting of reg classes by changing vehicle attributes as a producer decision.
Using GHG Credits to Enable Compliance
I asked the following question of EPA: I Does OMEGA2 ever look backwards as credits are expiring to consider
their use prior to expiration?
EPA Response: "As implemented for the demo, the producer will attempt to pay back any debits or use any
credits immediately. For the demo, credits only expire if the producer was unable to be over-target
sufficiently to use them all. Again, those credits that do expire can be observed in the credit logs. We have
experimented with the use of the base year banked credits via a couple of strategies (not implemented in the
57
-------
Peer Review Report
Work Assignment 3-05/4-05, Contract 68HE0C18C0001
demo) - for example, attempting to use only the expiring credits in a given year, or using the greater of the
expiring credits or an amortized amount of all available credits in a given year (which may result in a more
level usage of credits over before expiration).
In general, a multi-year credit management strategy involving, for example, being intentionally under- or
over-target, would need to be implemented via a higher-level algorithm. We have discussed such algorithms
within the OMEGA development team, but these have not been implemented at this time. "
I encourage EPA to develop methods to incorporate the use of credits to enable compliance by producers in
OMEGA2 (2.1?). Over the past several years, EPA's own Trends Reports demonstrate that producers have
both utilized their own credit banks and purchased credits from other producers to enable compliance. The
large size of producers' current credit banks and the role of these credits could play in enabling compliance is
a critical factor in EPA's recent GHG rulemaking addressing the 2023-2026 model year standards.
There are a number of simple concepts which could be used to reasonably project credit use and trading.
One, assume that all electric producers like Tesla will sell their credits to other producers. This has been
demonstrated in past Trends Reports.
Two, with exception, credits should not be allowed to expire. This exception pertains to the expiration of
2010-2015 credits at the end of MY 2021. The size of these credits is very large, the life of these credits is
essentially ending as we speak. To allow the expiration of later credits to expire is conceptually no different
than assuming a manufacturer will voluntarily consider some of their light trucks as passenger cars when
determining compliance. This is a simple giveaway of an asset which has real value. This is unlikely to occur to
a significant extent.
Three, producer's compliance costs vary due to their apparent preferences for particular vehicle styles (e.g.,
low or high mass per footprint, low or high horsepower to mass ratio, etc.). The prices paid for credit
purchases in the past are not public knowledge. However, it is reasonable to assume that producers would
somehow "split the difference" between the cost of generating a credit by a low cost producer and the
compliance cost of the high cost producer. OMEGA2 could output the final incremental cost of each
producer's compliance strategy. In OMEGA1, this would have been simple, since sales did not change with
the addition of technology, vehicle cost/price, etc. In OMEGA2, this would be more complicated. My personal
lack of experience with consumer choice models limits my ability to make specific recommendations here.
However, if OMEGA2 will be predicting a reduction in vehicle sales due to further GHG control, the lost sales
must apply to some producers. In general, one would expect the lost sales to be greater for those
manufacturers facing higher compliance costs than those facing lower compliance costs.
Four, some consideration of producers' desire to retain a cushion of credits to ensure compliance each year
and for future years could be acceptable.
EPA's Response (RR22): Please see our response to reviewer comment RR05.
Miscellaneous Comments
General
A Table of Contents for the User's Guide would be useful.
EPA's Response (RR23): The online version of the documentation is the primary version and contains a
table of contents. The pdf version was intended as a convenience for the peer reviewers but was not
formatted consistently with the primary online version.
Why does the output for an OMEGA2 run include a duplicate version of the source code used? This seems to
just require more storage space for each run. A list of the files and their dates would seem sufficient.
58
-------
Peer Review Report
Work Assignment 3-05/4-05, Contract 68HE0C18C0001
EPA's Response (RR24): The model code is saved in oderto provide a comprehensive record. Because
the code is made up of text files, the file sizes are only several MB in total, which is extremely small
compared to the overall model output.
GUI
When I entered a new batch file into the GUI, the comment box indicated that I had an invalid output
directory. One can only change the name of the batch file and output directory in two separate steps. Instead
of a comment that I did something wrong after completing the first step, I suggest an approach which
prompts the user more clearly. One possibility would be to add two separate "buttons" which indicate that
the use wants to update the input batch file or the output directory. Each button, when clicked would
highlight and allow the user to update the indicated file/directory. Also, if there were already output in the
output directory and the user clicked "run", it would be useful if OMEGA asked the user if they wanted to
replace the existing output, like Windows does when you try to save a new file using a previously existing file
name.
EPA's Response (RR25): Thank you for the recommendation.
Cars vs. Trucks and Hauling vs. Non-Hauling
I cannot find where OMEGA classifies individual vehicles into hauling and non-hauling segments. The vehicles
files labels vehicles as car or truck. The User's Guide refers to GHG targets for hauling and non-hauling
vehicles (p. 19). It is possible that non-hauling and hauling are synonyms for passenger car and light truck. If
so, I found this confusing.
EPA's Response (RR26): In the user-definable market class package, there is a get_vehicle_market_class
method that assigns a market class to each vehicle based on the user definition. The list of names of the
market classes is provided by the user in the 'market_classes' input file. Note that for whatever classes
the user wishes to define, the user will need to update other input files that contain market class
distinctions accordingly.
Control of the Consumer Module
I recommend that EPA add a button in the GUI which would disable consumer preferences, or describe what
values to enter in the sales_share_params-cntxt_a file which would effectively disable this feature. One, this
would allow the results of OMEGA2 to be compared to past analyses using OMEGA1. Two, it would reveal the
significance of the simulation of consumer preferences on the model's projections and alert the user or
observer to relative importance of this aspect of the modeling. For example, total vehicle sales increase by
roughly 250,000 vehicles in 2034 and 2035 under the Policy Alternative. This is substantial and will be noticed
and questioned as to its importance and justification.
EPA's Response (RR27): Thank you for the recommendation. There are too many various model settings
to include all of them in the GUI, especially since each session might need to be defined separately. In the
most recent version of the model, we have implemented a 'producer_shares_mode' developer option
that can be set in the batch file, which when set to 'True' would exhibit the behavior that the reviewer is
suggesting.
3.2. Running the Demo Example
A brief description of the purpose of the Configuration File would be helpful. I think that this filename comes
out of the blue.
59
-------
Peer Review Report
Work Assignment 3-05/4-05, Contract 68HE0C18C0001
Also, the passive tense in the sentence obscures who is performing these actions (e.g., "A model run is
created", "an existing Configuration File is used"). These options are not shown on the Run Model tab. How
does one load an existing Configuration File?
EPA's Response (RR28): The documentation contains a description of how the configuration file can be
used to simplify the loading of a previous run.
3.2.1. Creating a New Model Run From The Demo Example
A prompt asking the user if they want to save the Configuration File would be helpful.
The User's guide states: "Use the file menu to save the new Configuration File. (Optional)". References to the
"File" menu at the top of the window should be in quotes, bold or somehow highlighted in the User Guide
text so that the user knows what this is. Otherwise, it can be easily confused with a file directory where
OMEGA files are stored.
EPA's Response (RR29): Thank you for the recommendation. The documentation has been updated.
3.2.2.1. Set Model Run Options
Example Multiprocessor Batch File:
This could not be tested as the Event Monitor states that the multiprocessor option is disabled. However, if it
ever is enabled, it would be helpful to know what values or text needs to be modified by the user in the
example file and what is generic and does not need to be changed. For example, is it simply a matter of the
number of CPU cores? Or does the "%BasePath% need to be changed, as well?
Figure 3.8
The upper two graphs shown in Figure 3.8 show the C02e targets by market segment. The legend shows four
lines, hauling ICEVs, non-hauling ICEVs, hauling BEVs and non-hauling BEVs. However, only two lines are
shown in each of the two upper graphs, those pertaining to the two ICEV market segments. I don't believe
that this depiction is accurate. I don't think that the two curves describe the C02e targets just for ICEVs.
Instead, I believe that they apply to all hauling and all non-hauling vehicles. If I am correct, this should be
corrected.
EPA's Response (RR30): This portion of the GUI was used as an example feature and has been removed
as the detailed information is available in the output files.
The lower two graphs show the average certification C02e values for the same four market segments. Now
there are four distinct curves. I can't be sure whether these graphs are realistic or not. However, there are
some aspects of the two ICEV curves for the two GHG scenarios which appear unreasonable.
Starting with the NoActionPolicy scenario, upstream BEV emissions are first being counted in 2035. The C02e
emissions from hauling ICEVs decrease significantly (40-45 g/mi C02e, or 10-15% in relative terms) between
2034 and 2035. This decrease is unrealistically large for a single model year (see my comment above vehicle
redesign cycles). Since the best hybrid systems only reduce GHG emissions by ~20-25%, this implies that 50-
60% of hauling ICEVs were hybridized in one model year. Then, hauling ICEV emissions in 2036 and beyond
increase over 2035 levels. Is OMEGA2 removing technology between 2035 and 2036. If so, I believe that this
is problematic. These emissions then decrease significantly again between 2044 and 2045. I don't see any
substantial changes in the market shares for the four market segments occurring between 2044 and 2045.
What is the cause of this decrease in hauling ICEV emissions in 2045?
60
-------
Peer Review Report
Work Assignment 3-05/4-05, Contract 68HE0C18C0001
EPA's Response (RR31): The results in the demo example did not have redesign cycles. Please see our
response to RR11.
Figure 3.9
Why are certification and compliance C02e levels shown on both a MY and CY basis? There are differences
between the two types of years, but these differences have usually been ignored in prior versions of OMEGA
and in the Volpe CAFE model. While MY sales begin in the previous CY, the VMT of these early sales in the
previous CY is compensated to some extent by the fact that MY sales occurring in the same CY usually are not
driven for a full year in that CY.
EPA's Response (RR32): The model year curve on the auto-generated compliance plot shows the final Mg
values after credit transfers to or from that model year. The calendar year curve shows the Mg values for
new vehicles in that calendar year (i.e. model year age 0).
As in the past, for new vehicles we are not making any distinction between a calendar year and model
year for which month the year begins.
Figure 3-10
Under the NoActionPolicy scenario, there is a sudden increase in non-hauling BEV sales in 2040, due to a
minimum BEV sales requirement. I don't recall the User's Guide discussing minimum sales requirements for
specific market segments, though I found the input file which describes them. How do these minimum sales
requirements interact with the Consumer module? Does OMEGA simply prevent the Consumer module from
reducing relevant market shares below their minimum? In the absence of a minimum market share, when
OMEGA2 fails to converge on a solution, I presume that this is due to the Producer module wanting to
increase the sale of lower emitting vehicles (probably BEVs), but the Consumer module rejecting this because
of a consumer preference for ICEVs at the prices being provided to the module. How does OMEGA2 get
around this conundrum when there is a sales minimum?
Under the NoPolicyAction scenario, why is the minimum sales level for non-hauling BEVs in 2045-50 lower
than from 2040-45 (and worse, negative)?
Under the ActionAlternative scenario, non-hauling BEV sales increase substantially in 2048 in an attempt to
produce credits for carryback to 2046 and 2047. This attempt fails and the manufacturer is in non-
compliance. This demonstrates the need to include more broad and realistic simulation of credit generation
and use in OMEGA2. OMEGA2 apparently starts looking for ways to generate credits when the 3 year window
for carrying back credits is about to expire. This is good in and of itself, but hardly a robust use of credits to
aid compliance.
This section of the User's Guide simply states that the producer could not comply with the GHG standards
given the modeled consumer response and the specified limits on producer price cross-subsidization. I
suggest that the guide present the alternative that the limits on producer price cross-subsidization could be
expanded to determine what degree of subsidization would be require to enable compliance. This issue is
addressed in another comment and may be resolved through the modification to OMEGA2 suggested there.
In general, the guide is written with only one No Action scenario and one Action scenario in mind. In reality,
no model does everything well and realistically. I suggest a discussion where the user is instructed to review
the results for realism and determine if another run with slightly different inputs would yield more realistic
results. The current version of OMEGA2 is very flexible and doesn't set much in stone. The user is responsible
for nearly everything. This onus on the user should be made clear. The only exception would be the
reproduction of OMEGA2 runs made by EPA using input file provided by EPA.
61
-------
Peer Review Report Work Assignment 3-05/4-05, Contract 68HE0C18C0001
EPA's Response (RR33): Thank you for the recommendation. We will consider adding emphasis on the
importance of the user inputs and assumptions in a future version of the documentation.
62
-------
Peer Review Report
Work Assignment 3-05/4-05, Contract 68HE0C18C0001
APPENDIX A
Resumes of Selected Reviewers
A-l
-------
-------
Peer Review Report
Work Assignment 3-05/4-05, Contract 68HE0C18C0001
Sujit Das
VITA
OBJECTIVE
A professionally challenging R&D leadership position utilizing long years of experience in research,
analysis, and knowledge dissemination on energy, water, and material efficiency in the manufacturing
sector by developing and deployment, and promoting impactful practices, tools, and technologies
EDUCATION
MBA Management Science and Computer Science, University of Tennessee 1984
MS Metallurgical Engineering, University of Tennessee, 1982
B. Tech Metallurgical Engineering, Indian Institute of Technology, Kharagpur, India, 1979.
Ranked 2nd in class with Honors.
PROFESSIONAL EXPERIENCE
1984- 1992 Research Associate, Energy Division, Oak Ridge National Lab, Oak Ridge, TN
1992 - 2005 R&D Staff Member, Energy Division, Oak Ridge National Lab, Oak Ridge, TN
2005 - Present Sr. Research Staff Member, Energy and Transportation Science Division, Oak Ridge
National Laboratory, December 1984-present
A demonstrated long-term R&D team leader in the manufacturing energy efficiency research & analysis
supported by a wide range of U.S. federal agencies (notably by various DOE/EERE offices and Fossil
Energy), and industries. Technoeconomic, and life cycle energy and environmental, and manufacturing
supply chain analyses have been the foundation R&D analysis supported the technology field-validation,
verification, implementation of various early-stage various EERE office supported technologies to drive
American competitiveness and innovation in energy, water, and material efficiency in the manufacturing
sector. The analysis supported various ORNL program manager R&D activities across many divisions in
various directorates in a wide range of technology areas, e.g., lightweight materials including advanced
fibers and composites manufacturing, advanced clean energy manufacturing, vehicle manufacturing,
water supply and demand, digital manufacturing, and electrification and energy infrastructures.
The Strategic Analysis of Advanced Manufacturing project supported by the DOE Advanced
Manufacturing Office being led currently grew more than twice to a level of ~$900K in FY20 resulting in
several new staff hire. Other new initiatives have been the multi-year technoeconomic and life cycle
analysis of coal-to-products by the DOE Fossil Energy office. Several technoeconomic and life cycle
energy models of advanced materials (e.g., carbon fiber and its composites are the only models used
widely by various DOE/EERE offices, coal-to-products for the DOE Fossil Energy office), and clean energy
manufacturing technologies (e.g., Additive manufacturing and Roll-to-Roll manufacturing) were
developed. Decision-making tools for several resource markets such as petroleum and nuclear materials
were developed in the past including the recent supply chain competitiveness analysis of clean energy
manufacturing technologies (e.g., wide bandgap materials and semiconductor manufacturing), and
integrated energy and desalination system design.
Collaborated with several industry partners in the techno-economic analysis of materials technologies
for identifying commercialization opportunities in worldwide markets. Published research in referred
journals and presented in international conferences. Several energy projects in developing countries
A-3
-------
Peer Review Report
Work Assignment 3-05/4-05, Contract 68HE0C18C0001
such as Bangladesh and India were led in the past and recent invited talks on the life cycle analysis of
lightweight materials were in Canada, China, and Europe. The following list provides highlights of multi-
disciplinary and a wide range of manufacturing energy efficiency research and analysis projects in the
areas of energy, water, and material efficiency led:
• Strategic analysis of semiconductor, smart, carbon fiber composites and water use in the
advanced manufacturing for the EERE AMO Office
• Techno-economic and life cycle energy analysis of coal-to-products, e.g., carbon fiber
composites and building insulation materials for the DOE Fossil Energy Office
• Integrated energy and desalination system design and industrial water use for the DOE
• Lead author of Technology Assessment of Thermoelectric and Wide Band Gap materials for the
ongoing U.S. Dept. of Energy Quadrennial Technology Review
• Techno-economic and life cycle analysis of carbon fiber composites for the Wind Energy
Technologies office
• Life Cycle Energy and Environmental Assessment of Aluminum-Intensive Vehicle Design
• Life Cycle Energy and Environmental Assessment of Multi-Material Lightweight Vehicles
• Supply Chain Manufacturing Competitiveness Analysis of Additive Manufacturing, Carbon Fiber,
and Wide Band Gap Materials for U.S. Dept. of Energy
• Next generation materials with energy/emissions reduction potential in the U.S. industry for
DOE Advanced Manufacturing Office
• Systems Analysis for the U.S. Dept. of Energy Fuel Cell Technologies Office
• Manufacturing process modeling of high temperature stationary fuel cell systems in the 350-400
kW power range for DOE Fuel Cell Technologies Program
• Life cycle modeling of alternative lightweight engine design options for the DOE Propulsion
Materials Program
• Market potential and infrastructure assessment of ethanol and hydrogen as alternative
transportation fuels
• Cost modeling and life cycle analysis of advanced vehicles and lightweight materials
technologies for DOE Office of Vehicle Technologies
• Material technology assessments related to Partnership for A New Generation of Vehicles
PNGV)/Freedom Cooperative Automotive Research (FreedomCAR)
• Potential of renewable energy technologies in rural Bangladesh
• Biomass refinery analysis
• Economic analysis of advanced power electronics, electric motors, and intelligent transportation
systems
• Energy efficiency of distribution transformers
• Cost of alternative fuels
• Forecasting of petroleum and uranium supplies
A-4
-------
Peer Review Report
Work Assignment 3-05/4-05, Contract 68HE0C18C0001
• Estimation of flood-stage economic damages
• The economic viability of plastics and automobile recycling
• Environmental implications of privatization of the power sector in India
• Market assessments of energy efficient technologies such as home refrigerators in India
• Inspection and Maintenance of two-wheeler vehicles in India
• Assessment of uranium resources
Visiting Fellow, Tata Energy Research Institute (TERI), New Delhi, India, October 1992-June 1993.
Developed a comprehensive, computerized, and PC-based Energy-Economic-Environment database for
TERI - the first of its kind in India and provided technical support in their ongoing energy and economic
modeling activities.
Research Assistant, Energy and Economic Analysis Section, Oak Ridge National Laboratory, September
1982-December 1984.
Documented and evaluated several EIA, DOE maintained computers models, i.e., Headwater Benefit
Energy Gains Model and the Petroleum Allocation Model. Developed a computer software "BIOCUT" for
Economic Evaluation Model for Wood Energy Plantations.
LIST OF PUBLICATIONS
BOOK/CHAPTERS PUBLISHED
"Recycling and life cycle issues for lightweight vehicles," A Book Chapter in Materials, Design and
Manufacturing for Lightweight Vehicles 2e by Elsevier Inc. (Forthcoming 2020)
Two book chapters published in "Advanced Composite Materials for Automotive Applications: Structural
Integrity and Crashworthiness," Edited by Ahmed Elmarakbi, Univ. of Sunderland, UK and
published by Wiley & Sons (Aug.'13)
Chapter 3: Low Cost Carbon Fibre for Automotive Applications (Part 1: Low Cost Carbon Fibre
Development); Chapter 17: Low Cost Carbon Fibre for Automotive Applications (Part 2:
Applications, Performance and Cost Reduction Models)
"Recycling and Life Cycle Issues for Lightweight Vehicles," A Book Chapter in Materials, Design and
Manufacturing for Lightweight Vehicles, edited by P.K. Mallick, Woodhead Publishing Limited, pp.
309-330, 2010
"Material Use in Automobiles." A Book Chapter in Encyclopedia of Energy, published by Elsevier Inc.,
Vol. 3, pp. 859-869, 2004.
"Plastic Wastes: Management, Control, Recycling, and Disposal." Noyes Data Corporation, NJ (Co-
Authored with U.S. Environmental Protection Agency and T. R. Curlee), 1991.
SELECTED REFERRED ARTICLES/PRESENTATIONS (Out of 100+ articles and presentations)
"The Energy Footprint of Automotive Electronic Sensors," Sustainable Materials and Technologies,
Volume 25, September 2020, e00195, https://doi.Org/10.1016/j.susmat.2020.e00195 (Armstrong
et al.)
A-5
-------
Peer Review Report
Work Assignment 3-05/4-05, Contract 68HE0C18C0001
"Optimized Carbon Fiber Composites in Wind Turbine Blade Design/' SAND2019-14173, Sandia National
Laboratories, Albuquerque, NM Oct. (with Ennis et al.)
"Technoeconomic Analysis of Coal Pitch Carbon Fiber Manufacturing/' presentation at the Carbon 2019
conference, held in Lexington, KY on July 14-19, 2019 (Das, S.)
"Global Carbon Fiber Composites Supply Chain Competitiveness Analysis," Technical Report, Oak Ridge
National Laboratory, Oak Ridge, TN. (Das, S. et al.)
"Life Cycle Energy Impacts of Automotive Electronics," Smart and Sustainable Manufacturing Systems 1,
no. 1 (2017): 262-288; https://doi.Org/10.1016/j.susmat.2020.e00195 (Cassorla et al.)
"Vehicle Lightweighting Energy Use Impacts in US Light-Duty Vehicle Fleet,"Sustainable Materials and
Technologies, Vol. 8, July, p. 5-13 (2016).
"Innovation in Engineering Plastics, Carbon Fiber & Composites," invited presentation at the Advanced
Design & Manufacturing conference, held in Cleveland, OH on Mar. 29-30/17.
"OEM Strategies on Lightweight Metals" invited presentation at the Advanced Lightweight Vehicles and
Materials 2016 Forum, held on 13-14 Oct.'16, Berlin, Germany.
"Life Cycle Analysis of Automotive Electronics," paper presented at the ACLCA XIV conference, held on
Sept. 27-29/16 in Charleston, SC.
"Carbon Fiber Composites in High Volume Ground Transportation: Competition Between Alternatives,"
paper prepared for presentation and publication at CAMX to be held on Oct. 13-16/14 in Orlando,
Fl.
"Life Cycle Energy and Environmental Assessment of Aluminum-Intensive Vehicle Design," SAE Paper No.
14M-0325, Warrendale, PA (Apr. 14). Also published in theS/Af International Journal of Materials
and Manufacturing, June 2014.
"Aluminum's Environmental Superiority in Automotive Reaffirmed by Oak Ridge National Laboratory -
Aluminum Offers Smallest Total Carbon Footprint Among Competing Materials" - DRIVE
ALUMINUM press release on Apr. 9 (News Article published on April 10, 2014:
http://www.greencarreports.com/news/1091379_aluminum-vehicles-save-more-energy-than-it-
takes-to-build-them)
"Global Carbon Fiber Composites Supply Chain Competitiveness Analysis," draft report for the DOE
Strategic Planning Assessment Office, May 2014.
"Metal Flow, Recycling, Energy and Environment: Al, Cu, Mg and Ti", Paper presented at the TMS Annual
Meeting on Feb. 18, '14
"Cost of Ownership and Well-to-Wheels Carbon Emissions/Oil Use of Alternative Fuels And Advanced
Light-Duty Vehicle Technologies/' Energy for Sustainable Development\lo\. 17 (6), pp. 626-641,
December 2013
"Vehicle Lightweighting - A Sustainable Pathway in the Transport Sector?" Invited paper presentation at
IUMRS-ICAM2013 Intl. Conference on Advanced Materials, Qingdao, China, Sept. 22-28, 2013.
"Platinum Group Metals (PGM) for Light-Duty Vehicles", Fuel Cell Technologies Program Record, U.S.
Dept. of Energy, Sept. 2013.
"Manufacturing Process Modeling of 100-400 kWe Combined Heat and Power for Stationary Fuel Cell
Systems," Paper no. ESFuelCell2012-91183, Proceedings of the ASME 2012 6th International
A-6
-------
Peer Review Report
Work Assignment 3-05/4-05, Contract 68HE0C18C0001
Conference on Energy Sustainability & 10th Fuel Cell Science, Engineering and Technology
Conference, held on July 23-26, 2012, San Diego, CA.
Served as one of the expert reviewers for the following four recent U.S. DOT/U.S. EPA reports "NHTSA
fuel economy rule making draft report "Mass Reduction for Light-Duty Vehicles for Model Years
2017-25"(2016)
"EPA Advanced Light-Duty Powertrain Hybrid Analysis (ALPHA) Model"(2016) "Vehicle Mass Reduction
and Cost Analysis - Heavy Duty Pickup Truck and Light Commercial Vans" report for EPA (2016)
"Costs of Medium- and Heavy-Duty Vehicle Fuel Efficiency and Emissions Reduction Technologies," Tetra
Tech, Inc./NHTSA, Feb.'15.
"Mass Reduction and Cost Analysis - Light-Duty Pickup Trucks Model Years 2020-2025," FEV, Inc./EPA,
July,'14.
"Mass Reduction for Light-Duty Vehicles for Model Years 2017-2025," EDAG/The George Washington
University Report, Apr. 2012
"Light-Duty Technology Cost Analysis Pilot Study, FEV Draft Report, "Sept. 3, 2009 "An Assessment of
Mass Reduction Opportunities for a 2017-2020 Model Year Vehicle Program," Lotus Engineering
Inc., Mar. 2010
"Lightweighting Opportunities in the Global Automotive Industry," invited presentation at the 2011
International Automotive Lightweight Materials Development Forum, held in Chongqing, China, on
Mar. 24-25,'11. (Also at the 12th IUMRS International Conference on Advanced Materials, held in
Qingdao, China on Sept. 22-28, 2013)
"Importance of Economic Viability Assessment of Automotive Lightweight Materials" invited
presentation at the 3rd Annual Advanced Lightweight Materials for Vehicles conference held on
Aug. 11-12, '10, Detroit, Ml.
"Analysis of Fuel Ethanol Transportation Activity and Potential Distribution Constraints," Transportation
Research Record: Journal of the Transportation Research Board, No. 2168, Transportation
Research Board of the National Academies, Washington, DC, 2010, pp. 136-145.
"Reducing GHG Emissions in the United States' Transportation Sector" Energy for Sustainable
Development, 15 (2011) 117-136, May 11.
"Life Cycle Assessment of Carbon Fiber-Reinforced Polymer Composites," Intl. Journal of Life Cycle
Assessment, Volume 16, Issue 3, pp. 268-282, 2011
"Battle Green," an interview article published in American Metal Market, Oct. 2010, pp. 36-40.
"Shedding Pounds On a Magnesium Diet," Automotive Engg. International, Apr. 6, 2010, pp. 34-36,
interview article by Steven Ashley.
"Analysis of Fuel Ethanol Transportation Activity and Potential Distribution Constraints," Transportation
Research Record: Journal of the Transportation Research Board, No. 2168, Transportation
Research Board of the National Academies, Washington, DC, 2010, pp. 136-145.
"Low-Carbon Fuel Standard - Status and analytic issues," Energy Policy, vol. 38, No.l, Jan. 2010, pp. 580-
591.
"Importance of Economic Viability Assessment of Automotive Lightweight Materials," invited
presentation at the 3rd Annual Advanced Lightweight Materials for Vehicles," held in Detroit, Ml
on Aug. 11-12, 2010.
A-7
-------
Peer Review Report
Work Assignment 3-05/4-05, Contract 68HE0C18C0001
"A Comparative Life Cycle Assessment of Magnesium Front End Parts/' SAE Paper No. 2010-01-0275,
Society of Automotive Engineers, Warrendale, PA.
"Primary Magnesium Production Costs for Automotive Applications/' Journal of Metals, Vol. 60, No. 11,
2008, pp. 51-58.
"A Systems Approach to Life Cycle Truck Cost Estimation," SAE Paper No. 2006-01-3562, Society of
Automotive Engineers, Warrendale, PA.
"Automotive Lightweighting Materials Benefit Evaluation," ORNL/TM-2006/545, Oak Ridge National
Laboratory, Oak Ridge, TN, Nov. 2006
"Lightweight Opportunities for Fuel Cell Vehicles," SAE Paper No. 2005-01-0007, Society of Automotive
Engineers, Warrendale, PA.
"A Comparative Assessment of Alternative Powertrains and Body-in-White Materials for Advanced
Technology Vehicles," SAE Paper No. 2004-01-0573, Society of Automotive Engineers, Warrendale,
PA.
"Back To Basics? The Viability of Recycling Plastics by Tertiary Approaches," Working Paper #5, Program
on Solid Waste Policy, School of Forestry and Environmental Studies, Yale University, New Haven,
CT, September 1996.
"Determination Analysis of Energy Conservation Standards for Distribution Transformers. ORNL-6847,
Oak Ridge National Laboratory, Oak Ridge, TN, July 1996.
AWARDS & PROFESSIONAL ACTIVITIES
Ranked within the top 2% of world scientists in the Energy research discipline for 2019 Citation Impact
One of the instructors on the Lightweight Innovations for Tomorrow (LIFT) Advanced Materials for Case
Western Reserve University Siegel Continuing Professional Studies (CWRU) program - 2018
Recipient of the 2017 U.S.-lsrael Integrated Energy and Desalination System Design Challenge Winner of
the U.S. Department of Energy Policy and Systems Analysis Office
Recipient of the Society of Automotive Engineers 2015 Forest R. McFarland Award
Three-year (2016-2019) term as a Member-At-Large of SAE Engineering Meetings Board
American Center for Life Cycle Assessment Policy Committee member (2018-Present
Awarded 2004 Journal of Metals Best Paper by the Mineral, Metals, and Materials Society (TMS)
Chair of Society of Automotive Engineering (SAE) Sustainable Program Development Committee (Jan.
2013- Dec. 2014)
Member of Transportation Research Board (TRB) Committees (2008- 2016)
Transportation Economics
Alternative Transportation Fuels and Technologies
Invited Speaker on the Life Cycle Assessment of Materials by Beijing University of Technology, China
Conference Session Organizers for SAE and TRB
Research proposal review committee member of National Sciences and Engineering Council of Canada
A-8
-------
Peer Review Report
Work Assignment 3-05/4-05, Contract 68HE0C18C0001
External Peer Reviewer of EPA's draft report on vehicle mass reduction and cost analysis of light-duty car
and pickup truck, and medium- and heavy-duty vehicles
Peer Reviewer for National Sciences and Engineering Council of Canada
Peer Reviewer for Several Energy and Environmental Related Journals
A-9
-------
-------
Peer Review Report
Work Assignment 3-05/4-05, Contract 68HE0C18C0001
John M. German
PROFESSIONAL EMPLOYMENT
Sept. 2018 to present JG AUTOMOTIVE CONSULTING, LLC
• Consulting services for automotive technology, compliance and enforcement, policy,
rulemaking, and consumer behavior.
January 2009 to August 2018 SENIOR FELLOW, INTERNATIONAL COUNCIL ON CLEAN
TRANSPORTATION
• Primary responsibility for technology innovation and U.S. policy development.
• Part of team that conducted testing uncovering VW defeat devices and efforts to improve
enforcement worldwide.
• Wrote series of six technology assessment papers in cooperation with suppliers.
February 1998 to January 2009 MANAGER, ENVIRONMENTAL AND ENERGY ANALYSIS, PRODUCT
REGULATORY OFFICE, AMERICAN HONDA MOTOR CORPORATION
• Provide policy and technical analyses on vehicle emissions and energy issues.
• Liaison between Honda R&D, both in the U.S. and Japan, and external organizations, including
government agencies, environmental groups, other manufacturers, academia, and state
representatives.
• Primary Honda representative on fuel economy and global warming issues, including testifying
before Congress six times, writing testimony, writing responses to CAFE rulemaking, and making
presentations.
October, 1986 to January, 1998 SENIOR TECHNICAL ADVISOR, U.S. EPA OFFICE OF MOBILE SOURCES.
• Supervised up to 8 employees, managed development of regulations and guidance, and served
as technical consultant on a wide variety of issues.
• Technical manager for study on Tier II emission standards for cars and light trucks.
• Designed and managed extensive research project evaluating in-use driving behavior and its
impact on emissions in support of new Supplemental Federal Test Procedure regulations and
requirements. Created and managed extensive usage of teams across organizational
boundaries.
• Managed the development of the Nonroad Engine and Vehicle Emission Study.
• Managed rulemaking for Cold Temperature Carbon Monoxide Standards.
• Worked with transportation planners to help create and develop a computer simulation model
for vehicle emissions (as part of 2011-13 NCHRP project).
• EPA senior technical advisor on greenhouse gas and fuel economy issues, including CAFE
alternatives, in-use fuel economy factors, and advanced technology. Active member of EPA
global warming team and an inter-agency modeling team.
• Created and managed rulemaking assessing LDT CAFE test procedure adjustments.
• Developed policy guidance for driver-selectable devices, coastdown procedures, dynamometer
power absorption settings, and model year definition and duration.
May, 1985 to Sept., 1986 TEAM LEADER, U.S. EPA OFFICE OF MOBILE SOURCES.
• Supervised 3 employees and managed manufacturer motor vehicle emissions compliance
program.
• Wrote guidance on numerous certification procedure issues.
A-ll
-------
Peer Review Report
Work Assignment 3-05/4-05, Contract 68HE0C18C0001
December, 1981 to May, 1985 ENGINEERING SUPERVISOR, CHRYSLER POWERTRAIN.
• Supervised 6 engineers, supported product planning, and developed strategies to optimize
vehicle fuel economy and to ensure compliance with all fuel economy requirements.
Chrysler's principal technical advisor on fuel economy and methods to improve CAFE
Provided technical analyses and written responses to proposed regulations.
Represented Chrysler on fuel economy matters with EPA and NHTSA.
Provided CAFE projections and analyzed CAFE impacts of future product changes.
Team leader of a project to implement Shift Indicator Lights.
November, 1976 to December, 1981 ENGINEER, CHRYSLER POWERTRAIN.
Designed and implemented, from scratch, Chrysler's system to comply with extensive EPA fuel economy
regulations issued in 1975. Also coordinated fuel economy testing, served as liaison with EPA, helped
write responses to proposed regulations, and worked on special projects.
AWARDS and
2015
2010-13
2011-13
2008
2006
2004
2002-2003
2002-2003
2001-2002
1995
1994
1993
1992
1991
ADVISORY COMMITTEES
ECOBEST AWARD from Autobest.org, for uncovering the VW diesel defeat device
NATIONAL RESEARCH COUNCIL - Committee on Transitions to Alternative Vehicles and
Fuels; vehicle subcommittee lead.
NATIONAL COOPERATIVE HIGHWAY RESEARCH PROGRAM - Committee on Effects of
Changing Transportation Energy Supplies and Alternative Fuel Sources on State
Departments of Transportation
NATIONAL RESEARCH COUNCIL - Committee for a Study of Potential Energy Savings
and Greenhouse Gass Reductions from Transportation
SAE ENGINEERING MEETINGS OUTSTANDING ORAL PRESENTATION AWARD, for "It's a
High-MPG Vehicle Issue, Not a Hybrid Issue", SAE Government/Industry Mtg.
BARRY MCNUTT AWARD FOR EXCELLENCE IN AUTOMOTIVE POLICY ANALYSIS
1st recipient of annual award from the SAE
ADVISORY BOARD, Advanced Power Technology Alliance, Center for Automotive
Research, Ann Arbor, Ml
SAE INDUSTRIAL LECTURESHIP PROGRAM, to promote interaction between practicing
engineers and faculty and students via campus visits
SILVER MEDAL, U.S. EPA for strategies to reduce air pollution from nonroad engines
EPA SCIENCE ACHIEVEMENT AWARD in Air Quality. Only person in EPA's Office of
Mobile Sources ever to receive this award.
OUTSTANDING TECHNICAL COMMUNICATION in the 1992-93 Society for Technical
Communication of Southeastern Michigan Technical Publications Competition, for
"Nonroad Engine and Vehicle Emission Study"
BRONZE MEDAL, U.S. EPA for the "Nonroad Engine and Vehicle Emission Study"
BRONZE MEDAL, U.S. EPA for the Cold Temperature Carbon Monoxide Rulemaking
LEADERSHIP TRAINING
2000 Honda Leader's Program - Center for Creative Leadership
1997 Modeling and Computer Simulation of Internal Combustion Engine-U. of Mich, course
1996-7 Excellence in Government Fellows Program-Council for Excellence in Government
A-12
-------
Peer Review Report
Work Assignment 3-05/4-05, Contract 68HE0C18C0001
1995
Diversity Workshops - University of Michigan
1993
Total Quality Management
1992
Looking Glass Workshop: Leadership in Multilevel Organizations - Creative Leadership
1991
Use of Consultative Methods - EPA Institute
1990
Work Group Leadership - Conservation Foundation
1989
Regulation Development in EPA - EPA
1988
Planning Effective Meetings - EPA
1987
Zenger-Miller Supervision program on Behavior Modeling - EPA
1985
Personnel Management for Managers and Supervisors -OPM
1984
Interaction Management - Chrysler Institute
1982
Organizational Leadership and Productivity - Mansare Corp.
1982
Leadership Effectiveness Training - Chrysler Institute
1981
Supervisory Skills Training - Chrysler Institute
EDUCATION
1980-1984
University of Michigan. Completed 34 hours towards M.B.A. GPA: 7.9 (A=8.0)
1970-1975
University of Michigan, B.S., Physics (minor in Math).
Honors: National Merit Finalist, Honors Program, Dean's List
Activities: U. of Michigan Marching Band and Concert Band
PUBLICATIONS
J. German, "How things work: OMEGA modeling case study based on the 2018 Toyota Camry", working
paper published 2018.02.21 by the ICCT. https://theicct.org/publications/how-things-work-
omega-modeling-case-study-based-2018-toyota-camry
J. German, "U.S. fuel economy trends reflect a business strategy, not a technology challenge", posted 19
January 2018. https://theicct.org/blog/staff/us-fuel-economy-trend-reflects-business-strategy-
not-tech-challenge
J. German, "Technology Leapfrog: Or, all recent auto technology forecasts underestimate how fast
innovation is happening", posted 25 September 2017. https://theicct.org/blog/staff/technology-
leapfrogging
A. Isenstadt, J. German, "Diesel Engines", 2017, published by ICCT. http://www.theicct.org/diesel-
engines
A. Isenstadt, J. German (ICCT), M. Dorobantu (Eaton), "Naturally aspirated gasoline engines and cylinder
deactivation". 2017, published by ICCT. http://www.theicct.org/naturally-aspiratedgas-engines-
201606
Meszler, D., German, J., Mock, P., & Bandivadekar, A. (2016). C02 reduction technologies for the
European car and van fleet: A 2025-2030 assessment. International Council on Clean
Transportation, https://theicct.org/publications/co2-reduction-technologies-european-car-and-
van-fleet-2025-2030-assessment
A. Isenstadt, J. German (ICCT), M. Dorobantu (Eaton), D. Boggs (Ricardo), T. Watson (JCI), "Downsized,
boosted gasoline engines", 2016, published by ICCT. http://www.theicct.org/downsized-boosted-
gasoline-engines
A-13
-------
Peer Review Report
Work Assignment 3-05/4-05, Contract 68HE0C18C0001
A. Isenstadt, J. German (ICCT), P. Bubna, M. Wiseman (Ricardo), U. Venkatakrishnan, L. Abbasov (SABIC),
P. Guillen, N. Moroz (Detroit Materials), D. Richman (Aluminum Assoc.), G. Kolwich (FEV),
"Lightweighting technology development and trends in U.S. passenger vehicles", 2016, published
by ICCT. http://www.theicct.org/lightweighting-technology-development-and-trends-us-
passenger-vehicles
S. Osborne, J. Kopinsky, S. Norton (ITB), A. Sutherland, D. Lancaster, E. Nielsen (BorgWarner);
A.lsenstadt, J. German (ICCT), "Automotive thermal management technology", 2016, published by
ICCT. http://www.theicct.org/automotive-thermal-management-technology
John German, "Hybrid vehicles: Trends in technology development and cost reduction", 2015, published
by ICCT. http://www.theicct.org/hybrid-vehicles-trends-technology-development-and-cost-
reduction
V. Franco, F. Posada, J. German, P. Mock, "Real-world exhaust emissions from modern diesel cars",
October 2014, published by ICCT. http://theicct.org/real-world-exhaust-emissions-modern-diesel-
cars
P. Mock, U. Tietge, V. Franco, J. German, A. Bandivadekar (ICCT), N. Ligterink (TNO), U. Lambrecht (IFEU),
J. Kiihlwein (KISU), I. Riemersma (Sidekick Project Support), "From laboratory to road: A 2014
update", September 2014, published by ICCT. http://theicct.org/laboratory-road-2014-update
John German, "U.S. Tier 3 vehicle emissions and fuel quality standards, final rule", March 2014,
published by ICCT. http://theicct.org/us-tier-3-vehicle-emissions-and-fuel-quality-standards-final-
rule
D. Meszler, J. German, P. Mock, A. Bandivadekar, J. Tu, "Summary of Eastern EU labor rate impacts on
EU cost curves", February 2014, published by ICCT. http://theicct.org/summary-eastern-eu-labor-
rate-impacts-eu-cost-curves
S. Searle, F. Posada, C. Malins, J. German, "Technical barriers to the consumption of higher blends of
ethanol", February 2014, published by ICCT. http://theicct.org/technical-barriers-consumption-
higher-blends-ethanol
John German, "Invited Commentary: The Future of U.S. Natural Gas is Utilities, Export, and Trucks, not
Cars", Current Sustainable/Renewable Energy Reports, Springer International Publishing AG 2014.
A. Bandivadekar, T. DeFries, J. German, S. Kishan, F. Posada, M. Sabisch, "In-Use Fuel Economy and C02
Emissions Measurement using OBD Data on US Light-Duty Vehicles", SAE 2014-01-1623, April
2014.
F. Posada and J. German, "Measuring in-use fuel economy: Summary of pilot studies", December 2013,
published by ICCT. http://theicct.org/measuring-in-use-fuel-economy-summary-pilot-studies
A. Bandivadekar, J. German, U. Lambrecht, N. Ligterink, P. Mock, "From Laboratory to Road", May 2013,
published by ICCT. http://theicct.org/laboratory-road
A. Bandivadekar, J. German, D. Meszler, P. Mock, "Initial processing of Ricardo vehicle simulation
modeling C02 data", February 2013, published by ICCT. http://theicct.org/initial-processing-
ricardo-vehicle-simulation-modeling-co2-data-0
A. Bandivadekar, J. German, D. Meszler, P. Mock, "Mass reduction impacts on EU cost curves", February
2013, published by ICCT. http://theicct.org/mass-reduction-impacts-eu-cost-curves
D. Meszler, J. German, P. Mock, and A. Bandivadekar, "EU cost curve development methodology",
November 2012, published by ICCT. http://www.theicct.org/eu-cost-curve-development-
methodology
A-14
-------
Peer Review Report
Work Assignment 3-05/4-05, Contract 68HE0C18C0001
D. Meszler, J. German, P. Mock, A. Bandivadekar, "Initial processing of Ricardo vehicle simulation
modeling C02 data", July 2012, published by ICCT. http://www.theicct.org/initial-processing-
ricardo-vehicle-simulation-modeling-co2-data
F. Posada Sanchez, A. Bandivadekar, and J. German, Estimated Cost of Emission Reduction Technologies
for Light-Duty Vehicles, published by ICCT, June 2012. http://theicct.org/estimated-cost-emission-
reduction-technologies-ldvs
P. Mock, J. German, A. Bandivadekar, and I. Riemersma,"Discrepancies between type-approval and real-
world fuel consumption and C02 values in 2001-2011 European passenger cars", published by
ICCT, April 2012. http://www.theicct.org/fuel-consumption-discrepancies
D. Kodjak, A. Bandivadekar, J. German, and N. Lutsey, "The regulatory engine: How smart policy drives
vehicle innovation", published by ICCT, January 2011. http://www.theicct.org/regulatory-engine
John German, Hybrid Powered Vehicles, SAE Technology Profile T-119, 2nd edition, book published by
Society of Automotive Engineers, Warrendate, Pa., 2011.
J. German and N. Lutsey, "Size or Mass? The Technical Rationale for Selecting Size as an Attribute for
Vehicle Efficiency Standards", July 2010, published by ICCT. http://www.theicct.org/2010/08/size-
or-mass/
J. German and D. Meszler, "Best Practices for Feebate Program Design and Implementation", April 2010,
published by ICCT. http://www.theicct.org/2010/04/feebate-best-practices/
John German, "Leadtime, Customers, and Technology: Technology Opportunities and Limits on the Rate
of Deployment". Reducing Climate Impacts in the Transportation Sector. D. Sperling and J.
Cannon, Springer Press, 2008.
D. Greene, J. German, and M. Delucchi, "Fuel Economy: The Case for Market Failure ". Reducing Climate
Impacts in the Transportation Sector. D. Sperling and J. Cannon, Springer Press, 2008.
J. German, "Reducing Vehicle Emissions Through Cap and Trade Schemes". Driving Climate Change:
Cutting Carbon from Transportation. D. Sperling and J. Cannon, Elseview & Academic Press, 2006.
Hybrid Gasoline-Electric Vehicle Development, edited by John German, SAE PT-117, 2005.
John German, "Hybrid Electric Vehicles", Encyclopedia of Energy, Elsevier & Academic Press, 2004
John German, Hybrid Powered Vehicles, SAE Technology Profile T-119, book published by Society of
Automotive Engineers, Warrendate, Pa., 2003.
John German, "Hybrid Vehicles Go to Market", TR News #213, March-April 2001.
John German, "VMT and Emission Implications of Growth in Light Truck Sales", Air and Waste
Management Association Emission Inventory Conference proceedings, Oct. 1997.
J. Alson, J. German, K. Gold, R. Larson, and M. Wolcott, "Transportation Energy Demand Models: Why
They Underestimate Greenhouse Gas Emissions", Climate Change Analysis Workshop Proceedings,
June 6-7, 1996.
John German, "Off-Cycle Emission and Fuel Efficiency Considerations", Asilomar conference on
Transportation and Energy, 1995.
John German, "Observations Concerning Current Motor Vehicle Emissions", SAE 950812, Feb. 1995.
J. Koupal and J. German, "Real-Time Simulation of Vehicle Emissions Using VEMISS", CRC On-Road
Vehicle Emissions Workshop, April 1995.
S. Sheppard, J. Fieber, J. Cohen, and J. German, "Cold Start Motor Vehicle Emissions Model", Air and
Waste Management Association, Cincinnati, 1994. Paper ID: 94-RP107B.02
A-15
-------
Peer Review Report Work Assignment 3-05/4-05, Contract 68HE0C18C0001
P. Enns, J. German, and J. Markey, "EPA's Survey of In-Use Driving Patterns: Implications for Mobile
Source Emission Inventories", AWMA/CARB Specialty Conference on Emission Inventory,
Pasadena, CA, October, 1993.
A-16
-------
Peer Review Report
Work Assignment 3-05/4-05, Contract 68HE0C18C0001
Jeremy J. Michalek
CURRICULUM VITAE
Engineering and Public Policy • Mechanical Engineering • Carnegie Mellon University
ACADEMIC APPOINTMENTS
Professor, Carnegie Mellon University
Mechanical Engineering (July 2014 - present)
Engineering and Public Policy (July 2014 - present)
Associate Professor, Carnegie Mellon University
Mechanical Engineering (July 2010 - June 2014)
Engineering and Public Policy (July 2011 - June 2014)
Engineering and Public Policy, affiliated (July 2010 - June 2011)
Assistant Professor, Carnegie Mellon University
Mechanical Engineering (July 2005 - June 2010)
Engineering and Public Policy, affiliated (June 2007 - June 2010)
Research Economist
National Bureau of Economic Research (Apr 2018- Mar 2019)
Director, Carnegie Mellon Vehicle Electrification Group (Jan 2012 - present)
Co-Director, Carnegie Mellon Vehicle Electrification Group (Jan 2009-Jan 2012)
http://www.cmu.edu/cit/veg
Director, Carnegie Mellon Design Decisions Laboratory (July 2005 - present)
http://www.cmu.edu/me/ddl
Postdoctoral Research Fellow (Jan 2005 - June 2005)
Department of Mechanical Engineering
University of Michigan, Ann Arbor, Ml
EDUCATION
Ph.D. Mechanical Engineering, University of Michigan, 2005
M.S. Mechanical Engineering, University of Michigan, 2001
B.S. Mechanical Engineering, Minor in Engineering Design, Carnegie Mellon University, 1999
RESEARCH INTERESTS
Energy and transportation: Economic, environmental, social, technical, behavioral, and policy aspects of
vehicle and transportation systems
Vehicle electrification, automation, and sharing
Optimization methods and applications
Consumer choice models Jeremy J. Michalek • Curriculum Vitae
TEACHING INTERESTS
Engineering, economic, and environmental modeling, analysis, and decision-making
A-17
-------
Peer Review Report
Work Assignment 3-05/4-05, Contract 68HE0C18C0001
Mathematical modeling, numerical methods, and optimization
Energy and transportation: Economic, environmental, social, technical, behavioral, and policy aspects of
vehicle and transportation systems
Innovation and critical thinking in design, defining and working with open-ended problems
PUBLICATIONS
Peer-Reviewed Journal Publications
[1] Bruchon, M., I. Azevedo and J.J. Michalek (2021) "Effects of air emission externalities on optimal
ridesourcing fleet electrification and operations," Environmental Science & Technology, online
Feb 18, 2021.
[2] He, Guannan, J. Michalek, S. Kar, Q. Chen, D. Zhang and J. Whitacre (2021) "Utility-scale portable
energy storage systems," Joule v5 n2 p379-392.
[3] Ward, J., J.J. Michalek, C. Samaras, I. Azevedo, A. Henao, C. Rames, T. Wenzel (2021) "The
impact of Uber and Lyft on vehicle ownership, fuel economy & transit across U.S. cities,"
iScience v21 nl pl01933.
[4] Yip, A., K. Whitefoot and J.J. Michalek (2021) "Implications of competitor representation for
profit-maximizing optimal design," in press, ASME Journal of Mechanical Design.
[5] Ward, J., J.J. Michalek and C. Samaras (2021) "The air pollution, greenhouse gas and traffic
externality benefits and costs of shifting private vehicle travel to ridesourcing services," in press,
Environmental Science & Technology.
[6] Ward. J.W., J.J. Michalek, I.L. Azevedo, C. Samaras, P. Ferreira (2019) "Effects of on-demand
ridesourcing on vehicle ownership, fuel consumption, vehicle miles traveled, and emissions per
capita in US states," Transportation Research Part C: Emerging Technologies vl08 p289-301.
[7] Helveston. J.P., S.M. Seki, J. Min, E. Fairman, A.A. Boni, J.J. Michalek, I.M.L. Azevedo (2019)
"Choice at the pump: measuring preferences for lower-carbon combustion fuels,"
Environmental Research Letters, vl4 n8 084035.
[8] Jenn, A., I.L. Azevedo and J.J. Michalek (2019) "Alternative-fuel-vehicle policy interactions
increase U.S. greenhouse gas emissions," Transportation Research Part A: Policy and Practice,
vl24 p397-407.
[9] Ward, J.W., J.J. Michalek, I.L. Azevedo, C. Samaras, P. Ferreira (2019) "Effects of on-demand
ridesourcing on vehicle ownership, fuel consumption, vehicle miles traveled, and emisssions per
capita in U.S. states," Transportation Research Part C: Emerging Technologies, vl08 p289-301.
[10] Helveston, J., E. Feit and J.J. Michalek (2018) "Pooling stated and revealed preferences in the
presence of endogeneity," Transportation Research Part B: Methodological, vl09 p70-89.
[11] Yip, A., J.J. Michalek and K. Whitefoot (2018) "On the implications of using composite vehicles in
choice model prediction," Transportation Research Part B: Methodological, vll6 pl63-188.
[12] Sakti, A., I.M.L. Azevedo, E.R.H. Fuchs, J.J. Michalek, K.G. Gallagher and J.F. Whitacre (2017)
"Consistency and robustness of forecasting for emerging technologies: the case of Li-ion
batteries for electric vehicles," Energy Policy vl06 p415-426.
[13] Yuksel, T., S. Litster, V. Viswanathan, and J.J. Michalek (2017) "Plug-in hybrid electric vehicle
LiFeP04 battery life implications of thermal management, driving conditions, and regional
climate" Journal of Power Sources, v338 nl5 p49-64.
A-18
-------
Peer Review Report
Work Assignment 3-05/4-05, Contract 68HE0C18C0001
[14] Haaf, C.G., W.R. Morrow, I. Azevedo, E. Feit and J.J. Michalek (2016) "Forecasting light-duty
vehicle demand using alternative-specific constants for endogeneity correction vs. calibration,"
Transportation Research Part B: Methodology, v84 pl82-210.
[15] Jenn, A., I.L. Azevedo and J.J. Michalek (2016) "Alternative fuel vehicle adoption increases fleet
gasoline consumption and greenhouse gas emissions under United States corporate average
fuel economy policy and greenhouse gas emissions standards," Environmental Science &
Technology, v50 n5 p.2165-2174.
[16] Weis, A., P. Jaramillo and J.J. Michalek (2016) "Consequential life cycle air emissions externalities
for plug-in electric vehicles in the PJM interconnection," Environmental Research Letters, vll n2
024009.
[17] Yuksel, T., M. Tamayao, C. Hendrickson, I. Azevedo and J.J. Michalek (2016) "Effect of regional
grid mix, driving patterns and climate on the comparative carbon footprint of electric and
gasoline vehicles," Environmental Research Letters, vll n4 044007.
[18] Helveston, J.P., Y. Liu, E. Feit, E. Fuchs, E. Klampfl, and J.J. Michalek (2015) "Will subsidies drive
electric vehicle adoption? Measuring consumer preferences in the U.S. and China,"
Transportation Research Part A: Policy and Practice, v73 p96-112.
[19] Sakti, A., J.J. Michalek, E.R.H. Fuchs, and J.F. Whitacre (2015) "Atechno-economic analysis and
optimization of Li-ion batteries for light-duty passenger vehicle electrification," Journal of Power
Sources v273 p966-980.
[20] Sylcott, B., J.J. Michalek, and J. Cagan (2015) "Exploring the role of interaction effects in visual
conjoint analysis," ASME Journal of Mechanical Design, vl37 n9 p094503 1-5.
[21] Tamayao, M., J.J. Michalek, C. Hendrickson and I. Azevedo (2015) "Regional variability and
uncertainty of electric vehicle life cycle C02 emissions across the United States," Environmental
Science & Technology, v49 nl4 p8844-8855.
[22] Weis, A., J.J Michalek, P. Jaramillo and R. Lueken (2015) "Emissions and cost implications of
controlled electric vehicle charging in the US PJM interconnection," Environmental Science &
Technology, v49 n9 p5813-5819.
[23] Yuksel, T. and J.J. Michalek (2015) "Effects of regional temperature on electric vehicle efficiency,
range, and emissions in the United States," Environmental Science and Technology, v49 n6
p3974-3980.
[24] Griffin, W.M., J.J. Michalek, H.S. Matthews and M.N.A. Hassan (2014) "Availability of biomass
residues for co-firing in peninsular Malaysia: implications for cost and GHG emissions in the
electricity sector," v7 n2 p804-823.
[25] Haaf, C.G., J.J. Michalek, W.R. Morrow, and Y. Liu (2014) "Sensitivity of vehicle market share
predictions to discrete choice model specification," ASME Journal of Mechanical Design vl36
121402 pl-9.
[26] Khajavirad, A., J.J. Michalek and N.V. Sahinidis (2014) "Relaxations of factorable functions with
convex-transformable intermediates," Mathematical Programming, DOI 10.1007/sl0107-012-
0618-8, pl-34.
[27] Min, J., I. Azevedo, J.J. Michalek and W. Bruine de Bruin (2014) "Labeling energy cost on light
bulbs lowers implicit discount rates," Ecological Economics v97 p42-50.
[28] Weis, A., P. Jaramillo and J.J. Michalek (2014) "Estimating the potential of controlled electric
vehicle charging to reduce operational and capacity expansion costs for electric power systems
with high wind penetration," Applied Energy vll5 pl90-204.
A-19
-------
Peer Review Report
Work Assignment 3-05/4-05, Contract 68HE0C18C0001
[29] Traut, E., T.W. Cherng, C. Hendrickson, and J.J. Michalek (2013) "US residential charging
potential for electric vehicles/' Transportation Research Part D: Transport and Environment v25
pl39-145.
[30] Karabasoglu, O. and J.J. Michalek (2013) "Influence of driving patterns on life cycle cost and
emissions of hybrid and plug-in electric vehicle powertrains/' Energy Policy, v60 p445-461.
[31] Peterson, S. and J.J. Michalek (2013) "Cost effectiveness of plug-in hybrid electric vehicle battery
capacity and charging infrastructure investment for reducing US gasoline consumption/' Energy
Policy, v52 p429-438.
[32] Sakti, A., J.J. Michalek, S-E Chun and J.F. Whitacre (2013) "A validation study of lithium-ion cell
constant C-rate discharge simulation with Battery Design Studio©," International Journal of
Energy Research v37 nl2 pl562-1568.
[33] Resende, C.B., C.G. Heckmann and J.J. Michalek (2012) "Robust design for profit maximization
with aversion to downside risk from parametric uncertainty in consumer choice models," ASME
Journal of Mechanical Design, vl34 100901 pl-12.
[34] Traut, E., C. Hendrickson, E. Klampfl, Y. Liu and J.J. Michalek (2012) "Optimal design and
allocation of electrified vehicles and dedicated charging infrastructure for minimum life cycle
greenhouse gas emissions and cost," Energy Policy, v51 p524-534.
[35] Michalek, J.J., M. Chester, P. Jaramillo, C. Samaras, C.S. Shiau, and L. Lave (2011) "Valuation of
plug-in vehicle life cycle air emissions and oil displacement benefits" Proceedings of the National
Academy of Sciences, vl08 n40 pl6554-16558.
[36] Michalek, J.J., F.M. Feinberg, P. Ebbes, F. Adiguzel and P.Y. Papalambros (2011) "Enhancing
marketing with engineering: optimal product line design for heterogeneous markets,"
International Journal of Research in Marketing, v28 pl-12. [Best Article Award]
[37] Shiau, C.-S. and J.J. Michalek (2011) "Global optimization of plug-in hybrid vehicle design and
allocation to minimize life cycle greenhouse gas emissions," ASME Journal of Mechanical Design,
vl33 n8 p084502 1-6.
[38] Shiau, C-S, N. Kaushal, C.T. Hendrickson, S. Peterson, J. Whitacre, and J.J. Michalek (2010)
"Optimal plug-in hybrid electric vehicle design and allocation for minimum life cycle cost,
petroleum consumption, and greenhouse gas emissions," ASME Journal of Mechanical Design,
Special Issue on Sustainability, vl32 n9 p091013 1-11.
[39] Khajavirad, A and J.J. Michalek (2009) "A determinstic Lagrangian-based global optimization
approach for quasiseparable nonconvex mixed-integer nonlinear programs," ASME Journal of
Mechanical Design, vl31 p051009 1-8
[40] Khajavirad, A., J.J. Michalek and T.W. Simpson (2009) "An efficient decomposed multi-objective
genetic algorithm for solving the joint product family selection and design problem with
generalized commonality," Structural and Multidisciplinary Optimization, v39 pl87-201.
[41] Shiau, C.-S. and J.J. Michalek (2009) "Optimal product design under price competition," ASME
Journal of Mechanical Design, vl31 071003 pl-10.
[42] Shiau, C.-S., J.J. Michalek, and C.T. Hendrickson (2009) "A structural analysis of vehicle design
responses to corporate average fuel economy policy," Transportation Research Part A: Policy
and Practice, v43 p814-828.
[43] Shiau, C.-S., C. Samaras, R. Hauffe and J.J. Michalek (2009) "Impact of battery weight and
charging patterns on the economic and environmental benefits of plug-in hybrid vehicles,"
Energy Policy v37 p2653-2663.
A-20
-------
Peer Review Report
Work Assignment 3-05/4-05, Contract 68HE0C18C0001
[44] Shiau, C.-S. and J.J. Michalek (2009) "Should designers worry about market systems?" ASME
Journal of Mechanical Design, vl31 011011 pl-9.
[45] Khajavirad, A. and J.J. Michalek (2008) "A decomposed approach for solving the joint product
family selection and design problem with generalized commonality/' ASME Journal of
Mechanical Design, vl30 p071101.
[46] Li, Y., Z. Lu and J.J. Michalek (2008) "Diagonal quadratic approximation for parallelization of
analytical target cascading." ASME Journal of Mechanical Design vl30 n5 p051402-l-ll.
[47] Michalek, J.J., O. Ceryan, P.Y. Papalambros, and Y. Koren (2006) "Balancing marketing and
manufacturing objectives in product line design," ASME Journal of Mechanical Design, vl28 n6
pll96-1204.
[48] Michalek, J.J., F.M. Feinberg and P.Y. Papalambros (2005) "Linking marketing and engineering
product design decisions via analytical target cascading," Journal of Product Innovation
Management, v22 p42-62.
[49] Michalek, J.J. and P.Y. Papalambros (2005) "An efficient weighting update method to achieve
acceptable consistency deviation in analytical target cascading," ASME Journal of Mechanical
Design, vl27 p206-214.
[50] Michalek, J.J. and P.Y. Papalambros (2005) "Technical brief: weights, norms, and notation in
analytical target cascading," ASME Journal of Mechanical Design, vl27 p499-501.
[51] Michalek, J.J., P.Y. Papalambros, and S.J. Skerlos (2004) "A study of fuel efficiency and emission
policy impact on optimal vehicle design decisions," ASME Journal of Mechanical Design, vl26
pl062-1070.
[52] Michalek, J.J., R. Choudhary and P.Y. Papalambros (2002) "Architectural layout design
optimization," Engineering Optimization, v34 n5 pp461-484.
[53] Michalek, J.J. and P.Y. Papalambros (2002) "Interactive design optimization of architectural
layouts," Engineering Optimization, v34 n5 pp485-501.
Full-length Peer-Reviewed Conference Publications
[54] Yip, A.H.C., J.J Michalek and K.S. Whitefoot (2019) "Implications of competitor representation on
optimal design," ASME International Design Engineering Technical Conferences, Aug 2019,
Anaheim, CA.
[55] Heckmann, C.G., J.J. Michalek, W.R. Morrow, and Y. Liu (2013) "Sensitivity of vehicle market
share predictions to alternative discrete choice model specifications," ASME International
Design Engineering Technical Conferences, August 2013, Portland, OR.
[56] Sylcott, B, J.J. Michalek and J. Cagan (2013) "Understanding the role of interaction effects in
visual conjoint analysis," ASME International Design Engineering Technical Conferences, August
2013, Portland, OR.
[57] Yuksel, T. and J.J. Michalek (2012) "Development of a simulation model to analyze the effect of
thermal management on battery life," Society of Automotive Engineers World Congress, April
24-26, Detroit, Ml.
[58] Yuksel, T. and J.J. Michalek (2012) "Evaluation of the effects of thermal management on battery
life in plug-in hybrid electric vehicles," The Battery Congress, April 23-24, Ann Arbor, Ml.
[59] Michalek, J.J., C.T. Hendrickson and J. Cagan (2011) "Using economic input-output life cycle
assessment to guide sustainable design," ASME International Design Engineering Technical
Conferences, August 28-31, Washington DC.
A-21
-------
Peer Review Report
Work Assignment 3-05/4-05, Contract 68HE0C18C0001
[60] Resende, C., C.G. Heckmann and J.J. Michalek (2011) "Robust design for profit maximization
under uncertainty of consumer choice model parameters using the delta method/' ASME
International Design Engineering Technical Conferences, August 28-31, Washington DC.
[61] Sakti, A., S. Khan, T. Langer, J.J. Michalek, and J.F. Whitacre (2011) "Techno-economic analysis of
Lithium-ion batteries for personal vehicle electrification," National Academies Transportation
Research Board Annual Meeting, January 23-27, Washington D.C.
[62] Traut, E., C.T. Hendrickson, E. Klampfl, Y. Liu and J.J. Michalek (2011) "Optimal Design and
Allocation of Electrified Vehicles and Dedicated Charging Infrastructure for Minimum
Greenhouse Gas Emissions," National Academies Transportation Research Board Annual
Meeting, January 23-27, Washington D.C.
[63] Shiau, C.-S., S. Peterson and J.J. Michalek (2010) "Optimal plug-in hybrid electric vehicle design
and allocation for minimum life cycle cost, petroleum consumption and greenhouse gas
emissions," Proceedings of the ASME International Design Engineering Technical Conferences,
Advance Vehicle and Tire Technologies Conference, August 15-18, Montreal, Quebec, Canada.
[64] Shiau, C.-S. and J.J. Michalek (2010) "A mixed-integer nonlinear programming model for
deterministic global optimization of plug-in hybrid vehicle design and allocation," Proceedings of
the ASME International Design Engineering Technical Conferences, Design Automation
Conference, August 15-18, Montreal, Quebec, Canada.
[65] Kaushal, N., C.-S. Shiau and J.J. Michalek (2009) "Optimal plug-in hybrid electric vehicle design
and allocation for diverse charging patterns," Proceedings of the International Design
Engineering Technical Conferences, Aug 30-Sept 2, San Diego, CA, USA.
[66] Shiau, C.-S., J.J. Michalek and C. T. Hendrickson (2009) "A structural analysis of vehicle design
responses to corporate average fuel economy standards," Transportation Research Board
Annual Meeting, Washington D.C.
[67] Shiau, C.-S., C. Samaras, R. Hauffe and J.J. Michalek (2009) "Impact of battery weight and
charging patterns on the economic and environmental benefits of plug-in hybrid vehicles,"
Transportation Research Board Annual Meeting, Washington D.C.
[68] Hauffe, R., C. Samaras and J.J. Michalek (2008) "Plug-in hybrid vehicle simulation: How battery
weight and charging patterns impact cost, fuel consumption, and C02 emissions," Proceedings
of the ASME International Design Engineering Technical Conferences, August 3-6, Brooklyn, NY,
USA.
[69] Khajavirad, A. and J.J. Michalek (2008) "A deterministic Lagrangian-based global optimization
approach for large scale decomposable problems," Proceedings of the ASME International
Design Engineering Technical Conferences, August 3-6, Brooklyn, NY, USA.
[70] McGaughey, A. and J. J. Michalek (2008) "Wiki-based learning in the mechanical engineering
classroom," American Society for Engineering Education Annual Conference, June 22-25,
Pittsburgh, PA, USA.
[71] Shiau, C.-S. and J.J. Michalek (2008) "Should designers worry about market systems?"
Proceedings of the ASME International Design Engineering Technical Conferences, August 3-6,
Brooklyn, NY, USA.
[72] Shiau, C.S. and J.J. Michalek (2008) "Optimal product design under price competition,"
Proceedings of the ASME International Design Engineering Technical Conferences, August 3-6,
Brooklyn, NY, USA.
A-22
-------
Peer Review Report Work Assignment 3-05/4-05, Contract 68HE0C18C0001
[73] Li, Y., Z. Lu and J.J. Michalek (2007) "Diagonal quadratic approximation for parallelization of
analytical target cascading." Proceedings of the 2007 ASME International Design Engineering
Technical Conferences, Sept 4-6, Las Vegas, NV, USA.
[74] Khajavirad, A. and J.J. Michalek (2007) "A single-stage gradient-based approach for solving the
joint product family platform selection and design problem using decomposition," Proceedings
of the 2007 ASME International Design Engineering Technical Conferences, Sept 4-6, Las Vegas,
NV, USA.
[75] Khajavirad, A. and J.J. Michalek (2007) "An extension of the commonality index for product
family optimization," Proceedings of the 2007 ASME International Design Engineering Technical
Conferences, Sept 4-6, Las Vegas, NV, USA.
[76] Orsborn, S., S. Swamy, J. Michalek and J. Cagan (2007) "Measurement of headlight form
preference using choice-based conjoint analysis," Proceedings of the 2007 ASME International
Design Engineering Conferences, Sept 4-6, Las Vegas, NV, USA.
[77] Shiau, C.S., I.H. Tseng, D. Heutchy and J. Michalek (2007) "Design optimization of a laptop
computer using aggregate and mixed logit demand models with consumer survey data."
Proceedings of the 2007 ASME International Design Engineering Conferences, Sept 4-6, Las
Vegas, NV, USA.
[78] Shiau, C.S. and J. Michalek (2007) "A game theoretic approach for finding market equilibria for
automotive design under environmental regulation," Proceedings of the 2007 ASME
International Design Engineering Technical Conferences, Sept 4-6, Las Vegas, NV, USA.
[79] Michalek, J.J. and P.Y. Papalambros (2006) "BB-ATC: Analytical target cascading using branch
and bound for mixed integer nonlinear programming," Proceedings of the 2006 ASME
International Design Engineering Technical Conferences, Sept. 10-13, Philadelphia, PA, USA.
[80] Michalek, J.J., O. Ceryan, P.Y. Papalambros, and Y. Koren (2005) "Manufacturing investment and
allocation in product line design decision-making," Proceedings of the 2005 ASME International
Design Engineering Technical Conferences, DETC2005-84812, Sept. 26-28, Long Beach, CA, USA.
[Best Paper Award]
[81] Choudhary, R. and J. Michalek (2005) "Design optimization in computer aided architectural
design," International Conference of the Association for Computer Aided Architectural Design
Research In Asia, April 28-30, 2005, New Delhi, India.
[82] Michalek, J.J., F.M. Feinberg and P.Y. Papalambros (2004) "An optimal marketing and
engineering design model for product development using analytical target cascading,"
Proceedings of the 2004 Tools and Methods for Competitive Engineering Conference, Lausanne,
Switzerland, April 12-16, 2004.
[83] Michalek, J.J. and P.Y. Papalambros (2004) "An efficient weighting update method to achieve
acceptable consistency deviation in analytical target cascading," Proceedings of the 2004 ASME
Design Engineering Technical Conferences, DETC2004-57134, Sept. 28 - Oct. 2, Salt Lake City,
Utah, USA.
[84] Michalek, J.J., P.Y. Papalambros and S.J. Skerlos (2003) "A study of emission policy effects on
optimal vehicle design decisions," Proceedings of the ASME Design Engineering Technical
Conferences, DETC2003/DAC-48767, September 2-6, Chicago, IL, USA.
A-23
-------
Peer Review Report
Work Assignment 3-05/4-05, Contract 68HE0C18C0001
Other Conference Publications
[85] Khajavirad, A., J.J. Michalek and T.W. Simpson (2007) "An efficient decomposed genetic
algorithm for joint product family selection and optimization," Proceedings of the 2007 AIAA
Multidisciplinary Design Optimization Specialists Conference, April 23-26, Honolulu, Hawaii,
USA.
[86] Cooper, A.B., P. Geortiopoulos, J.J. Michalek, and P.Y. Papalambros (2004) "A simulation-based
vehicle design strategy for acquisition and requirements validation," Proceedings of the 2004
Society of Automotive Engineers World Congress, March 11-18, Detroit, Ml. A shortened version
also appeared at the US-ASMO SMART Conference, Dearborn, Michigan, Sept. 8-11, 2003.
Book Chapters
[87] Khajavirad, A., J.J. Michalek and T.W. Simpson (2014) "Solving the joint product platform
selection and product family design problem: An efficient decomposed multiobjective genetic
algorithm with generalized commonality," in Advances in Product Family and Product Platform
Design: Methods & Applications, Eds: T. Simpson, R. Jiao, Z. Siddique, K. Holtta-Otto, Springer,
New York.
[88] Michalek, J.J. (2007) "Designing better products by coordinating marketing research and
engineering," in Modern Marketing Research: Concepts, Methods and Cases, eds: F. Feinberg, T.
Kinnear and J. Taylor.
[89] Skerlos, S.J., W.R. Morrow, and J.J. Michalek (2006) "Sustainable design engineering and science:
selected challenges and case studies," chapter 3.10 in Sustainability Science and Engineering,
Volume 1: Defining Principles, edited by M. Abraham, Elsevier Science.
Magazine Publications:
[90] Michalek, J.J. and V.J. Karplus (2021) "Choosing electric vehicle policies with care," Issues in
Science and Technology, 37, n3 (Spring 2021).
[91] Michalek, J.J., M. Chester and C. Samaras (2012) "Getting the most out of electric vehicle
subsidies," Issues in Science and Technology, Summer 2012, p25-27.
[92] Michalek, J.J. (2008) "Design for market systems: Integrating social, economic, and physical
sciences to engineer product success," Mechanical Engineering, v 130, nil, p32-36.
Policy Briefs:
[93] Michalek, J.J. (2017) Electric Vehicle Benefits and Costs in the United States, policy brief,
Carnegie Mellon University, May 2017. Presented at the U.S. House of Representatives, National
Governors Association, U.S. Department of Transportation and Office of U.S. Senator Toomey.
[94] Michalek, J.J. (2017) Electric Vehicle Adoption Potential in the United States, policy brief,
Carnegie Mellon University, May 2017. Presented at the U.S. House of Representatives, National
Governors Association, U.S. Department of Transportation and Office of U.S. Senator Toomey.
[95] Michalek, J.J. (2015) Electric Vehicle Benefits and Costs in the United States, policy brief,
Carnegie Mellon University, May 2015. Presented to members of the U.S. Environmental
Protection Agency, California Air Resources Board, California Energy Commission, California
State Senate and Assembly, and several national labs and universities, 2015.
[96] Michalek, J.J. (2015) Electric Vehicle Adoption Potential in the United States, policy brief,
Carnegie Mellon University, May 2015. Presented to members of the U.S. Environmental
Protection Agency, California Air Resources Board, California Energy Commission, California
State Senate and Assembly, and several national labs and universities, 2015.
A-24
-------
Peer Review Report
Work Assignment 3-05/4-05, Contract 68HE0C18C0001
[97] Michalek, J.J., M. Chester, P. Jaramillo, and C. Samaras (2011) Air Emissions and Oil
Displacement Benefits from Plug-in Vehicles, policy brief, Carnegie Mellon University,
September 2011. Presented to members of the Congressional Budget Office, the Congressional
Research Service, the Senate Energy and Natural Resources Committee, the Senate Commerce,
Science and Transportation Committee, and members of the U.S. House of Representatives,
March 13-14, 2012.
[98] Michalek, J.J. and C. Samaras (2009) Economic, Environmental, and Security Implications of Plug-
in Hybrid Electric Vehicles, policy brief, Carnegie Mellon University, April 2009. Presented to
members of the House Energy and Commerce Committee, the House Committee on Science and
Technology, the Select Committee on Energy Independence and Global Warming, the
Congressional Research Service, and members of the U.S. House and Senate, April 16-20, 2009.
Commentary and Testimony
[99] Bruchon, M. and J. Michalek, "Comment on California Air Resources Board's Proposed Clean
Miles Standard," Public Comment, May 17, 2021.
[100] Michalek, J. and K. Whitefoot, "Comment on the Notice of Proposed Rulemaking for the 'Safer
Affordable Fuel-Efficient (SAFE) Vehicles Rule for Model Years 2021-2026 Passenger Cars and
Light Trucks'" Public Comment, Docket No. NHTSA-2018-0067 and EPA-HQ-OAR-2018-0283,
October 26, 2018, U.S. Federal Register.
[101] Michalek, J., "Statement for Hearing on the Notice of Proposed Rulemaking for the 'Safer
Affordable Fuel-Efficient (SAFE) Vehicles Rule for Model Years 2021-2026 Passenger Cars and
Light Trucks,"' Sept 26, 2018.
[102] Michalek, J. "Problems with the fuel economy rollback," op-ed, The Hill, Sept 2018.
[103] Adams, P., J. Apt, I. Azevedo, E. Casman, J. Cohon, A. Davis, N. Donahue, D. Dzombak, W. Griffin,
P. Jaramillo, T. Krishnamurti, M. Mauter, J. Michalek, M. Morgan, N. Muller, I. Nair, J. Peha, E.
Rubin, M. Sirbu, M. Small, E. Subrahamanian, P. Vaishnav, J. Whitacre, K. Whitefoot, H. Zhai,
"Comment submitted with respect to the EPA Proposed Rule on 'Strengthening Transparency in
Regulatory Science' 40 CFR Part 30 [EPA-HQ-OA-2018-0259; FRL-9977-40-ORD] RIN 2080-AA14."
May 11, 2018.
[104] Whitefoot, K., J.J. Michlalek and I. Azevedo (2017) "Comment on [Docket No. HNTSA-2017-0059]
Civil Penalties Rate for Violations of Corporate Average Fuel Economy Standards," U.S. Federal
Register.
[105] Whitefoot, K., J.J. Michalek, and I. Azevedo (2017) "Comment on Docket No.: EPA-HQ-OAR-
2015-0827 and NHTSA-2016-0068 Reconsideration of the Final Determination of the Mid-Term
Evaluation of Greenhouse Gas Emissions Standards for Model Year 2022-2025 Light-Duty
Vehicles; and Model Year 2021 Greenhouse Gas Emissions Standards," U.S. Federal Register.
[106] Whitefoot, K., J.J. Michalek, and I. Azevedo (2016) "Comment on [Docket No.: EPA-HQ-OAR-
2015-0827] Proposed Determination on the Appropriateness of the Model Year 2022-2025
Light-Duty Vehicle Greenhouse Gas Emissions Standards under the Midterm Evaluation"
Dissertations and Theses:
[107] Michalek, J.J., (2005) Preference Coordination in Engineering Design Decision-Making, Ph.D.
Dissertation, Department of Mechanical Engineering, University of Michigan. Committee: Panos
Papalambros*, Fred Feinberg, Steven Skerlos, Richard Gonzalez.
A-25
-------
Peer Review Report
Work Assignment 3-05/4-05, Contract 68HE0C18C0001
[108] Michalek, J.J. (2001) Interactive Layout Design Optimization, M.S. Thesis, Department of
Mechanical Engineering, University of Michigan, Ann Arbor, Michigan, USA, 2001. Committee:
Panos Papalambros*, Kazuhiro Saitou.
ADVISING
Ph.D. Research
Shiau, Ching-Shin Norman - Design Decision Making for Market Systems and Environmental Policy
with Vehicle Design Applications, Ph.D. Mechanical Engineering, (2006-2010). Present position:
Design Engineering Staff Engineer, Seagate Technologies, Taiwan.
Khajavirad, Aida - Convexification Techniques for Global Optimization of Nonconvex Nonlinear
Optimization Problems, Ph.D. Mechanical Engineering (2006-2011), co-advised with Nick Sahinidis.
Present position: Assistant Professor, Operations Research and Industrial Engineering, Department
of Mechanical Engineering, University of Texas at Austin.
Karabajoglu, Orkun - Influence of Driving Patterns and Optimal Robust Powertrain Combined
Design and Control of Plug-in Vehicle Cost, Life Cycle Emissions, Component Sizing, and Battery
Stress, Ph.D. Mechanical Engineering (2008-2013). Present position: Assistant Professor, Electrical
and Computer Engineering, SYSU-CMU Joint Institute of Engineering, Visiting Professor, Electrical
and Computer Engineering, Carnegie Mellon University.
Traut, Elizabeth - Life Cycle Cost and Environemtnal Implications of U.S. Electric Vehicle and
Charging Infrastructure Scenarios, Ph.D. Mechanical Engineering (2008-2013), co-advised with Prof.
Chris Hendrickson. Present position: Postdoctoral Research Fellow, Civil and Environmental
Engineering, Carnegie Mellon University.
Resende, Camilo- Ph.D. student, Mechanical Engineering (2009-2010), left program for medical
school.
Sakti, Apurba - Quantification of Performance and Cost Trajectory of Li-ion Battery Designs for
Personal Vehicle Electrification in the Near Future, Ph.D. Engineering & Public Policy (2010-2013), co-
advised with Prof. Jay Whitacre. Present Position: Postdoctoral Associate, Energy Initiative and
Chemical Engineering, Massachusetts Institute of Technology.
Haaf, Christine Grace Heckmann - Vehicle Demand Forecasting with Discrete Choice Models, Ph.D.
Mechanical Engineering (2010-2014). Present Position: Visiting Researcher, King Abdullah Petroleum
Studies and Research Center (KAPSARC).
Jenn, Alan - Advanced and Alternative Fuel Vehicle Policies: Regulations and Incentives in the United
States, Ph.D. Engineering and Public Policy (2014), officially advised by Prof. Ines Azevedo but I co-
advised several studies. Present Position: Postdoctoral Researcher, University of California, Davis.
Min, Jihoon - Energy Efficient Lighting: Consumer Preferences, Choices, and System Wide Effects,
Ph.D. Engineering and Public Policy (2010-2014), co-advised with Prof. Ines Azevedo. Present
Position: Research Scholar, International Institute for Applied Systems Analysis.
Tamayao, Mili-Ann - Urbanization and Vehicle Electrification in the U.S.: C02 Emissions Estimation
and Implications for Climate Policy, Ph.D. Engineering and Public Policy (2012-2014), co-advised with
Prof. Chris Hendrickson. Present Position: Assistant Professor, Industrial Engineering and Operations
Research, and Deputy Executive Director, National Engineering Center, University of the Philippines.
Weis, Allison - Electric Vehicles and the Grid: Interactions and Environmental and Health Impacts,
Ph.D. Engineering and Public Policy (2010-2014), co-advised with Prof. Paulina Jaramillo. Present
Position: Senior Stationary Storage Modeling Engineer, Tesla Motors.
A-26
-------
Peer Review Report
Work Assignment 3-05/4-05, Contract 68HE0C18C0001
Yuksel, Tugfe - Quantification of Temperature Implications and Investigation of Battery Design
Options for Electrified Vehicles, Ph.D. Mechanical Engineering (2010-2015). Present Position: Sabanci
University.
Helveston, John - Development and Adoption of Plug-in Electric Vehicles in China: Markets, Policy,
and Innovation, Ph.D. Engineering and Public Policy (2011-2016), co-advised with Prof. Erica Fuchs.
Present Position: postdoctoral researcher, Boston University.
Yip, Arthur - Integrated Modelling of Consumer Choice, Producer Decisions, and Policy Design in the
Automotive Market, Ph.D., Engineering and Public Policy (2015 - 2020), co-advised with Prof. Kate
Whitefoot.
Ward, Jacob - The Energy and Environmental Effects of New and Future Mobility: Econometric and
Simulation Analysis of Ridesourcing Services Uber and Lyft, Ph.D., Engineering and Public Policy
(2016 - 2020), co-advised with Prof. Ines Azevedo and Prof. Costa Samaras.
Bruchon, Matthew - Ph.D. candidate, Engineering and Public Policy (2017 - pres), co-advised with
Prof. Ines Azevedo.
Forsythe, Connor - PhD. student, Mechanical Engineering (2018- pres), co-advised with Prof. Kate
Whitefoot.
Burns, Aaron - Ph.D. student, Engineering and Public Policy (2020 - pres), co-advised with Prof.
Constantine Samaras.
Koling, Adam - PhD. student, Engineering and Public Policy (2021 - pres), co-advised with Prof.
Daniel Armanios.
Vicente, Johnathan - Ph.D. student, Mechanical Engineering (2021 - pres), co-advised with Prof.
Kate Whitefoot.
M.S. Research Project-Based Degree
Li, Yanjing - Diagonal Quadratic Approximation for Parallel Computing with Analytical Target
Cascading, M.S. Mathematical Sciences, 2006
Swamy, Surya - Global Optimization of Mixed-Integer Nonlinear Systems using Decomposition and
Lagrangian Branch-and-Cut, M.S. Mechanical Engineering, 2007
Kaushal, Nikhil - Optimal Plug-in Hybrid Electric Vehicle Design and Allocation for Diverse Charging
Patterns, M.S. Mechanical Engineering, 2009
Krishnakumar, Varun - Global Design Optimization using Black-Box Optimization Techniques, M.S.,
Mechanical Engineering, 2009
Tsu-Wei Cherng- Charging Infrastructure for Electric Vehicles, Mechanical Engineering, 2011-2012
Gao, Nan - Battery Degradation and Thermal Management for Plug-in Vehicles, M.S. Mechanical
Engineering, 2013-2014
Jennings, Stephanie - Forecasting Technology Purchase Choice Using Composites, M.S. Mechanical
Engineering, 2016-2018, co-advised with Kate Whitefoot.
Chen, Zihao (Lance) - M.S. Engineering and Public Policy, 2020-pres.
M.S. Independent Research Project
Hsieh, Sandy-Review of Vehicle Simulation Software -M.S. Mechanical Engineering, 2006
Chen, Esther - Review of Multiattribute Utility Theory - M.S. Mechanical Engineering, 2008
Yuksel, Burak - Optimal Design and Control of Plug-in Vehicles - Independent Study, Istanbul
Technical University.
A-27
-------
Peer Review Report
Work Assignment 3-05/4-05, Contract 68HE0C18C0001
Yoon, Hyungchul Paul - Plug-in Vehicle Optimization for Real-World Driving, Mechanical
Engineering, 2011
Li, Guo - Optimal Design and Control of Plug-in Vehicles, Mechanical Engineering, 2012-2013
Pan, Lu - Impact of Terrain on Life Cycle Implications of Conventional and Electrified Vehicles,
Mechanical Engineering, 2012-2013
Mehta, Darshit - Optimization of Electric Vehicle Battery and Thermal Management Systems,
Mechanical Engineering, 2015.
Undergraduate Research
Gitomer, Ali - Review of Automotive Demand Models - Visiting student, Northwestern University
Hamilton, Andrew - Study of Consumer Preferences for Hybrid Vehicles - B.S. Mechanical
Engineering, 2008
Hauffe, Richard - Effects of Battery Weight on Plug-in Hybrid Vehicle Performance - B.S. Mechanical
Engineering, 2008
Lee, Jonghyun - Review of Environmental Policy for Vehicle Design - B.S. Mechanical Engineering,
2008
Lewis, Anne Marie - Optimization of Infrastructure for Ethanol Distribution - B.S. Mechanical
Engineering, 2008
Mikkilineni, Sarat- Uncertainty in Choice Modeling - B.S. Mechanical Engineering, 2008
Khwaja, Osman - Effects of Terrain on Vehicle Efficiency of Electrified Powertrains, Independent
Study, Princeton University, 2011.
Kimball, Paul - Control Systems for Electrified Vehicles, 2011-2012
Stabile, Rebecca - Synthetic Drive Cycle Generation and Fuel Economy Estimation, 2011-2012.
Finlayson, Andrew - Synthetic Drive Cycles and Plug-in Vehicle Simulation, 2012-2013.
Ph.D. Committee Service
Olson, Jesse -The Collective Potential: Achieving Organizational Potential by Design, Ph.D.
Mechanical Engineering, Carnegie Mellon University, 2005 [chair: Jonathan Cagan],
Orsborn, Seth -Quantifying aesthetic preference through statistics applied to an agent-based shape
grammar implementation, Ph.D. Mechanical Engineering, Carnegie Mellon University, 2007 [chair:
Jonathan Cagan],
Wakeley, Heather-Alternative Transportation Fuels: Infrastructure Requirements and
Environmental Impacts for Ethanol and Hydrogen, Ph.D. Civil and Environmental Engineering,
Carnegie Mellon University, 2007 [chair: Chris Hendrickson]
Logue, Jennifer - Characterizing Air Toxics Exposure and Risk in Allegheny County and Evaluating
EPA Modeling Tools for Policy Making, Ph.D. Engineering and Public Policy, Mechanical Engineering,
Carnegie Mellon University, 2009 [chair: Allen Robinson],
You, Fengqi - Mixed Integer Nonlinear Programming Models and Algorithms for Enterprise-Wide
Supply Chain Optimization under Uncertainty, Ph.D. Chemical Engineering, Carnegie Mellon
University, 2009 [chair: Ignacio Grossmann],
Hannah, Lindsay- Combinatory Adaptive Optimization with Multi-Agent Systems Ph.D., Mechanical
Engineering, Carnegie Mellon University, 2009 [chair: Jonathan Cagan],
Hatton, Ross - Geometric Mechanics of Locomotion and Optimal Coordinate Choice, Ph.D., Carnegie
Mellon University, Mechanical Engineering, 2011 [chair: Howie Choset],
A-28
-------
Peer Review Report
Work Assignment 3-05/4-05, Contract 68HE0C18C0001
Hassan, Mohd Nor Azman - GHG Emissions and Costs of Developing Biomass Energy in Malaysia:
Implications for Energy Security in the Transportation and Electricity Sectors, Ph.D. Carnegie Mellon
University, Engineering and Public Policy, 2012 [chair: W. Michael Griffin],
Nadadur, Gopal - Anthropometry-based Sustainable Design for Multiple Global Populations, Ph.D.,
The Pennsylvania State University Department of Mechanical and Nuclear Engineering, 2012 [chair:
Matthew Parkinson],
Peterson, Scott - Plug-in Hybrid Electric Vehicles: Battery Degradation, Grid Support, Emissions, and
Battery Size Tradeoffs, Ph.D., Carnegie Mellon University, Engineering and Public Policy, 2012 [chair:
Jay Apt],
Venkatesh, Aranya - Towards Robust Energy Systems Modeling: Examining Uncertainty in Fossil
Fuel-Based Life Cycle Assessment Approaches, Ph.D. Carnegie Mellon University, Civil and
Environmental Engineering 2012 [chair: Chris Hendrickson],
Kamath, Ravindra - Strategies for Optimization and Heat Integration in Integrated Gasification
Combined Cycle Systems, Ph.D. Chemical Engineering, Carnegie Mellon University, 2012 [chair: Larry
Biegler],
Hess, Kacy - In Situ, Through-Thickness Potential and Current Distribution Measurements in
Electrochemical Energy Conversion and Storage Devices, Ph.D. Mechanical Engineering, Carnegie
Mellon University, 2013 [chair: Shawn Litster],
Sylcott, Brian - Understanding How Consumers Process Product Form and Function and How They
Balance Their Preferences for Each, Ph.D. Mechanical Engineering, Carnegie Mellon University, 2013
[chair: Jonathan Cagan],
Baker, Kyri- Coordination of Resources across Areas for the Integration of Renewable Generation:
Operation, Sizing, and Siting of Storage Devices, Ph.D. Electrical and Computer Engineering, Carnegie
Mellon University, 2014 [chair: Gabriela Hug-Glanzmann],
Jenn, Alan - Analysis of Transportation Policies: Regulations and Incentives in the United States,
Ph.D. Engineering and Public Policy, Carnegie Mellon University, 2014 [chair. Ines Azevedo],
Heo, Jinhyok- Evaluation of Air Quality Impacts on Society: Methods and Application, Ph.D.
Engineering and Public Policy, Carnegie Mellon University, 2014 [chair: Peter Adams],
Dowling, Alex - Equation Oriented Flowsheet Optimization for Advanced Energy Processes, Ph.D.
Chemical Engineering, Carnegie Mellon University, 2015 [chair: Larry Biegler],
Zhang, Qi - Enterprise-wide Optimization for Industrial Demand Side Management, Ph.D. Chemical
Engineering, Carnegie Mellon University, 2016 [chair: Ignacio Grossmann],
Seki, Stephanie - Evaluating the Economic, Environmental and Policy Impacts ofEthanol as a
Transportation Fuel in Pennsylvania, Ph.D. Engineering and Public Policy, Carnegie Mellon University,
2016 [chairs: Chris Hendrickson and W. Michael Griffin],
Tong, Fan - The Good, the Bad, and the Ugly: Economic and Environmental Implications of using
Natural Gas to Power On-Road Vehicles in the United States, Ph.D. Engineering and Public Policy,
Carnegie Mellon University, 2016 [chairs: Paulina Jaramillo and Ines Azevedo],
Mersky, Avi - Environmental and Policy Implications of Vehicle Automation and Electrification, Civil
and Environmental Engineering, Carnegie Mellon University, 2017 [chair: Constantine Samaras],
Li, Zhongju (Hugh) - Urban Aerosol: Spatiotemporal Variation and Source Characterization,
Mechanical Engineering, Carnegie Mellon University, 2018 [chair: Albert Presto],
A-29
-------
Peer Review Report
Work Assignment 3-05/4-05, Contract 68HE0C18C0001
Ciez, Rebecca - Battery Energy Storage for Maturing Markets: Performance, Cost, Perceptions, and
Environmental Impacts, Ph.D., Engineering & Public Policy, Carnegie Mellon University, 2018 [Chair:
Jay Whitacre],
Chen, Mo - Vehicle and Vehicle-related Policy Analysis in China, Engineering and Public Policy,
Carnegie Mellon University (current) [chair: Paul Fischbeck],
Styler, Alex - Learning, Prediction, and Optimization for Hybrid Vehicle Energy Management, School
of Computer Science, Carnegie Mellon University (current) [chair: lllah Nourbakhsh],
Tee, Chin Yen - Market Design and Risk Management for Flexibility in the Future Electric
Transmission System, Engineering and Public Policy, Carnegie Mellon University, 2017 [chair: Marija
II ic].
Chen, Qi - Computational Tools for Process Synthesis, Chemical Engineering, Carnegie Mellon
University, 2020 [chair: Ignacio Grossmann],
Lu, Quanyang - Mobile Source Emission Profiles and the Social Cost of Air Pollution, Mechanical
Engineering, Carnegie Mellon University, 2020 [chair: Robinson],
He, Guannan - Storage for Low-Carbon Energy System: Decisions, Economics, Business Model, and
Policy, Engineering and Public Policy, Carnegie Mellon University, 2019 [chair: Jay Whitacre],
Farquharson, DeVynne - Sustainable Energy Transitions in Sub-Saharan Africa: Impacts on Air
Quality, Economics, and Fuel Consumption, Engineering and Public Policy, Carnegie Mellon
University, 2019 [chair: Jaramillo],
Braaten, Jonathan - Understanding Catalyst Layer Morphology and Degradation and their Impact
on Critical Oxygen Transport in PEMFC Cathodes, Mechanical Engineering, Carnegie Mellon, 2021
[chair: Litster],
Gowharji, Waleed - Mechanical Engineering, Carnegie Mellon (current) [chair: Kate Whitefoot]
Mohan, Aniruddh - Mechanical Engineering, Carnegie Mellon (current) [chair: Parth Vaishnav and
Venkat Viswanathan]
Choi, Paul - (TBD) Mechanical Engineering, Carnegie Mellon (current) [chair: Shawn Litster]
TEACHING EXPERIENCE
19-668 Electric Vehicles: Technology, Economics, Environment and Policy (2021-pres)
In this course, students read academic literature, government documents, and popular press to
develop a broad understanding of the technology, economic, environmental and policy dimensions
of electric vehicles. Topics include (1) Technology: Battery technology, design, application,
degradation and innovation; electric vehicle technologies and designs; the electric power grid; (2)
Economics: cost; consumer behavior; infrastructure; electricity dispatch; automotive externalities;
the Gruenspecht effect; (3) Environment: life cycle assessment; air pollution; greenhouse gas
emissions; marginal grid emission factors; renewables; vehicle to grid; hydrogen; (4) Policy:
effectiveness, efficiency, uncertainty and equity; short-run versus long-run effects; fleet standards;
incentives; mandates; policy interactions; intellectual property; and policies in the US, China, EU,
Japan, and local jurisdictions. Fundamentals covered at an introductory level to support readings
include time value of money, economies of scale, social welfare analysis, externalities, valuation of
reduced mortality risk; choice modeling, regression, life cycle assessment, and other topics.
24-785 Engineering Optimization (2010-pres)
This course introduces students to 1) the process of formally representing an engineering design or
decision-making problem as a mathematical problem and 2) the theory and numerical methods
A-30
-------
Peer Review Report
Work Assignment 3-05/4-05, Contract 68HE0C18C0001
needed to understand and solve the mathematical problem. Theoretical topics focus on constrained
nonlinear programming, including necessary and sufficient conditions for local and global optimality
and numerical methods for solving nonlinear optimization problems. Additional topics such as linear
programming, mixed integer programming, global optimization, and stochastic methods are briefly
introduced. Model construction and interpretation are explored with metamodeling and model
reformulation techniques, study of model boundedness, constraint activity, and sensitivity analysis.
Matlab is used in homework assignments for visualization and algorithm development, and students
apply theory and methods to a topic of interest in a course project.
19-670 / 24-680 Quantitative Entrepreneurship: Analysis for New Technology Commercialization
(2008-pres)
Formerly 19-484/24-484/19-784/24-784 Decision Tools for Engineering Design and
Entrepreneurship; Co-instructors: Prof. Erica Fuchs, Prof. Kate Whitefoot
This course provides engineers with a multidisciplinary mathematical foundation for integrated
modeling of engineering design and enterprise planning decisions in an uncertain, competitive
market. Topics include economics in product design, manufacturing and operations modeling and
accounting, consumer choice modeling, survey design, conjoint analysis, decision-tree analysis,
optimization, model integration, and professional communication skills. Students will apply theory
and methods to a team project for a new product or emerging technology of their choice,
developing a business plan to defend technical and economic competitiveness. Students may select
emerging technologies from research at Carnegie Mellon for study in the course, and in some years
venture capitalists and other industry leaders may take part in critiquing student projects. This
course assumes fluency with calculus and some prior programming experience.
24-441 / 24-442 Engineering Design II: Conceptualization and Realization, Carnegie Mellon University
(2006-pres)
In this course, students gain hands-on, practical experience applying engineering principles,
theories, thought processes, and problem-solving approaches to the design and prototyping of a
physical product. Students develop skills for working in teams, working with open-ended problems,
and making appropriate engineering assumptions. Students are expected to research the topic area,
identify opportunities and design criteria, generate creative concepts, synthesize detailed design of
the concept, analyze the design on a number of criteria to make improvements, and prototype and
communicate the final solution.
24-789C Quantitative Methods for Product Design and Development, Carnegie Mellon
University (2006)
This course provides a multidisciplinary mathematical foundation for integrated modeling in product
design and development. In this course, students learn introductory theory and methods for
building and solving optimization models integrating producer, consumer, competitor, regulator,
and designer perspectives. Topics include optimization methods with a focus on nonlinear
programming, choice modeling, conjoint analysis, game theory, policy analysis, decomposition and
model integration. Students apply methods to a team project and engage in independent research
to deepen knowledge on a relevant topic of interest.
ENGR490 Engineering for Community, University of Michigan (2003-2004)
Worked to co-design, develop, and implement a new interdisciplinary engineering course focusing
on learning through experience and field work, international and local community development,
social and cultural awareness, sustainability, appropriate technology, and communication skills.
A-31
-------
Peer Review Report
Work Assignment 3-05/4-05, Contract 68HE0C18C0001
Graduate Student Instruction Mentor, University of Michigan (2003-2004)
University of Michigan Center for Research on Learning and Teaching
Trained and assisted graduate student instructors, facilitated student feedback on teaching and
provided teaching observations and evaluations.
ME 499/599 Analytical Product Design, University of Michigan (2003, 2004)
Helped to develop and implement a new interdisciplinary course on model-based product
development. Gave lecturers, provided student support, and developed computer tools for
students.
ME 555 Design Optimization, University of Michigan (2003)
Acted as a course aid for the graduate-level course on design optimization. Developed new material
and gave lectures and demonstrations.
Rackham-CRLT Seminar on College Teaching, University of Michigan (2003)
Participated in the selective Center for Research on Learning and Teaching Preparing Future Faculty
Seminar on College Teaching to sharpen teaching philosophy, learn about the structure of higher
education institutions in the United States, and develop teaching skills.
Multicultural Classroom Facilitation Training, University of Michigan (2003)
Participated in a training course emphasizing teaching through dialogue and focusing on issues of
multiculturalism, diversity, and social and cultural awareness.
Detroit Area Pre-College Engineering Program, University of Michigan (2002-2003)
Exposed middle-school students from minority and low-income areas to engineering by providing
hands-on examples and activities in order to improve accessibility of the field. Position: instructor.
Michigan Mentor Program, University of Michigan (2000-2002)
Worked with two middle school students one-on-one to provide exposure to the engineering
disciplines and work with them to explore career possibilities in engineering and beyond.
ME 450 Design and Manufacturing III, University of Michigan (2000-2001)
Assisted in teaching and advising the capstone mechanical engineering design course. Developed
new course material and restructured the syllabus. Initiated and supported interdisciplinary student
coursework and project interaction with the University Of Michigan School of Art and Design and
the Parsons School of Design. Developed an interactive course web portal to support distance
collaboration.
Tutor, Carnegie Mellon University (1997)
Tutored undergraduates in programming the C++ language.
CONSULTING EXPERIENCE
New York State Energy Research and Development Authority (NSERDA) (2014-2017)
With Industrial Economics, Inc.
INDUSTRY EXPERIENCE
Research Engineer, Intern
Xerox-The Document Company, Rochester, NY (summer 1999)
Designed and implemented improvements to a distributed embedded digital control system for
paper path handling.
A-32
-------
Peer Review Report
Work Assignment 3-05/4-05, Contract 68HE0C18C0001
Design Engineer, Intern
General Motors, Warren, Ml (summer 1998)
Worked with a multidisciplinary team including engineering, industrial design, and sculpture to
design and prototype a future concept vehicle with engineering documentation to show concept
feasibility.
Reliability Engineer, Intern
General Motors, Warren, Ml (summer 1997)
Developed and implemented a procedure to quantify the reliability of commercial components of
machinery and equipment in the assembly plant environment.
Designer
Linear Systems Corporation, Rochester, Ml (1994-1996)
Designed automotive tooling. Trained employees in AutoCAD. Developed new CAD tools.
ORGANIZATIONAL LEADERSHIP AND MEMBERSHIP
Leadership
Director, Vehicle Electrification Group (since 2012)
Co-Director, Vehicle Electrification Group (2009-2012)
Chair, EPP Master of Science Program Committee (2020-pres)
Chair, EPP Communications and Diversity Committee (2017-2018)
Acting Co-Director, Master in Product Development program (2008-2009)
Director, Design Decisions Laboratory, Carnegie Mellon University (since 2005)
Chair, Publicity and Newsletter Committee, ASME Design Engineering Division (2007-2009)
Membership
Member, Center for Climate and Energy Decision Making, Carnegie Mellon University (since 2011)
Member, Electricity Industry Center, Carnegie Mellon University (since 2011)
Member, Center for Product Strategy and Innovation, Carnegie Mellon University (2008-2010)
Member, Green Design Institute, Carnegie Mellon University (since 2006)
Member, ASME: American Society of Mechanical Engineers (since 1997)
Member, INFORMS: Institute for Operations Research and the Management Sciences (since 2005)
Member, TRB: National Academies Transportation Research Board Alterative Transportation Fuels
and Technologies Committee (2014-present)
Graduate Student Activities
Engineers Without Borders / BLUElab, University of Michigan, Education Chair (2003-2005)
Graduate Student Mentor (2003-2004)
Amnesty International Student Group on Economic and Social Rights (2003-2004)
Public Service Announcement Director, WCBN-FM Ann Arbor (2002-2004)
University of Michigan Graduate Student Symposium, Design and Manufacturing Chair (2001)
Interdisciplinary Antilium Education and Research Collaboration Initiative (2001-2005)
New Foreign Graduate Student Mentor (2000-2002)
A-33
-------
Peer Review Report
Work Assignment 3-05/4-05, Contract 68HE0C18C0001
New Graduate Student Recruiter (2000)
RESEARCH GRANTS AND AWARDS
Principal Investigator: (total $2.6M)
[1] Mobility21, "Ridehailing Service Equity in Normal and Rare Conditions/' Nov 2020.
[2] ATLAS Moonshot, "Autonomous Taxis: Understanding and Improving Sustainability and Equity/'
Aug 2020.
[3] U.S. Department of Energy "Dynamic Effects of Technology-Forcing ZEV Policy under Vehicle
Fleet Standards/' Aug 2019
[4] National Bureau of Economic Research, "Effects of On-Demand Ridesourcing on U.S. Vehicle
Ownership, Travel Patterns, and Energy Use Externalities" March 2018.
[5] Steinbrenner Institute for Environmental Education and Research, "How Do Shared Mobility and
Autonomous Taxis Affect Energy Consumption, Vehicle Use, and Emissions of Greenhouse Gases
and Criteria Air Pollutants?" August 2017.
[6] Scott Institute for Energy Innovation, "How Do Shared Mobility and Autonomous Taxis Affect
Energy Consumption, Vehicle Use, and Greenhouse Gas and Criteria Pollution Emissions?"
January 2017.
[7] Toyota Motor Corporation, "Advanced Vehicle Research," January, 2015.
[8] Toyota Motor Corporation, "Vehicle Electrification Research and Policy," August 2013.
[9] Toyota Motor Corporation, "Autonomous Vehicle Research," [co-PI Chris Hendrickson] June
2013.
[10] Toyota Motor Corporation, "Advanced Vehicle Research," October 2012.
[11] Toyota Motor Corporation, "PHEV Research," [co-PI Jay Whitacre], September 2011.
[12] Ford Motor Company, "Electrified Vehicles in China: Identifying Consumer Preferences and Key
Factors that Impact Adoption," Sept 2011 - Aug 2014.
[13] Toyota Motor Corporation, "Life Cycle Cost and Environmental Assessment of Plug-in Vehicles,"
[co-PIs Jay Whitacre, Constantine Samaras], August 2010.
[14] Toyota Motor Corporation, "Systems Analysis of Plug-in Hybrid Electric Vehicles," [co-PIs Jay
Whitacre, Constantine Samaras], April 2009.
[15] National Science Foundation, CAREER Award, "Driving Design: Modeling the Influence of Market
Forces and Public Policy on Engineering Design Decision-Making," Aug 2008 - Aug 2013.
[16] Ford Motor Company, "Engineering and Market Simulation for Optimal Product Planning under
Environmental Regulation," Aug 2008 - Aug 2011.
[17] Carnegie Mellon Institute for Entrepreneurship, Innovation and Technology course development
grant for "Quantitative Methods for Product Design and Development," Aug 2007 - July 2012.
[18] Pennsylvania Infrastructure Technology Alliance, "A Systems Decomposition Approach for
Optimization of Product Families to Balance Market and Engineering Needs," Jan - Dec 2007
[Co-PIs Peter Boatwright, Tim Simpson],
[19] Pennsylvania Infrastructure Technology Alliance, "Hierarchical Design Optimization of Complex
Systems," Jan - Dec 2006 [Co-PIs Jon Cagan, Zhaosong Lu],
Co-Investigator (total $15M)
[20] Center for Applied Environmental Law and Policy, "Changing Consumer Preferences for EVs Over
Time," 2020.
A-34
-------
Peer Review Report
Work Assignment 3-05/4-05, Contract 68HE0C18C0001
[21] U.S. Department of Energy, "Understanding and Improving Energy Efficiency of Regional
Mobility Systems Leveraging System-Level Data/' 2018.
[22] U.S. Department of Energy, "Drones, Delivery Robots, Driverless Cars, and Intelligent Curbs for
Increasing Energy Productivity of First/Last Mile Goods Movement," 2018.
[23] Carnegie Mellon University, "Carnegie Mellon Equipment Grant Proposal," 2017.
[24] Electric Power Research Institute, "Estimating Consumer Preferences for Diverse Electric
Services," 2016.
[25] Carnegie Mellon University College of Engineering, "Workshop on Techno-Economic Assessment
Methods for Energy Technologies," 2016.
[26] Scott Institute for Energy Innovation, "Techno-Economic Assessment Methods for Energy
Technologies," 2016.
[27] Environmental Protection Agency Air, Climate, and Energy (ACE) Center, "Center for Air, Climate,
and Energy Solutions (CACES)", 2015-2020.
[28] Carnegie Mellon University Transportation Center, "Evaluating the Opportunities for Cost
Savings and Environmental Benefits of Coupling Solar Energy and Electric Vehicles in City of
Pittsburgh Municipal Operations," 2015.
[29] Fuels Institute "Comparative Analysis of the Economic and Environmental Impacts of CNG and
LNG for the Transportation Sector," 2014-2015.
[30] Fuel Freedom Foundation "Assessment of Comparative Economic and Environmental Impacts of
Alternative Light Duty Vehicle Liquid Fuels Produced from Natural Gas," 2014-2015.
[31] Carnegie Mellon Metro 21 "Evaluating the Opportunities for Cost Savings and Environmental
Benefits of Coupling Solar Energy and Electric Vehicles in City of Pittsburgh Municipal
Operations," 2014-2015 [PI: Constantine Samaras],
[32] Research for Advanced Manufacturing in Pennsylvania "Manufacturing Modeling Tools for
Domestic Energy Storage Production: Process-Based Cost Modeling," 2012-2014 [PI: Erica
Fuchs],
[33] Carnegie Institute of Technology "Institutionalizing and Disseminating Engineering
Entrepreneurship," 2012-2013 [PI: Erica Fuchs],
[34] National Science Foundation, SciSIP Program (2011) "GOALI: Think Globally Act Locally - China
and the Future of Energy-Saving Vehicle Technologies," Sept 2011 - Aug 2014 [PI: Erica Fuchs],
[35] Mellon Foundation award for "Course Instructor Outreach to Carnegie Mellon's Center for
Technology Transfer," 2008-2009 [PI: Erica Fuchs],
[36] National Science Foundation, MUSES Program "Material Use, Infrastructure Change, and
Environmental Impacts for Alternative Fuels and Vehicles," Sept 2006 - Aug 2011 [PI: Lester
Lave. Co-PIs: Chris Hendrickson, H. Scott Matthews, W. Michael Griffin],
[37] United States Environmental Protection Agency P3 Award People Prosperity and the Planet
Award for project "AWARE: A Step Toward Building a Sustainable Economy by Informing
Consumer Purchasing Decisions at the Point of Sale," Sept 2004 - May 2005 [PI: Steven Skerlos,
co-investigator W. Ross Morrow],
Student Awards (total $0.3M)
[38] Dwight David Eisenhower Transportation Fellowship "How to Autonomy, Connectivity, and the
Sharing Economy Affect Passenger Travel? Informing the Technical and Policy Framework for a
New Mobility Future" student: Ward 2017.
A-35
-------
Peer Review Report
Work Assignment 3-05/4-05, Contract 68HE0C18C0001
[39] Link Foundation "Environmental Implications of Consumer Preferences and Policy Incentives for
Plug-in Vehicles in China and the U.S." student: Helveston 2014-2015.
[40] National Science Foundation Graduate Research Fellowship "Can Controlled Charging of Electric
Vehicles Reduce the Economic and Environmental Implications of Integrating Wind Power into
the Electricity Grid?" student: Allison Weis, 2011, June 2012 - May 2015.
[41] National Science Foundation Graduate Research Fellowship "How Does Energy Policy Affect
Vehicle Design?" student: Elizabeth Traut, 2009, June 2009 - May 2012.
[42] Steinbrenner Institute for Environmental Education and Research "How Does Energy Policy
Affect Vehicle Design?" student: Elizabeth Traut, August 2009 - July 2010.
HONORS AND AWARDS
Faculty
Steven J. Fenves Award for Systems Research, Carnegie Mellon University (2018)
Thar Energy Design Award, American Society of Mechanical Engineers (2016)
Philip L. Dowd Fellowship Award, College of Engineering, Carnegie Mellon University (2015)
Best Article Award, European Marketing Academy, International Journal of Research in Marketing, Best
Article in 2011
Best Course, awarded by the Carnegie Mellon Mechanical Engineering Class of 2010
American Society of Mechanical Engineers, Design Automation Outstanding Young Investigator Award
(2009)
George Tallman Ladd Research Award for outstanding research and professional accomplishments and
potential (2008)
Best Use of Technology in the Classroom, awarded by the Carnegie Mellon Mechanical Engineering
Class of 2006
American Society of Mechanical Engineers Design Automation Conference Best Paper Award (2005)
Student
NSF Engineering Research Center for Reconfigurable Manufacturing Systems Ph.D. Student of the Year
Award (2005)
Elaine Harden Award for Outstanding Leadership in College, University and Community Activities,
(2005) - University of Michigan BLUElab Executive Committee
Martin Luther King Jr. Spirit Award for co-development of new Engineering for Community course
(2004) - University of Michigan
Michigan Teaching Fellow (2003) - University of Michigan Horace H. Rackham School of Graduate
Studies and the Center for Research on Learning and Teaching
University of Michigan Mechanical Engineering Graduate Student Council Second Place Award for
Research Poster Competition at the Graduate Student Symposium (2003)
Rackham Interdisciplinary Institute Fellowship (2001-2002)
Devlieg Fellowship and Scholarship (2000)
Engineering Graduate Fellowship (1999)
Graduated first in the ME class of 1999 at Carnegie Mellon with University Honors
Motorola Second Place Research Award for Sigma Xi Undergraduate Research Symposium (1999)
A-36
-------
Peer Review Report
Work Assignment 3-05/4-05, Contract 68HE0C18C0001
Carnegie Institute of Technology Third Place Research Award for Sigma Xi Undergraduate Research
Symposium (1999)
Bennett Award for Academic Achievement (1999)
Department Research Honors, Carnegie Mellon Mechanical Engineering (1999)
Student Leadership Award, Carnegie Mellon University (1999)
ACADEMIC SERVICE (since 2005)
Editor:
ASME Design Engineering Division Newsletter Editor (2006-2009)
Concurrent Engineering: Research and Applications, An International Journal (guest editor, Special
Issue on Managing Modularity and Commonality in Product and Process Development, 2005-2006)
Reviewer:
ACS Omega
AIAA Journal
AMA American Marketing Association Conference
Applied Energy
ASME Journal of Mechanical Design
ASME Design Engineering Technical Conferences
Concurrent Engineering - Research and Applications
Design Science
Energies
Energy Journal
Energy Policy
Engineering Optimization
Environmental Research Letters
Environmental Science and Technology
IEEE Spectrum
IEEE Transactions on Automation Science and Engineering
IEEE Transactions on Engineering Management
IEEE Transactions on Power Systems
IEEE Transactions on Smart Grid
International Journal of Hydrogen Energy
International Journal of Information Technology and Decision Making
International Journal of Manufacturing Technology and Management
International Journal of Product Development
Journal of Cleaner Production
Journal of Intelligent Manufacturing
Marketing Science (INFORMS)
National Petroleum Council
National Renewable Energy Laboratory
A-37
-------
Peer Review Report
Work Assignment 3-05/4-05, Contract 68HE0C18C0001
National Research Council of the National Academies
National Science Foundation
Nature Energy
Proceedings of the National Academy of Sciences of the United States of America
Research in Engineering Design
Science Advances
Structural and Multidisciplinary Optimization
Sustainable Transportation
Transport Policy
Transportation Research Board
Transportation Research Part A: Policy and Practice
Transportation Research Part D: Transport and Environment
Transportation Science
Union of Concerned Scientists
World Electric Vehicle Journal
Conference Session Organizer / Review Coordinator:
ASME International Design Engineering Technical Conferences
INFORMS Annual Meeting
Transportation Research Board Annual Meeting
Conference Session Chair:
ASME Design Engineering Technical Conferences
INFORMS Annual Meeting
Transportation Research Board Annual Meeting
Outreach and Public Service:
Committee on Current Methods for Life Cycle Analysis of Low-Carbon Transportation Fuels in the
United States, The National Academies of Science, Engineering and Medicine (2021-pres)
Testimony at public hearing of Environmental Protection Agency and National Highway Traffic Safety
Administration on proposed changes to the federal light-duty vehicle fuel economy and greenhouse
gas emissions standards (2018)
Commentary on Pennsylvania House Bill 1446 for Office of PA Representative Dan Frankel (2018)
Policy Briefing, U.S. House of Representatives (2017)
Policy Briefing, National Governors Association (2017)
Policy Briefing, U.S. Department of Transportation (2017)
Policy Briefing, Office of U.S. Senator Toomey (2017)
Policy Briefing, National Resources Defense Council (2016)
Policy Briefing, National Renewable Energy Laboratory (2016)
Policy Briefing, Environmental Protection Agency (2016)
Policy Briefing: California Energy Commission (2015)
Policy Briefing: California Air Resources Board (2015)
A-38
-------
Peer Review Report
Work Assignment 3-05/4-05, Contract 68HE0C18C0001
Policy Briefing: California State Senate Transportation Committee (2015)
Policy Briefing: California State Assembly Transportation Committee (2015)
Policy Briefing: Office of State Senator Fran Pavley (2015)
Policy Briefing: California State Assembly Natural Resources Committee (2015)
Policy Briefing, Union of Concerned Scientists (2015)
Policy Briefing, U.S. Congressional Budget Office (2012)
Policy Briefing, U.S. Congressional Research Service (2012)
Policy Briefing, U.S. Senate Energy and Natural Resources Committee (2012)
Policy Briefing, U.S. Senate Commerce, Science and Transportation Committee (2012)
Policy Briefing, Office of U.S. Representative Levin (2012)
Policy Briefing, National Academy of Engineering, Maxine Savitz, Vice President (2012)
National Petroleum Council study on Future Transportation Fuels, Electricity Subgroup (2010-2012)
Policy Briefing, U.S. House of Representatives Energy and Commerce Committee (2009)
Policy Briefing, U.S. House of Representatives Committee on Science and Technology (2009)
Policy Briefing, U.S. House of Representatives Select Committee on Energy Independence and Global
Warming (2009)
Policy Briefing, U.S. Congressional Research Service (2009)
Policy Briefing, Office of U.S. Senator Specter (2009)
Policy Briefing, Office of U.S. Representative Markey (2009)
Green Design Apprenticeship, 6 day course for high-school students (2007-present)
Society of Women Engineers High School Day Workshop (2005-2008)
Society Service:
Transportation Research Board of the National Academies, Alternative Transportation Fuels and
Technologies Committee (member since 2014, friend since 2013)
Transportation Research Board of the National Academies, Transportation Energy Committee (friend
since 2013)
ASME Design Engineering Division, Chair: Publicity and Newsletter Committee (2007-2009)
ASME Design Automation Committee (since 2005)
University Service:
Engineering and Public Policy MS Program Committee Chair (2020-pres)
Engineering and Public Policy Strategic Review Committee (2019-2020)
Engineering and Public Policy Department Head Search Committee (2019-2020)
Engineering and Public Policy Communications and Diversity Committee (2016-2018)
Mechanical Engineering Awards Committee (2019-pres)
Mechanical Engineering Undergraduate Education Committee (2016-pres)
Engineering and Public Policy Communications and Diversity Committee, Chair (2016-2017)
Carnegie Institute of Technology Ad Hoc Promotion and Tenure Committee (2014-2015)
Chair, Carnegie Mellon Mechanical Engineering Communications Committee (2013-2014)
Department Head Search Committee, Engineering and Public Policy (2013-2014)
A-39
-------
Peer Review Report
Work Assignment 3-05/4-05, Contract 68HE0C18C0001
Chair, Carnegie Mellon Engineering and Public Policy Graduate Education Committee (since 2013)
Department Head Search Committee, Mechanical Engineering (2012)
Acting co-director, Master in Product Development program, (2008-2009)
Mechanical Engineering Curriculum Assessment Committee (since 2010)
Mechanical Engineering Undergraduate Education Committee (2006-2011)
Mechanical Engineering Library Committee (2006-2008)
Engineering and Public Policy Qualifying Examination Service (since 2006)
Mechanical Engineering Qualifying Examination Service (since 2005)
Mechanical Engineering Seminar Series Committee (2005-2006)
INVITED SEMINARS
[1] Uber and Lyft Implications: Vehicle Ownership, Transit, Electrification, Air Emissions, Traffic
Externalities and Equity, CMU Smart Mobility Connection, March 2021 (online).
[2] Externalities of Policy-Induced Scrappage: The Case of Automotive Regulations, U.S.
Environmental Protection Agency, May 2021 (online).
[3] Evolving Consumer Preferences for Electric Vehicles, U.S. Environmental Protection Agency,
March 2021 (online).
[4] Alternative-Fuel-Vehicle Policy Interactions Increase U.S. GHG Emissions, Industry Studies
Association, June 2020 (online).
[5] Effects of On-Demand Ridesourcing on U.S. Vehicle Ownership, Travel, Energy, and
Environmental Outcomes, National Bureau of Economic Research, 2019, Washington D.C.
[6] U.S. Light-Duty Vehicle Fleet Efficiency and Greenhouse Gas Emissions Policy - Past, Present,
and Future, Webinar provided by Carnegie Mellon Transportation Electrification Group, 2018.
[7] Effects of On-Demand Ridesourcing on U.S. Vehicle Ownership, Travel Patterns, and Energy Use
Externalities, National Bureau of Economic Research, 2018, Washington, D.C.
[8] Automobile Air Emissions and the Transition to Electric Powertrain Technology, Carnegie Mellon
Department of Mechanical Engineering, Mar 23, 2018, Pittsburgh, PA.
[9] Carnegie Mellon Vehicle Electrification Group, Department of Energy Vehicle Technologies
Office Meeting, Oct 25, 2017, Pittsburgh, PA.
[10] Carnegie Mellon Vehicle Electrification Group, Federal Highway Administration Meeting, Sept
25, 2017, Pittsburgh, PA.
[11] Carnegie Mellon Vehicle Electrification Group, National Renewable Energy Laboratory Smart
PGH Meeting, April 19, 2017, Pittsburgh, PA.
[12] Implications of Utility-Controlled Charging for Electric Vehicle Cost and Emissions, Vehicle Grid
Integration Workshop, Transportation Research Board Annual Meeting, Jan 8, 2017,
Washington, D.C.
[13] Techno-Economic Analysis for Differentiated Products, Institute for Techno-Economic
Assessment Methods, Carnegie Mellon University, Nov 18, 2016, Pittsburgh, PA.
[14] Electric Vehicle Adoption Potential in the United States, University of California, Davis, Apr 12,
2016, Davis, CA.
[15] Electric Vehicles: Benefits, Costs, and Policies in the United States, National Resources Defense
Council, Apr 11, 2016, San Francisco, CA.
A-40
-------
Peer Review Report
Work Assignment 3-05/4-05, Contract 68HE0C18C0001
[16] Upstream Emissions from Electric Vehicle Charging, UN Economic Commission for Europe EVE
Working Group, Apr 11, 2016, delivered remotely to China.
[17] Electric Vehicles: Benefits, Costs, and Policies in the United States, Stanford University Precourt
Institute for Energy, Apr 8, 2016, Stanford, CA.
[18] Alternative Fuel Vehicle Adoption Increases U.S. Fleet Gasoline Consumption and Greenhouse
Gas Emissions under Federal Corporate Average Fuel Economy and Fleet Greenhouse Gas
Emission Policy, King Abdullah Petroleum Studies and Research Center Workshop, April 1, 2016,
San Francisco, CA.
[19] Electric Vehicles: Benefits, Costs, and Policies in the United States, Energy Institute at Haas,
University of California, Berkeley, March 31, 2016, Berkeley, CA.
[20] Electric Vehicles: Benefits, Costs, and Policies in the United States, Lawrence Berkeley National
Laboratory, Mar 29, 2016, Berkeley, CA.
[21] Electric Vehicles: Benefits, Costs, and Policies in the United States, Renewable and Appropriate
Energy Laboratory, Energy Resources Group, University of California, Berkeley, Mar 16, 2016,
Berkeley, CA.
[22] Electric Vehicles: Benefits, Costs, and Policies in the United States, Ford Motor Company, Mar 3,
2016, Palo Alto, CA.
[23] Electric Vehicle Adoption Potential in the United States, National Renewable Energy Laboratory,
Feb 2, 2016, Golden, CO.
[24] Electric Vehicles: Benefits, Costs, and Policies in the United States, National Renewable Energy
Laboratory, Feb 2, 2016, Golden, CO.
[25] Electric Vehicles: Benefits, Costs, Policies, and Adoption Potential in the United States,
Environmental Protection Agency, Office of Transportation and Air Quality, Jan 20, 2016, Ann
Arbor, Ml.
[26] Electric Vehicles: Benefits, Costs, Policies, and Adoption Potential in the United States, Centro
para a Excelencia e Inovagao na Industria Automovel, Dec 15, 2015, Matosinhos, Portugal.
[27] Electric Vehicles: Benefits, Costs, Policies, and Adoption Potential in the United States, Instituto
Superior Technico, Universidade de Lisboa, Dec 11, 2015, Lisbon, Portugal.
[28] Electric Vehicles: Benefits, Costs, and Adoption Potential in the United States, California Energy
Commission, June 2015, Sacramento, CA.
[29] Electric Vehicles: Benefits, Costs, and Adoption Potential in the United States, California Air
Resources Board, June 2015, Sacramento, CA.
[30] Electric Vehicles: Benefits, Costs, and Adoption Potential in the United States, University of
California, Davis, June 2015, Davis, CA.
[31] Electric Vehicles: Benefits, Costs, and Adoption Potential in the United States, Union of
Concerned Scientists, June 2015, Oakland, CA.
[32] Carnegie Mellon Vehicle Electrification Group, Toyota Motor Sales, September 12, 2014,
Torrance, CA.
[33] Vehicle Automation: Implications for Advanced Vehicle Energy Technologies, University of
Michigan Transportation Research Institute Global Symposium (invited by Jacob Ward, US DOE),
April 23, 2014, Ann Arbor, Ml.
[34] Plug-in Vehicle Life Cycle Benefits and Costs: Implications and Strategies, University of Illinois at
Urbana-Champaign, February 28, 2013, Champaign, IL.
A-41
-------
Peer Review Report
Work Assignment 3-05/4-05, Contract 68HE0C18C0001
[35] The Costs and Benefits of Plug-in Vehicles: How Much Can We Control? University of California,
San Diego, February 22, 2013, San Diego, CA.
[36] Air Emissions and Oil Displacement Benefits from Plug-in Vehicles, Society of Automotive
Engineers Hybrid and Electric Vehicle Symposium, February 20, 2013, Anaheim, CA.
[37] Effect of Location and Driving Conditions on Plug-in Vehicle Benefits, Transportation Research
Board of the National Academies, January 13, 2013, Washington D.C.
[38] Quantifying Plug-in Vehicle Benefits, Argonne National Laboratory, Aug 17, 2012, Argonne, IL.
[39] Thoughts on the Field of Design for Market Systems, Northwestern University, Aug 16, 2012,
Evanston, IL.
[40] Are Plug-in Vehicles Worth the Cost? Valuing Air Emissions and Oil Displacement Benefits in the
U.S., March 14, 2012 Toyota Motor North America, Washington, DC.
[41] Are Plug-in Vehicles Worth the Cost? Valuing Air Emissions and Oil Displacement Benefits in the
U.S., Sept 2011, Cambridge University, U.K.
[42] Life Cycle Cost, Air Emissions, and Oil Displacement Potential of Plug-in Vehicles, Ford Motor
Company, July 2011, Dearborn, Ml.
[43] On the Life Cycle Implications of Plug-in Hybrid Electric Vehicles, University of California
Berkeley, Oct 22, 2010, Berkeley, CA.
[44] On the Life Cycle Implications of Plug-in Hybrid Electric Vehicles, Stanford University, Precourt
Energy Efficiency Center, Oct 21, 2010, Stanford, CA.
[45] Product Design in Strategic Firm Decision-Making, Stanford Graduate School of Business,
Operations, Information and Technology Group, Oct 20, 2010, Stanford, CA.
[46] On the Life Cycle Implications of Plug-in Hybrid Electric Vehicles, University of California Davis,
ITS-STEPS Seminar, Oct 19, 2010, Davis, CA.
[47] Do More Batteries Make a Plug-in Better? Economic and Environmental Analysis of Plug-in
Hybrid Electric Vehicles, Society of Automotive Engineers Government-Industry Meeting,
January 28, 2010, Washington, D.C.
[48] Market Forces and Public Policy in Engineering Systems Optimization, Massachusetts Institute of
Technology, Engineering Systems Division, June 12, 2009, Boston, MA.
[49] Design for Market Systems: Integrating Social, Economic, and Physical Sciences to Engineer
Product Success, University of Maryland - College Park, Design and Reliability of Systems
Division, January 14, 2009, College Park, MD.
[50] Design for Market Systems: Integrating Social, Economic, and Physical Sciences to Engineer
Product Success, The Pennsylvania State University, Industrial and Manufacturing Engineering,
October 16, 2008, State College, PA.
[51] Driving Design: Modeling Market Forces and Public Policy in Vehicle Design, State University of
New York at Buffalo, Mechanical and Aerospace Engineering, April 10, 2008, Buffalo, NY.
[52] Should Designers Worry about Market Structure?, The Pennsylvania State University,
Engineering Design, April 4, 2008, State College, PA.
[53] Modeling Energy Policy and Consumer Choice in Vehicle Design Optimization, Ford Motor
Company, Systems Analytics and Environmental Sciences, July 25, 2007, Detroit, Ml.
[54] Realizable Product Line Optimization: Coordinating Product Positioning and Design for
Heterogeneous Markets, Tepper School of Business, Carnegie Mellon University, Feb. 24, 2006,
Pittsburgh, PA.
A-42
-------
Peer Review Report
Work Assignment 3-05/4-05, Contract 68HE0C18C0001
[55] A Model for Studying the Impact of Fuel Economy and Emission Policy on Profit-Driven Vehicle
Design Decisions in a Competitive Market, Green Design Institute, Carnegie Mellon University,
Oct. 20, 2005, Pittsburgh, PA.
[56] Preference Coordination in Engineering Design Decision-Making, Mechanical Engineering,
Northwestern University, March 31, 2005, Evanston, IL.
[57] Preference Coordination in Engineering Design Decision-Making, General Motors Research and
Development Seminar, Oct. 27, 2004, Detroit, Ml.
[58] Individual, Social and Economic Preference in Engineering Design Decision-Making, Eindhoven
University of Technology, April 22, 2004, Eindhoven, The Netherlands.
[59] Automotive Design and Environmental Policy in the US Market, Delft University of Technology,
April 21, 2004, Delft, The Netherlands.
[60] Coordination of Preferences Using Hierarchical Optimization of Complex Systems, Ecole Centrale
de Nantes, April 19, 2004, Nantes, France.
[61] Automotive Design and Environmental Policy in the US Market, Technischen Universitat Berlin,
April 8, 2004, Berlin, Germany.
PROFESSIONAL PRESENTATIONS
[62] Externalities of Policy-Induced Scrappage: The Case of Automotive Regulations, Transportation
Research Board Annual Meeting, Jan 2021 (with C. Forsythe)
[63] The Dynamic Costs and Benefits of Technology-Forcing Policy Nested in a Broader Performance
Standard: The Case of ZEV and CAFE, Transportation Research Board Annual Meeting, Jan 2021
(with A. Yip)
[64] Effects of Air Emissions Externalities on Optimal Ride-Hailing Fleet Electrification and
Operations, National Bureau of Economics, Energy Use in Transportation Conference, June 2020,
Washington D.C.
[65] Alternative Fuel Vehicle Policy Interactions Increase U.S. Greenhouse Gas Emissions, Industry
Studies Association, June 2020 (online).
[66] Effects of On-Demand Ridesourcing on U.S. Vehicle Ownership, Energy and Environmental
Outcomes in the United States, January 2020, Washington D.C. (with J. Ward)
[67] Ride-Hailing Fleets Increase Vehicle Electrification and Reduce Emissions, January 2020,
Washington D.C. (with M. Bruchon)
[68] Alternative Fuel Vehicle Policy Interactions Increase U.S. Greenhouse Gas Emissions,
Transportation, Economics, Energy and Environment Conference, Oct 2019, Ann Arbor, Ml.
[69] Effects of On-Demand Ridesourcing on U.S. Vehicle Ownership, Travel, Energy, and
Environmental Outcomes, National Bureau of Economic Research, May 2019, Washington D.C.
[70] U.S. Federal and State Policy Interactions for Alternative Fuel Vehicles, Transportation Research
Board Annual Meeting, Jan 10, 2018, Washington, D.C.
[71] On-Demand Ridesourcing Has Reduced Per-Capita Vehicle Registrations and Gasoline Use in U.S.
States, Transportation Research Board Annual Meeting, Jan 2018, Washington D.C.
[72] Using Composite Vehicles in Choice Model Simulations: Implications on Prediction and Policy,
Transportation Research Board Annual Meeting, Jan 2018, Washington D.C.
[73] U.S. Alternative Fuel Vehicle Policy Interactions Increase Greenhouse Gas Emissions,
Transportation Research Board Annual Meeting, Jan 2017, Washington D.C.
A-43
-------
Peer Review Report
Work Assignment 3-05/4-05, Contract 68HE0C18C0001
[74] Consequential Life-Cycle Air Emissions Externalities for Plug-in Electric Vehicles in the PJM
Interconnection, Transportation Research Board Annual Meeting, Jan 2016, Washington, D.C.
[75] Effect of Regional Grid Mix, Driving Patterns, and Climate on the Comparative Carbon Footprint
of Electric and Gasoline Vehicles, Transportation Research Board Annual Meeting, Jan 2016,
Washington, D.C.
[76] Unintended Consequences: Why U.S. Alternative Fuel Vehicle Adoption Increases Fleet Gasoline
Consumption and Greenhouse Gas Emissions under Federal Corporate Average Fuel Economy
and Greenhouse Gas Emission Policy, United States Association for Energy Economics, Oct 27,
2015, Pittsburgh, PA.
[77] Life Cycle Air Emissions Externality Implications of Electric Vehicle Adoption in the United States:
A Comparison of Empirical and Normative Approaches, United States Association for Energy
Economics, Oct 27, 2015, Pittsburgh, PA (speaker: Jaramillo).
[78] Emissions from Electric Vehicle Charging in the United States, Climate and Energy Decision
Making Center Annual Meeting, May 21, 2015, Pittsburgh, PA.
[79] Forecasting Light-Duty Vehicle Demand using Alternative-Specific Constants for Endogeneity
Correction vs. Calibration, Transportation Research Board Annual Meeting, Jan 14, 2015,
Washington, DC.
[80] Emissions and Cost Implications of Controlled Electric Vehicle Charging in the PJM
Interconnection, Transportation Research Board Annual Meeting, Jan 14, 2015, Washington, DC.
[81] Energy Implications of Partial Vehicle Automation, Transportation Research Board Annual
Meeting, Jan 13, 2015, Washington, DC. (speaker: Hayeri).
[82] Greenhouse Gas Emissions from Alternative Fuel Vehicle Incentives in CAFE Policy,
Transportation Research Board Annual Meeting, Jan 13, 2015 (poster), Washington, DC.
[83] A Techno-Economic Analysis and Optimization of Li-ion Batteries for Light-Duty Passenger
Vehicle Electrification, Transportation Research Board Annual Meeting, Jan 13, 2015 (poster),
Washington, DC.
[84] Regional Uncertainty and Variability of Electric Vehicle Life Cycle C02 Emissions in the U.S.,
Transportation Research Board Annual Meeting, Jan 13, 2015 (poster), Washington, DC.
[85] Effects of Regional Temperature on Electric Vehicle Efficiency, Range and Emissions in the
United States, Transportation Research Board Annual Meeting, Jan 12, 2015, Washington, DC
(speaker: Yuksel).
[86] A Techno-Economic Analysis and Optimization of Li-ion Batteries for Light-Duty Passenger
Vehicle Electrification, Carnegie Mellon Electrochemical Systems Group, Nov 14, 2014,
Pittsburgh, PA.
[87] Regional Emissions from Electric Vehicles, Carnegie Mellon Electricity Industry Center, Oct 21,
2014, Pittsburgh, PA.
[88] Costs and Emissions Implications of Controlled Electric Vehicle Charging, Climate and Energy
Decision Making Center Annual Meeting, May 20, 2014, Pittsburgh, PA.
[89] Consumer Preferences for Hybrid and Electric Vehicles in China and the United States:
Implications for Policy and Environment, Transportation Research Board Annual Meeting,
January 2014, Washington D.C. (speaker: Helveston).
[90] Sensitivity of Vehicle Market Share Predictions to Alternative Discrete Choice Model
Specifications, Transportation Research Board Annual Meeting, January 2014, Washington D.C.
A-44
-------
Peer Review Report
Work Assignment 3-05/4-05, Contract 68HE0C18C0001
[91] Comparative Life-Cycle Cost of Electric Vehicle Battery Exchange Versus Fast Charging Stations,
Transportation Research Board Annual Meeting, January 2014, Washington D.C.
[92] Influence of Driving Patterns on Life-Cycle Cost and Emissions of Hybrid and Plug-in Electric
Vehicle Powertrains, Transportation Research Board Annual Meeting, January 2014, Washington
D.C.
[93] Global Control Optimization of Electric Vehicles with Supercapacitor-Battery Systems Over a Set
of Real-World Speed and Elevation Profiles via Dynamic Programming, Transportation Research
Board Annual Meeting, January 2014, Washington D.C.
[94] Life Cycle Cost of Electric Vehicle Fast Charging and Battery Swapping Stations, INFORMS Annual
Meeting, October 2013, Minneapolis, MN.
[95] Consumer Preferences for Hybrid and Electric Vehicles in China and the United States, INFORMS
Annual Meeting, October 2013, Minneapolis, MN (speaker: Helveston).
[96] Sensitivity of Vehicle Market Share Predictions to Alternative Discrete Choice Model
Specifications, ASME International Design Engineering Technical Conferences, August 2013,
Portland, OR (speaker: Haaf).
[97] Toward Understanding the Role of Interaction Effects in Visual Conjoint Analysis, ASME
International Design Engineering Technical Conferences, August 2013, Portland, OR (speaker:
Sylcott).
[98] Electric Vehicles: A Techno-Economic and Environmental Assessment of Costs, Benefits,
Challenges, and Strategies, Carnegie Mellon University (departmental seminar), April 26, 2013,
Pittsburgh, PA.
[99] Cost-Effectiveness of PHEV Battery Capacity and Charging Infrastructure, Transportation
Research Board of the National Academies, poster presentation, January 15, 2013, Washington,
D.C.
[100] Evaluation of the Effects of Thermal Management on Battery Life in PHEVs, Transportation
Research Board of the National Academies, poster presentation, January 15, 2013, Washington,
D.C. (speaker: Yuksel)
[101] U.S. Residential Charging Potential for Plug-in Vehicles, Transportation Research Board of the
National Academies, poster presentation, January 15, 2013, Washington, D.C. (speaker: Traut)
[102] Valuation of Plug-in Vehicle Life Cycle Air Emissions and Oil Displacement Benefits,
Transportation Research Board of the National Academies, poster presentation, January 15,
2013, Washington, D.C.
[103] Influence of Driving Patterns on Life Cycle Benefits of Hybrid and Plug-in Electric Vehicles,
International Mechanical Engineering Congress and Exposition, November 2012, Houston, TX.
(speaker: Karabasoglu)
[104] Optimal Combined Design and Control of Electrified Vehicles for Globally Minimum Life Cycle
Cost, International Mechanical Engineering Congress and Exposition, November 2012, Houston,
TX. (speaker: Karabasoglu)
[105] Supercapacitor-Battery System Design and Control for Plug-in Electric Vehicles and Life Cycle
Economic and Environmental Implications, International Mechanical Engineering Congress and
Exposition, November 2012, Houston, TX. (speaker: Karabasoglu)
[106] Controlled Plug-in Vehicle Charging in High Wind Penetration Scenarios, INFORMS Annual
Meeting, Oct 2012, Phoenix, AZ. (speaker: Weis)
A-45
-------
Peer Review Report
Work Assignment 3-05/4-05, Contract 68HE0C18C0001
[107] Globally Optimal Robust Design and Control of Plug-in Hybrid Electric Vehicles, INFORMS Annual
Meeting, Oct 2012, Phoenix, AZ. (speaker: Karabasoglu)
[108] Global Control Optimization of Supercapacitor-Battery Electric Vehicles, INFORMS Annual
Meeting, Oct 2012, Phoenix, AZ. (speaker: Karabasoglu)
[109] Driving Design: Modeling the Influence of Market Forces and Public Policy on Vehicle Design
Decisions, National Science Foundation Civil, Mechanical, and Manufacturing Innovation
Conference, July 23, 2012, Boston, MA.
[110] Development of a Simulation Model to Analyze the Effect of Thermal Management on Battery
Life, SAE World Congress, April 25, 2012, Detroit, Ml (speaker: Yuksel)
[111] Evaluation of the Effects of Thermal Management on Battery Life in Plug-in Hybrid Electric
Vehicles, The Battery Congress, April 23, 2012, Ann Arbor, Ml (speaker: Yuksel)
[112] Are Plug-in Vehicles Worth the Cost? Valuing Air Emissions and Oil Displacement Benefits in the
U.S., INFORMS Annual Meeting, Nov 2011, Charlotte, N.C.
[113] Using Economic Input-Output Life Cycle Assessment to Guide Sustainable Design, INFORMS
Annual Meeting, Nov 2011, Charlotte, N.C. (speaker: Traut)
[114] Optimal Design and Allocation of Electrified Vehicles and Dedicated Charging Infrastructure for
Minimum Greenhouse Gas Emissions, INFORMS Annual Meeting, Nov 2011, Charlotte, N.C.
(speaker: Traut)
[115] Minimizing the Integration Costs of Wind Using Curtailment and Electric Vehicle Charging, U.S.
Association for Energy Economics North American Conference, Oct 2011, Washington D.C.
(speaker: Weis)
[116] Robust Design for Profit Maximization under Uncertainty of Consumer Choice Model
Parameters Using the Delta Method, ASME International Design Engineering Technical
Conferences, Aug 2011, Washington D.C. (speaker: Heckmann)
[117 Using Economic Input-Output Life Cycle Assessment to Guide Sustainable Design, ASME
International Design Engineering Technical Conferences, Aug 31, 2011, Washington D.C.
[118] Are Plug-in Vehicles Worth the Cost?, ASME International Design Engineering Technical
Conferences, Aug 30, 2011, Washington D.C.
[119] A Perspective on Rebound Effects and Demand/Supply Equilibrium, 2011 Climate and Energy
Decision Making Workshop, June 2011, Washington D.C. (speaker: Hendrickson)
[120] NHTS Survey Day Driving Distance and Estimated Variability to inform Electric Vehicle Range
Design, Using National Household Travel Survey (HNTS) Data for Transportation Decision Making
Workshop, June 2011, Washington, D.C. (speaker: Traut)
[121] Costs and Benefits of Plug-in Vehicles, Carnegie Mellon Steinbrenner Media Fellowship, June
2011, Pittsburgh, PA.
[122] Optimal Design and Allocation of Electrified Vehicles and Dedicated Charging Infrastructure for
Minimum Greenhouse Gas Emissions, International Society of Industrial Ecology Conference,
June 2011, Berkeley, CA. (speaker: Traut)
[123] Techno-Economic Analysis of Lithium-Ion Batteries for Personal Vehicle Electrification, with
Apurba Sakti, Technology Management and Policy Conference, May 2011, State College, PA.
[124] Optimal Design and Allocation of Electrified Vehicles and Dedicated Charging Infrastructure for
Minimum Greenhouse Gas Emissions, Mascaro Center's Engineering Sustainability Conference
2011, April 2011, Pittsburgh, PA.
A-46
-------
Peer Review Report
Work Assignment 3-05/4-05, Contract 68HE0C18C0001
[125] Air Emissions and Oil Displacement Benefits from Plug-in Vehicles, policy brief presented to
members of the Congressional Budget Office, the Congressional Research Service, the Senate
Energy and Natural Resources Committee, the Senate Commerce, Science and Transportation
Committee, and members of the U.S. House of Representatives, March 13-14, 2012,
Washington, D.C.
[126] Techno-Economic Analysis of Lithium-Ion Batteries for Personal Electrification, with Apurba
Sakti. National Academies Transportation Research Board Annual Meeting, January 23-27, 2011,
Washington, D.C.
[127] Optimal Design and Allocation of Electrified Vehicles and Dedicated Charging Infrastructure for
Minimum Greenhouse Gas Emissions, with Elizabeth Traut, National Academies Transportation
Research Board Annual Meeting, January 23-27, 2011, Washington D.C.
[128] Driving Design: Modeling the Influence of Market Forces and Public Policy on Vehicle Design
Decisions, Civil, Mechanical and Manufacturing Innovation Conference, January 4-7, 2011,
Atlanta, GA.
[129] Material Use, Infrastructure Change, and Environmental Impacts of Alternative Fuels and
Vehicles, Civil, Mechanical and Manufacturing Innovation Conference, January 4-7, 2011,
Atlanta, GA.
[130] Are Plug-in Vehicles Worth the Cost?, INFORMS Annual Meeting, Nov 9, 2010, Austin, TX.
[131] A MINLP Model for Global Optimization of Plug-in Vehicle Design and Allocation to Minimize Life
Cycle Cost and Greenhouse Gas Emissions, INFORMS Annual Meeting, Nov 8, 2010, Austin, TX.
[132] Why Your Plug-in Hybrid Electric Vehicle with a 40-mile Battery May Only Go 25, with Orkun
Karabasoglu, INFORMS Annual Meeting, Nov 9, 2010, Austin, TX.
[133] Optimal Plug-in Hybrid Electric Vehicle Design and Allocation for Minimum Life Cycle Cost,
Petroleum Consumption, and Greenhouse Gas Emissions, ASME International Design
Engineering Technical Conferences, Advanced Vehicle and Tire Technology Conference, Aug
2010, Montreal, CA.
[134] A MINLP Model for Global Optimization of Plug-in Hybrid Electric Vehicle Design and Allocation
to Minimize Life Cycle Greenhouse Gas Emissions, ASME International Design Engineering
Technical Conferences, Design Automation Conference, Aug 2010, Montreal, CA.
[135] MINLP Global Optimization of Plug-in Hybrid Vehicle Design and Allocation for Minimum Cost
and GHG Emissions, INFORMS Conference on Energy, Sustainability and Climate Change, Feb
2010, Gainesville, FL, USA.
[136] Do More Batteries Make a Plug-in Better? Economic and Environmental Analysis of Plug-in
Hybrid Electric Vehicles, Society of Automotive Engineers Government-Industry Meeting, Jan
2010, Washington D.C., USA.
[137] Do More Batteries Make a Plug-in Better? Economic and Environmental Analysis of Plug-in
Hybrid Electric Vehicles, Institute for Operations Research and the Management Sciences
(INFORMS) Annual Meeting, Oct 11, 2009, San Diego, CA, USA.
[138] Optimal Plug-in Hybrid Electric Vehicle Design and Allocation for Diverse Driving Patterns, ASME
Design Engineering Technical Conferences, Sept 1 2009, San Diego, CA, USA.
[139] Do More Batteries Make a Plug-in Better? Economic and Environmental Analysis of Plug-in
Hybrid Electric Vehicles, MIT Engineering Systems Symposium, June 17, 2009, Boston, MA, USA.
A-47
-------
Peer Review Report
Work Assignment 3-05/4-05, Contract 68HE0C18C0001
[140] A Structural Analysis of Vehicle Design Responses to Corporate Average Fuel Economy (CAFE)
Policy, poster presentation, MIT Engineering Systems Symposium, June 16, 2009, Boston, MA,
USA.
[141] Economic, Environmental, and Security Implications of Plug-in Hybrid Electric Vehicles,
Presented to staff members of the House Energy and Commerce Committee, the House
Committee on Science and Technology, the Select Committee on Energy Independence and
Global Warming, the Congressional Research Service, and offices of U.S. House and Senate
members, April 16 and April 20, 2009, Washington D.C., USA.
[142] A Structural Analysis of Vehicle Design Responses to CAFE Policy, National Academies
Transportation Research Board Annual Meeting, January 2009, Washington DC, USA.
[143] Impact of Battery Weight and Charging Patterns on the Economic and Environmental Benefits of
Plug-in Hybrid Vehicles, National Academies Transportation Research Board Annual Meeting,
January 2009, Washington DC, USA.
[144] A Structural Analysis of Vehicle Design Responses to CAFE Policy, Institute for Operations
Research and Management Sciences (INFORMS) Annual Meeting, November 2008, Washington,
DC, USA.
[145] A Deterministic Lagrangian-Based Global Optimization Approach for Decomposable Nonconvex
Mixed-Integer Problems, Institute for Operations Research and Management Sciences
(INFORMS) Annual Meeting, November 2008, Washington, DC, USA.
[146] Optimal Product Design in a Competitive Market, Institute for Operations Research and
Management Sciences (INFORMS) Annual Meeting, November 2008, Washington, DC, USA.
[147] Consumer Heterogeneity and Channel Structures in Optimal Product Design, Institute for
Operations Research and Management Sciences (INFORMS) Annual Meeting, November 2008,
Washington, DC, USA.
[148] Global Optimization of the Joint Product Family Platform Selection and Design Problem, Institute
for Operations Research and Management Sciences (INFORMS) Annual Meeting, November
2008, Washington, DC, USA.
[149] Impact of Battery Weight and Charging Patterns on Plug-in Hybrid Vehicles, ASME International
Design Engineering Technical Conferences, August 2008, New York, NY, USA.
[150] Should Designers Worry about Market Systems?, ASME International Design Engineering
Technical Conferences, August 2008, New York, NY, USA.
[151] Optimal Product Design under Price Competition, ASME International Design Engineering
Technical Conferences, August 2008, New York, NY, USA.
[152] A Deterministic Lagrangian-based Global Optimization Approach for Large-Scale Decomposable
Problems, ASME International Design Engineering Technical Conferences, August 2008, New
York, NY, USA.
[153] Wiki-Based Learning in the Mechanical Engineering Classroom, American Society for Engineering
Education (ASEE) Annual Conference and Exposition, June 2008, Pittsburgh, PA, USA.
[154] Applications of Lagrangian Branch and Cut for Hierarchical Engineering Systems, Institute For
Operations Research and the Management Sciences (INFORMS) Annual Meeting, November
2007, Seattle, WA, USA.
[155] A Quantitative Approach to Achieving Optimal Balance between Product Variety and
Manufacturability, Institute For Operations Research and the Management Sciences (INFORMS)
Annual Meeting, November 2007, Seattle, WA, USA.
A-48
-------
Peer Review Report
Work Assignment 3-05/4-05, Contract 68HE0C18C0001
[156] Product Line Design Optimization for Heterogeneous Markets, Institute For Operations Research
and the Management Sciences (INFORMS) Annual Meeting, November 2007, Seattle, WA, USA.
[157] Diagonal Quadratic Approximation for Parallelization of Analytical Target Cascading, ASME
International Design Engineering Technical Conferences, September 2007, Las Vegas, NV, USA.
[158] A Game Theoretic Approach to Finding Market Equilibria for Automotive Design under
Environmental Regulation, ASME International Design Engineering Technical Conferences,
September 2007, Las Vegas, NV, USA.
[159] Design Optimization of a Laptop Computer using Aggregate and Mixed Logit Demand Models
with Consumer Survey Data, ASME International Design Engineering Technical Conferences,
September 2007, Las Vegas, NV, USA.
[160] Measurement of Headlight Form Preference using Choice-Based Conjoint Analysis, ASME
International Design Engineering Technical Conferences, September 2007, Las Vegas, NV, USA.
[161] An Extension of the Commonality Index for Product Family Optimization, ASME International
Design Engineering Technical Conferences, September 2007, Las Vegas, NV, USA.
[162] A Single-Stage Gradient-Based Approach for Solving the Joint Product Family Platform Selection
and Design Problem using Decomposition, ASME International Design Engineering Technical
Conferences, September 2007, Las Vegas, NV, USA.
[163] A Decomposed Genetic Algorithm for Solving the Joint Product Family Optimization Problem,
AIAA Multidisciplinary Design Optimization Specialists Conference, April, Honolulu, Hawaii, USA.
[164] Balancing Marketability and Manufacturability in Product Line Design Optimization, Institute For
Operations Research and the Management Sciences (INFORMS) Annual Conference, November
2006, Pittsburgh, PA.
[165] The Impact of Environmental Policy on Profit-Driven Vehicle Design Optimization, Institute For
Operations Research and the Management Sciences (INFORMS) Annual Conference, November
2006, Pittsburgh, PA.
[166] Analytical Target Cascading using Branch and Bound for Mixed Integer Nonlinear Programming,
ASME International Design Engineering Technical Conferences, September 2006, Philadelphia,
PA, USA.
[167] Balancing Marketing and Manufacturing Objectives in Product Line Design, ASME International
Design Engineering Technical Conferences, September 2005, Long Beach, CA, USA.
[168] An Efficient Weighting Update Method to Achieve Acceptable Consistency Deviation in
Analytical Target Cascading, ASME International Design Engineering Technical Conferences,
September 2004, Salt Lake City, Utah, USA.
[169] An Efficient Weighting Update Method to Achieve Acceptable Consistency Deviation in
Analytical Target Cascading, Automotive Research Center Conference, May 2004, Ann Arbor, Ml,
USA.
[170] An Optimal Marketing and Engineering Design Model for Product Development Using Analytical
Target Cascading, Tools and Methods of Competitive Engineering Conference, April 2004,
Lausanne, Switzerland.
[171] A Simulation-based Vehicle Design Strategy for Requirements Validation, Automotive Research
Center Conference, May 12, 2003, Chrysler Headquarters, Auburn Hills, Ml, USA.
[172] A Study of Emission Policy Effects on Optimal Vehicle Design Decisions, ASME Design
Engineering Technical Conferences, September 2003, Chicago, IL, USA.
A-49
-------
Peer Review Report
Work Assignment 3-05/4-05, Contract 68HE0C18C0001
VIDEOS
[1] Did You Know That Extreme Weather Affects an Electric Car's Range? Inside Science, Nov 2015
[2] Electric Vehicle Benefits & Costs in the U.S., Carnegie Mellon University, June 2015
[3] Electric Vehicle Adoption Potential in the U.S., Carnegie Mellon University, June 2015
[4] Is Manufacturing in China a Wise Decision for a Small, Innovative US Company? Carnegie Mellon
University Scott Institute for Energy Innovation, June 2015
[5] Jeremy Michalek: Vehicle Electrification, Carnegie Mellon University, April 2015
[6] CMU Energy Interview: Jeremy Michalek, Carnegie Mellon University, July 2013.
[7] Do Hybrid and Plug-in Cars Really Save the Environment? Carnegie Mellon University Scott
Institute for Energy Innovation, Nov 2013
[8] CMU Energy Presentation: Plug-in Vehicles, Carnegie Mellon University, Sep 2012
RADIO AND PODCASTS
[9] On Point, National Public Radio, WBUR: America's Electric Vehicle Future, June 2021
[10] Energy Bite, WESA-Pittsburgh: Are electric vehicles the right choice for you? July 2015
[11] Energy Bite, WESA-Pittsburgh: If I buy a plug-in electric vehicle, how much will it help the
environment? July 2015
[12] Energy Bite, WESA-Pittsburgh: Will everyone be driving a plug-in electric vehicle in the future?
July 2015
[13] Energy Bite, WESA-Pittsburgh: If I buy a plug-in electric vehicle, when is the most efficient time
to charge it? July 2015
[14] Energy Bite, WESA-Pittsburgh: Will automated vehicles help save energy? July 2015
[15] Carnegie Mellon Engineering: What do American and Chinese consumers want in an electric
vehicle? May 2015
[16] Carnegie Mellon Engineering: How does climate affect your electric car's performance? April
2015
PRESS RELEASES
[1] Should Uber and Lyft be electrifying more vehicles? Feb 19, 2021.
[2] When Uber and Lyft enter cities, vehicle ownership increases, Jan 6, 2021.
[3] Want an Electric Vehicle? The One You Should Buy Depends on Where You Live, Apr 20, 2016.
[4] Federal Policy Reverses Benefits of Alternative Fuel Vehicles, Mar 3, 2016.
[5] Researchers Show Coal Retirement Needed for Electric Vehicles to Reduce Air Pollution, Feb 23,
2016.
[6] Charging Electric Vehicles at Night Can Cause More Harm Than Good, Says CMU Study, Feb 22,
2016.
[7] Which Vehicle Holds Smallest Carbon Footprint? Electrics in Some Regions, Hybrids in Others,
Says Carnegie Mellon Study, July 16, 2015
[8] Carnegie Mellon Study Shows Electric Vehicle Range and Emissions Vary With Climate, Feb 24,
2015
[9] CMU Study Finds Chinese Consumers May Adopt Electric Vehicles First, Impacting Auto Market,
Feb 16, 2015.
A-50
-------
Peer Review Report
Work Assignment 3-05/4-05, Contract 68HE0C18C0001
[10] Big Factories Won't Solve High Cost of Electric Vehicles, Carnegie Mellon Researchers Say, Oct
21, 2014
[11] Carnegie Mellon Study Says Electric Vehicles Could Be Cheaper to Recharge if Electricity
Providers Control Charging Speeds, Jan 23, 2014
[12] Carnegie Mellon Researchers Find Consumers Choose More Efficient Light Bulbs when Energy
Costs are Labeled, Jan 10, 2014.
[13] Carnegie Mellon Researchers Find Limited Residential Parking a Barrier to Electric Vehicle
Adoption, Nov 11, 2013.
[14] Carnegie Mellon Researchers Report Hybrid Cars are Greener for City Drivers, June 17, 2013.
[15] Fiscal Cliff Bill Tax Credits for Electric Vehicle Chargers Found Not Cost Effective, Jan 7, 2013.
[16] Carnegie Mellon Study Finds Benefits of Plug-in Vehicles Depend on Battery Size, Sept 26, 2011.
OP-EDS
[1] MarketWatch: I'm an EV expert, and I'm skeptical about how quickly electric cars will go
mainstream in the U.S., June 2021.
[2] Michalek, J., "Problems with the fuel economy rollback," Op-Ed, The Hill, Sept 26, 2018.
SELECTED MEDIA STORIES
[3] On Point, National Public Radio, WBUR: America's Electric Vehicle Future, June 2021
[4] The New York Times: How green are electric vehicles? Mar 2, 2021.
[5] The Hill: Buttigieg sets goals for electric, automated freight vehicles, Feb 17, 2021.
[6] Scientific American: Can California Eliminate Gas Cars? Nov 8, 2017.
[7] The New York Times: Climate Protection Advocates Fear a Rollback of Emissions Standards, Nov
14, 2016.
[8] Science: The Best - and Worst - Places to Drive Your Electric Car, Feb 20, 2015.
[9] National Public Radio On-Point: A Dire Climate Change Report, and the Possibility for Change,
Nov 6, 2014.
[10] The New York Times: EVs Could Be a Key Part of a Changing Electricity Grid, Jan 23, 2014.
[11] The Daily Beast: Electric Vehicles May Be the Green Car of the Future, But Hybrids Are the Green
Car of the Present, Jan 24, 2013.
[12] The Washington Post: Hybrid, electric or gas: What's a car buyer interested in the environment
to do? March 19, 2012.
[13] Science: Editors' Choice: The Pros (and Cons) of Plugging In, V331, Feb 18 2011.
ALL MEDIA STORIES
[1] WHYY: SEPTA's cracking battery buses raise questions about the future of electric transit, July
15, 2021.
[2] Vice (Motherboard): Uber and Lyfts can be electrified now, but drivers aren't so sure, June 28,
2021.
[3] On Point, National Public Radio, WBUR: America's Electric Vehicle Future, June 17, 2021
[4] IEEE Spectrum: How is this a good idea? EV battery swapping, May 13, 2021.
[5] The Seattle Times: Are electric vehicles really as green as advertised? March 6, 2021.
A-51
-------
Peer Review Report
Work Assignment 3-05/4-05, Contract 68HE0C18C0001
[6] The New York Times: How green are electric vehicles? Mar 2, 2021.
[7] Green Car Congress: CMU study finds taxes on emissions would result in more rapid
electrification by ridesharing companies, Feb 21, 2021.
[8] The Hill: Buttigieg sets goals for electric, automated freight vehicles, Feb 17, 2021.
[9] Shift: Oh, great - Uber and Lyft are actually increasing car ownership in US cities, Jan 13, 2021.
[10] Ladders: Ride-sharing apps had this surprising effect on cities, Jan 11, 2021.
[11] StreetsBlog USA: Study: e-taxis increase private car ownership in many cities, Jan 8, 2021.
[12] Inverse: Uber and Lyft are shaping cities in one deeply unexpected way, Jan 7, 2021.
[13] Vice: Uber and Lyft say they reduce car ownership, but that might not be true, Jan 7, 2021.
[14] New Scientist: Uber and Lyft operating in US cities linked to rises in car ownership, Jan 6, 2021.
[15] Automotive News: Electrification expert Jeremy Michalek says California policies are key to US
adoption of EVs, Dec 20, 2020.
[16] The Globe and Mail: Automakers use solar panels to boost electric range, aiming to replace
charging stations altogether, Dec 7, 2020.
[17] Forbes: Will a solar-powered mass market car ever be possible? Nov 30, 2020.
[18] E&T Magazine: The EV lifecycle conundrum, July 20, 2020.
[19] Digital Trends: Inside the light-speed race to build a solar-powered commuter car, Mar 17, 2020.
[20] Watershed Sentinel: Electric vehicles: the good, the bad, the context, Feb 20, 2020.
[21] WIRED: Want a Tax Credit for Buying an Electric Vehicle? Move Fast, Dec 18, 2019.
[22] Quartz: Sunshine will cover your daily commute, Nov 13, 2019.
[23] CarbonBrief: Factcheck: How electric vehicles help to tackle climate change, May 13, 2019.
[24] Quartz: The physics of why we don't have solar powered cars, Oct 15, 2018.
[25] WIRED: California says "nope" to the EPA's car emissions rules, Oct 4, 2018.
[26] The Hill: Problems with the fuel economy rollback, Sept 26, 2018.
[27] PublicSource: Pittsburgh leaders, citizens attack Trump's proposed rollback of cleaner vehicle
emissions standards, Sept 26, 2018.
[28] Pittsburth Tribune Review: Pittsburgh speakers rally against Trump plan to weaken fuel
standards, Sep 26, 2018.
[29] PublicSource: Will Trump nix Obama-era emissions requirements? You can weight in right in
Pittsburgh, Sep 14, 2018.
[30] Earther: Scott Pruitt's clean car rollback could have global consequences, Apr 4, 2018.
[31] GLT89.1FM: Searching for clues into Rivian's electric vehicle future, Jan 9, 2018.
[32] SiriusXM channel 11, Knowledge@Wharton: Will losing the electric vehicle tax credit kill plug-in
cars? Dec 6, 2017.
[33] San Francisco Chronicle: For drivers without garages, charging a big barrier to electric cars, Nov
24, 2017.
[34] Scientific American: Can California Eliminate Gas Cars? Nov 8, 2017.
[35] WIRED: Congress' Plan to Kill the Electric Car Tax Credit Could Kill Electric Cars, Nov 4, 2017.
[36] La Presse: Environment - Un Choix Ecologique Pour L'avenir, Aug 26, 2017.
[37] La Presse: Voiture Electrique - Le Dilemme Environnemental, Aug 26, 2017.
[38] E&E News Climate Wire: Tesla Logging 1,800 Model 3 Reservations Daily, Aug 3, 2017.
A-52
-------
Peer Review Report
Work Assignment 3-05/4-05, Contract 68HE0C18C0001
[39] Pittsburgh Post-Gazette: Electricy Pennsylvania transportation system - Here's how the state
should spend its share of the Volkswagen settlement, July 28, 2017.
[40] MIT Technology Review: Tesla's Model 3 is a Long Way from Elon Musk's Grand Goal, July 28,
2017.
[41] Fortune: Could the U.S. Ever Ban Gas-Powered Cars Like France Wants to Do?, July 6, 2017.
[42] E21: Not Yet Time for Tesla, July 5, 2017.
[43] WIRED: 2017 will be the year Tesla reigns supreme - or finally flops, Dec 30, 2016.
[44] ClimateWire: 'We need more customers!' Calif, revisits clean car rule, Dec 16, 2016.
[45] The New York Times: Climate Protection Advocates Fear a Rollback of Emissions Standards, Nov
14, 2016.
[46] Vox: Millions of used electric car batteries will help store energy for the grid. Maybe., Aug 29,
2016.
[47] NGT News: Does buying an electric vehicle make sense? Apr 27, 2016.
[48] Boston Globe: Study says 87% of our vehicles could be electric, Aug 16, 2016.
[49] Bloomberg: How many of our vehicles could be electric? How does 87% strike you?, Aug 15,
2016.
[50] Wards Auto: Sustainability, Innovation Driving Tesla CEO Musk, July 26, 2016.
[51] Clean Technica: Alt-fuel vehicle incentives could increase fuel consumption & emissions in short
term, May 18, 2016.
[52] Green Car Reports: Buying an electric car can increase fuel use allowed by CAFE rules, May 11,
2016.
[53] The Star: Tesla becomes the first 488km electric car, Apr 29, 2016.
[54] Autoblog: This map reveals the cleanest vehicles based on location, Apr 28, 2016.
[55] The New York Times: Demand for the new Tesla is wild, but limited to tech fans, Apr 12, 2016
(AP story, 100s of outlets).
[56] Truthdig: Day-to-Day Human Activity Determines the Finer Effects of Climate Change, Mar 31,
2016.
[57] MIT Technology Review: Here's Why You Might be an Electric Car Owner a Decade from Now,
Mar 30, 2016.
[58] Phys.org: Federal Policy Reverses Benefits of Alternative Fuel Vehicles, Mar 9, 2016.
[59] Chemical & Engineering News: Incentive for Electric Cars Increases C02 Pollution, Mar 2, 2016.
[60] Hybrid Cars: Study: Overnight Charging in Coal-Intensive Grids May Be Worse For The
Environment, Feb 24, 2016.
[61] Tech Times: Charging Electric Vehicles at Night Leads to Higher Greenhouse Gas Emissions, Feb
24, 2016.
[62] Gas2: Charging Your Electric Vehicle at Night Could Be Bad for the Environment, Feb 24, 2016.
[63] Ars Technica: Could more electric cars mean greater fleet emissions and fuel consumption? Feb
19, 2016.
[64] Atlanta Journal Constitution: Electric Cars Going Mainstream Sooner, Jan 27, 2016.
[65] Eco-Business: Boosting EVs with Cloud Power, Dec 17, 2015.
[66] Inside Science TV: Extreme Weather Affects an Electric Car's Range, Dec 13, 2015.
[67 SFGate: Could Climate Impact Electric Car Range? Nov 12, 2015.
A-53
-------
Peer Review Report
Work Assignment 3-05/4-05, Contract 68HE0C18C0001
[68] Futurity: Prius or Leaf? Carbon Footprint Hinges on Location, July 20, 2015.
[69] Bloomberg: The $5 Billon Race to Build a Better Battery, April 14, 2015.
[70] CMU Engineering Podcast: How does climate affect your electric car's performance?, April 10,
2015.
[71] WIRED: Electric Cars can Explain our Highway Funding Fiasco, Apr 8, 2015.
[72] Takepart: Find Out if You Live in the Best - or Worst - State for Electric Cars, Mar 11, 2015.
[73] NDTV: Weather Conditions Affect Mileage of Electric Cars, Mar 4, 2015.
[74] Green Car Reports: Chinese Car Buyers May Be More Open to Electric Cars than Americans:
Study, Mar 3, 2015.
[75] Ecomento.com: Electric Cars Appeal to More Chinese than American Buyers, Mar 3, 2015.
[76] The Economic Times: Efficiency of Electric Cars Depends on Weather, Mar 2, 2015.
[77] Fast Company: Where It's Best and Worst to Drive an Electric Vehicle, Mar 2, 2015.
[78] Mother Nature Network: Cold Weather Kills Electric Car Range, Feb 27, 2015.
[79] Charged Electric Vehicles Magazine: New Study: Chinese More Receptive to EVs than Americans,
Feb 26, 2015.
[80] Laboratory Equipment: Location Influences Electric Car Range, Emissions, Feb 26, 2015.
[81] Product Design & Development: EV Range Depends on Where You Live, Feb 25, 2015.
[82] Engineering.com: China May Adopt EV Before Other Markets, Feb 23, 3015.
[83] Science: The Best - and Worst - Places to Drive Your Electric Car, Feb 20, 2015.
[84] Design News: Chevy Bolt Announcement Signals Battery Improvement, Feb 17, 2015.
[85] Discovery News: Electric vs. Fuel Cell Cars: 'Green' Autos Explained, Jan 30, 2015.
[86] Live Science: Electric vs. Fuel Cell Vehicles: 'Green' Auto Tech Explained, Jan 28, 2015.
[87] WIRED: Chevy Could Beat Tesla to Building the First Mainstream Electric Car, Jan 13, 2015.
[88] CQ Roll Call - Policy Pulse: Looking Into the Future on Electric Vehicle Adoption, Jan 12, 2015.
[89] Pittsburgh Tribune-Review: Alcoa, Israeli Company Collaborate on Aluminum-Air Battery, Jan 10,
2015.
[90] Design News: Chevy Bolt Announcement Signals Battery Improvement, Feb 17, 2015.
[91] The Engineer: Improved Battery Design is Key to Cost Savings, Nov 4, 2014.
[92] New York Times Automobiles: Wheelies: The Real Deal Jeap Edition, Oct 24, 2014.
[93] Network World: Carnegie Mellon: Bigger May Not be Better with Battery Makers, Oct 21, 2014.
[94] Automotive News: Pulling the plug: EV battery module exchange put to test, Oct 20, 2014.
[95] Pittsburgh Post-Gazette: Alternative vehicle fuels a growing niche market, April 29, 2014.
[96] Pittsburgh Post-Gazette, "Shedding light on true cost of bulbs," Feb 7, 2014.
[97] WESA, Pittsburgh's NPR News Station: New Method of Charging Electric Cars Could Bring
Savings, Feb 3, 2014.
[98] The Daily Fusion: EVs Could Be Recharged Cheaper With Controlled Charging Speeds, Jan 27,
2014.
[99] The Pittsburgh Tribune-Review: Plugging electric vehicles in at cheapest time could save on
power bills, CMU finds, Jan 24, 2014.
[100] The New York Times: EVs Could Be a Key Part of a Changing Electricity Grid, Jan 23, 2014.
A-54
-------
Peer Review Report
Work Assignment 3-05/4-05, Contract 68HE0C18C0001
[101] EV World: CMU Study Sees Parking as Next Electric Car Barrier, Jan 14, 2014.
[102] Design News, "Will parking problems slow the rise of electric vehicles?" January 14, 2014.
[103] Futurity, "Labels sway consumers to pick 'green' bulbs," January 13, 2014.
[104] Earth Techling, "Good labeling might increase green light bulb sales," January 2014.
[105] Bloomberg Businessweek: Toyota Sees Hydrogen Car Fill-Ups at Dealers to Trash Dumps, Jan 7,
2014.
[106] Allegheny Front: Charging a Challenge for Electric Car Owners, Nov 22, 2013.
[107] WESA 90.5FM: Lack of Parking May Be One of the Electric Car's Biggest Hurdles, Nov 18, 2013.
[108] The Car Connection: Electric Car Adoption Lagging Because Drivers Have Nowhere to Park &
Charge Them, Nov 13, 2013.
[109] The New York Times: Electric Cars Give Rise to a Recharging Industry, Nov 12, 2013.
[110] The Daily Fusion: Insufficient Parking Space Hinders Electric Vehicle Adoption, Nov 12, 2013.
[111] Green Car Congress: CMU study finds limited dedicated residential parking and charging a
significant barrier to mainstream EV adoption, Nov 12, 2013.
[112] Autoblog Green: In city driving, hybrids really outshine the competition, June 22, 2013.
[113] EV World: Hybrids better choice for city drivers, CMU study finds, June 18, 2013.
[114] Pittsburgh Post Gazette: Business news briefs: Study says hybrid, electric vehicles not always
beneficial, June 18, 2013.
[115] ASME.org, "The Expo Advantage," March 2013.
[116] Pocono Record: Hybrids Deserve Better, Feb 18, 2013.
[117] The Allegheny Front on WESA, Pittsburgh's NPR News Station: "CMU studies electric vs. hybrid
cars," Feb 9, 2013.
[118] The Daily Beast: "Electric vehicles may be the green car of the future, but hybrids are the green
car of the present," Jan 24, 2013.
[119] CBS-KDKA Ch.2: "CMU engineering students developing new consumer products," Dec 10, 2012.
[120] NBC-WTAE Ch.4: "Young engineers showcase prototypes at CMU expo ," Dec 10, 2012.
[121] Pittsburgh Post-Gazette: "Innovation 101: CMU students iron out details to improve consumer
products," Dec 10, 2012.
[122] Green Car Congress: "CMU/Ford study assesses optimal mix of conventional, hybrid, plug-in
hybrid and electric vehicles for minimizing GHG and cost," Oct 31, 2012.
[123] Green Car Congress: "CMU study finds small battery PHEVs and gasoline hybrids the least-cost
policy solution to reducing gasoline consumption," Oct 29, 2012.
[124] Design News: "Could Pure Electrics Emit More than Hybrids?" Nov 15, 2012.
[125] The Street: "Tesla Motors pollutes today for a greener tomorrow," Nov 7, 2012.
[126] Edmunds: "California ZEV mandate - Would a gas tax be better?" Oct 17, 2012.
[127] Automotive World: "US: PHEVs give better subsidy yields than EVs - study", July 26, 2012.
[128] CBS-KDKA Ch.2: "The Sunday Business Page," Interview with J. Michalek, P. White and W. Shim,
May 13, 2012.
[129] CBS-WTAE Ch. 2 News: "CMU engineering students present inventions," May 7, 2012.
[130] ABC-WTAE Ch. 4 Action News: "CMU students showcase designs at senior design expo," May 7,
2012.
A-55
-------
Peer Review Report
Work Assignment 3-05/4-05, Contract 68HE0C18C0001
[131] WPXI-Pittsburgh Ch. 11 Nightly News: "CMU students show off their new and improved
products/' May 7, 2012.
[132] The Washington Post: "Even hybrid and plug-in cars may not really be green/' March 20, 2012.
[133] The Washington Post: "Hybrid, electric or gas: What's a car buyer interested in the environment
to do?" March 19, 2012.
[134] Autoblog Green: "Why plug-in vehicles with "small" battery packs should be subsidized in a big
way/' Oct 6, 2011.
[135] The Truth About Cars: "Report knocks 'big battery' plug-in subsidies, will the DOE notice?" Sept
28, 2011.
[136] TheGreenCarWebsite.co.uk: "Are small batteries the future for hybrid cars," Sept 28, 2011.
[137] Greenwire: "Incentives should target hybrids, not pure EVs - report," E&E Publishing,
Washington D.C., Sept 28, 2011.
[138] K-Car Network (2011) l$l, Sept 27,
2011.
[139] Green Car Congress: "CMU study finds that HEVs and PHEVs with small battery packs offer more
emissions and oil displacement benefits per dollar spent than large pack PHEVs and BEVs; policy
implications," Sept 27, 2011.
[140] Alt Energy Stocks: (2011) "Plug-in vehicles have been weighed in the balance and found
wanting," Sept 27, 2011.
[141] HybridCars (2011) "Carnegie Mellon study says hybrids are better target for government
investment than EVs," Sept 26, 2011.
[142] Bloomberg Businessweek: "U.S. Battery, Plug-in Car Push Costs Exceed Rewards, New Study
Says," Sept 26, 2011.
[143] EV World: "The Case for Small Electric Hybrids," Dec 14, 2010.
[144] Green Car Congress: "CMU Study Explores Optimizing PHEV Design and Allocation to Minimize
Life Cycle Cost, Petroleum Consumption, and GHG Emissions," Dec 6, 2010.
[145] The Wall Street Journal, "High Battery Cost Curbs Electric Cars," Oct 17, 2010
[146] MSNBC: "Toyota's Straight Talk on Plug-in Vehicles," July 12 2010.
[147] IEC Engineering and Technology Magazine: "Batteries Struggle for Greater Power," n9, June 16,
2010.
[148] Minyanville: "Welcome Chevy Volt, GM's Next Flop," April 1, 2010.
[149] IEEE Spectrum: "Loser: Why the Chevy Volt will Fizzle," January 2010.
[150] CNET News: "Q&A: Will the Chevy Volt make money for GM?," October 30, 2009.
[151] Pittsburgh Tribune-Review: "Newsmaker: Jeremy J. Michalek," July 4, 2009.
[152] Nippon Television Network (2009) "The Future of the Auto Industry," Nightly News, April 29,
2009.
[153] IEEE Spectrum: "Why Plug-Ins will Make (Dollars and) Sense," March 2009.
[154] IEEE Spectrum: "Why the Chevy Volt Doesn't Add Up," March 9, 2009.
[155] Automobile Magazine: "Carnegie Mellon Study Finds High Capacity Plug-In Hybrids Cost
Inefficient," March 5, 2009.
[156] CNNMoney.com Fortune Magazine: "Taking the Charge out of Chevy's Volt: A new study casts
doubt on the benefits of the automaker's much-hyped plug-in hybrid," March 3, 2009.
A-56
-------
Peer Review Report
Work Assignment 3-05/4-05, Contract 68HE0C18C0001
[157] U.S. News and World Report "Study: Chevy Volt 'Not Cost Effective in Any Scenario/' March 2,
2009.
[158] Bloomberg News: "GM Volt Cost May Limit Value to Drivers, Study Finds," February 26, 2009.
[159] MIT Technology Review: "Toyota to Deliver Plug-In Hybrids: The new Prius is designed so that its
battery pack can be swapped out for a plug-in lithium-ion battery," January 13, 2009.
[160] Pittsburgh Tribune-Review: "CMU Engineers Innovation at Student Exposition," December 8,
2008.
[161] CBS-KDKA-Pittsburgh Ch.2 News: "CMU Students Show Off Inventions," December 8, 2008.
[162] WPXI-Pittsburgh Ch. 11 News: "CMU Students Host Innovative Product Expo," December 8,
2008.
[163] Pittsburgh Tribune-Review: "Newsmaker: Jeremy J. Michalek," Jan 30, 2008.
A-57
-------
-------
Peer Review Report
Work Assignment 3-05/4-05, Contract 68HE0C18C0001
Richard A. Rykowski
EDUCATION
Bachelor Degree in Chemical Engineering, University of Michigan, 1975
Master's Degree in Chemical Engineering, University of Michigan, 1977
Business, 30 credits towards MBA, University of Michigan, 1987-1992
EMPLOYMENT
Independent engineering contractor to Environmental Defense Fund, 2016 to present
U.S. Environmental Protection Agency, Ann Arbor, Michigan 1977-1993, 1996-2010
15 years as a Senior Technical Advisor in the Office of Transportation Air Quality, focusing on
technological and economic analysis used in regulatory development
Lead architect for development of OMEGA, used in support of the EPA GHG standards for
passenger cars and light trucks for model years 2012-2016 and 2017-2025
Co-team leader for development of Tier 2 standards for light-duty vehicles, 10 ppm gasoline sulfur
standard, 10 ppm on-road and non-road diesel fuel sulfur standards
Project manager for reformulated gasoline standards, gasoline RVP standards and the initial fine
PM standards for cars and light trucks and heavy-duty vehicles, as well as later standards for fine
PM emissions from heavy-duty vehicles and nonroad equipment
Air Improvement Resource, 1993-1996
Chemical engineering consultant, clients included: ARCO Chemical, Motor Vehicle Manufacturers
Association, General Motors, Engine Manufacturers Association
PUBLICATIONS
Rykowski, Richard, "The Benefits of Protective Advanced Clean Car Standards in Colorado An
Examination of Cost Savings, Greenhouse Gas Emission Reductions, and Health Benefits", for EDF,
May 2018, The Benefits of Protective Clean Car Standards CO.pdf (edf.org)
Rykowski, Richard A., "Appendix B, Review of the Agencies' Technical Analysis Supporting the SAFE
Vehicle NPRM, Supporting Report for Environmental Defense Fund Comment on Environmental
Protection Agency and National Highway Traffic Safety Administration Proposed Rule, The Safer
Affordable Fuel-Efficient (SAFE) Vehicles Rule for Model Years 2021-2026 Passenger Cars and
Light Trucks", 83 Fed. Reg. 42986 (Aug. 24, 2018), October 26, 2018
EDF3 Comment Appendix B Rvkowski.pdf
Rykowski, Richard A., "Colorado Zero Emission Vehicle Program Will Deliver Extensive Economic, Health
and Environmental Benefits, for EDF, July 2019, FINAL-EDF-Colorado-ZEV-report-2019.pdf .
Cackette, Tom and Rick Rykowski, "Technical Assessment of C02 Emission Reductions for Passenger
Vehicles in the Post-2025 Timeframe", February 2017,
final public white paper post 2026 co2 reductions2.27 clean.pdf (edf.org)
Numerous EPA reports, U.S. Environmental Protection Agency I US EPA, search on "Rykowski"
Authored numerous sections of the preambles and regulatory impact analyses associated with the
above-mentioned EPA rulemakings
A-59
-------
Peer Review Report
Work Assignment 3-05/4-05, Contract 68HE0C18C0001
AWARDS
1984 EPA Engineer of the Year, Numerous Gold and Silver Medals for Meritorious Service, received both
as an individual and as a member of a project team
A-60
-------
Peer Review Report
Work Assignment 3-05/4-05, Contract 68HE0C18C0001
Appendix B
Charge to Reviewers
B-l
-------
-------
Peer Review Report
Work Assignment 3-05/4-05, Contract 68HE0C18C0001
Technical Charge to External Peer Reviewers
Contract No. 68HE0C18C0001
Work Assignment 3-05
November 2021
External Peer Review of EPA's OMEGA Model
BACKGROUND
EPA created OMEGA version 1.0 to analyze new GHG standards for light-duty vehicles proposed in 2011.
In the period since the release of the initial version, there have been significant changes in the light duty
vehicle market including technological advancements and the introduction of new mobility services.
Advancements in battery electric vehicles (BEVs) with greater range, faster charging capability, and
expanded model availability, as well as potential synergies between BEVs, ride-hailing services and
autonomous driving are particularly relevant when considering pathways for greater levels of emissions
reduction in the future. OMEGA version 2.0 has been developed with these trends in mind. The model's
interaction between consumer and producer decisions allows a user to represent consumer responses
to these new vehicles and services. The model now also has been designed to have expanded capability
to model a wider range of GHG program options, which is especially important for the assessment of
policies that are designed to address future GHG reduction goals.
Like the prior releases, this latest version is intended primarily to be used as a tool to support regulatory
development by providing estimates of the effects of policy alternatives under consideration. These
effects include the costs associated with emissions-reducing technologies and the monetized effects
normally included in a societal benefit-cost analysis, as well as physical effects that include emissions
quantities, fuel consumption, and vehicle stock and usage.
ELEMENTS TO BE ADDRESSED
EPA's OMEGA estimates the technology necessary for vehicle manufacturers to meet a
specified greenhouse gas (GHG) standard. The model is contained in the enclosed computer
program and the documentation located at: https://omeqa2.readthedocs.io/en/2.0.0/. This
report illustrates the concepts and methodologies behind EPA's OMEGA model. No
independent data analysis will be required for this review.
Specifically, EPA is seeking your expert opinion on the concepts and methodologies upon which
the model relies and whether or not the model will execute these algorithms correctly.
In making your comments, please 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 should be sufficiently clear and detailed to allow a thorough
understanding by EPA or other parties familiar with the model.
B-3
-------
Peer Review Report
Work Assignment 3-05/4-05, Contract 68HE0C18C0001
Please review and comment on the following items:
1. The overall approach to the specified modeling purposes, the specific approaches chosen for
modeling individual modules, and the particular methodologies chosen to achieve that purpose.
2. The appropriateness and completeness of the contents of the sample input files. EPA is not
seeking comment on the particular values of the contents of the input files, which are samples
only.
NOTE: The types of information which can be input to the model point to both the flexibilities
and constraints of the model.
3. The accuracy and appropriateness of the model's conceptual algorithms and equations for
technology application, market impacts, and calculation of compliance.
4. Clarity, completeness, and accuracy of the documentation.
5. The congruence between the conceptual methodologies and the program execution. NOTE: This
can be verified by comparing spreadsheet calculations to the outputs provided by EPA or by
changing the input values and examining the results with good engineering judgment.
6. Clarity, completeness, and accuracy of the model's visualization output, in which the technology
application is displayed.
7. Recommendations for any functionalities beyond what EPA has described as "future work."
B-4
------- |