FEV Inc. Report "Light Duty Technology
Cost Analysis, Power-Split and P2 Hybrid
Electric Vehicle Case Studies" -
Response to Peer Reviewer Comments
&EPA
United States
Environmental Protection
Agency
-------
FEV Inc* Report "Light Duty Technology
Cost Analysis, Power-Split and P2 Hybrid
Electric Vehicle Case Studies" -
Response to Peer Reviewer Comments
Assessment and Standards Division
Office of Transportation and Air Quality
U.S. Environmental Protection Agency
and
FEV, Inc.
EPA Contract No. EP-C-07-069
Work Assignment No. 3-3
United States
Environmental Protection
Agency
EPA-420-R-11-017
November 2011
-------
This document provides responses to peer review comments on the FEV Report, "Light Duty
Technology Cost Analysis, Power-Split and P2 Hybrid Electric Vehicle Case Studies" The peer
review for the FEV Report was conducted by ICF International, and the documentation of their
peer review process and analysis of its findings have been published in: Peer Review of FEV
Inc. Report "Light Duty Technology Cost Analysis, Power-Split and P2 Hybrid Electric Vehicle
Case Studies", Final Report, March 31, 2011.
The original text of the ICF Report is used in this document for the purpose of directly
addressing each comment topic or comment theme with a response from EPA and FEV. Thus,
the EPA/FEV responses are incorporated into the body of the ICF comments summary, and are
identified by blue text and a response number (e.g. "Response #:...) In the interest of clarity
and context for the reader, the Introduction and Peer Review Process sections of the ICF
Report are included in this document, but the large Appendices containing peer reviewer
qualifications and the original review submittals are not.
A-1
-------
Introduction (from ICF Report)
The U.S. Environmental Protection Agency's (EPA's) Office of Transportation and Air Quality
(OTAQ) is developing programs to control greenhouse gas (GHG) emissions from light-duty
highway vehicles, which require an evaluation of the costs of technologies likely to be used to
meet any standards. EPA contracted with FEV Incorporated to perform this cost analysis
through tearing down vehicles, engines, and components, both with and without these
technologies, and evaluating, part by part, the observed differences in size, weight, materials,
machining steps and other cost-affecting parameters. Though complex and time-consuming,
EPA believes this approach has great potential for determining accurate technology costs, a
goal that is of paramount importance in the setting of appropriate GHG standards.
Although the teardown and analysis work is ongoing, FEV wrote a report detailing the
methodology it and its subcontractor are using to cost out technologies and describing the
results of the cost-out work to date.1 To assure that this work incorporates the highest quality
science, EPA contracted with ICF International (ICF) to determine appropriate independent peer
reviewers for the FEV report, "Light Duty Technology Cost Analysis, Power-Split and P2 Hybrid
Electric Vehicle (HEV) Case Studies" and document their feedback on the costing methodology
it presents. The reviewers selected were independent subject matter experts and their
reviews were conducted in compliance with EPA peer review guidelines.2
This report presents the findings of the reviews conducted by four subcontracted subject matter
experts. The peer reviewers were:
1. Mr. Ted Bohn, Argonne National Laboratories
2. Dr. Linos Jacovides, Delphi (Retired)
3. Ms. Linda Miller, independent consultant
4. Dr. Deepa Ramaswamy, Hybrid Chakra
The Peer Review Process (from ICF Report)
From December 2010 to April 2011, EPA contracted with ICF to coordinate this peer review.
ICF implemented the peer review in compliance with EPA's Peer Review Handbook (3rd
Edition).2 EPA requested that the peer reviewers represent subject matter expertise in
manufacturing cost estimating and/or automotive design.
ICF developed a list of candidate peer reviewers from the following sources: (1) ICF experts in
this field with knowledge of relevant professional society membership, industry, academia, and
other organizations, and (2) suggestions from EPA staff. ICF identified 25 qualified individuals
as candidates to participate in the peer review. ICF sent each of these individuals an
introductory screening email to describe the needs of the peer review and to gauge the
candidate's interest and availability. ICF attached to the email the reviewer charge to ensure
each candidate was familiar with the scope of work. ICF also asked candidates to provide an
updated resume or curriculum vitae (CV). Several candidate reviewers were unable to
participate in the peer review due to previous commitments, and several others did not respond.
ICF reviewed the responses and evaluated the resumes/CVs of the interested and available
individuals for relevant experience and demonstrated expertise in the above areas, as
1 Draft Report FEV07-069-303F, February 22, 2011.
2 EPA's Science Policy Council. Peer Review Handbook, 3rd Edition (http://www.epa.gov/peerreview/).OMB's Information Quality
Bulletin for Peer Review and Preamble (also in the EPA's Peer Review Handbook, Appendix B).
A-2
-------
demonstrated by educational degrees attained, research and work experience, publications,
awards, and participation in relevant professional societies.
ICF reviewed the interested, available, and qualified candidates with the following concerns in
mind. As stated in the EPA's Peer Review Handbook, the group of selected peer reviewers
should be "sufficiently broad and diverse to fairly represent the relevant scientific and technical
perspectives and fields of knowledge; they should represent a balanced range of technically
legitimate points of view." As such, ICF selected peer reviewers to provide a complimentary
balance of expertise of the above criteria.
EPA reviewed ICF's proposed peer reviewers and concurred with ICF's recommendations;
these peer reviewers were listed in the introduction. Exhibit 1 shows the representation of the
peer reviewers in the required areas of expertise.
Exhibit 1. Chart of Peer Reviewer Expertise Areas and Affiliation
Expertise Areas/
Affiliation
HEVs
Cost Modeling
Manufacturing
Mass Production
Tier 1 Supplier
Original Equipment
Manufacturer (OEM)
T. Bohn,
Argonne
National
Laboratories
S
S
L. Jacovides,
Delphi
(Retired)
^
^
^
^
L. Miller,
independent
consultant
-------
The Peer Reviewer Charge provided peer reviewers with general guidelines, as well as example
questions, for preparing their overall review, with particular emphasis on methodologies and
cost results. In addition, EPA asked each reviewer to provide recommendations on the "overall
adequacy of the model for predicting future battery prices, and on any improvements that might
reasonably be adopted by the authors to improve the model."
A mid-review teleconference was held on March 8, 2011, to discuss the charge, the purpose of
the review, and to answer any outstanding questions the reviewers might have. The call was
moderated by ICF and attended by reviewers Mr. Bohn, Dr. Jacovides, Ms. Miller, and Dr.
Ramaswamy, as well as EPA staff Brian Nelson, and FEV, Inc. staff Greg Kolwich who were
familiar with the report.
A-4
-------
1. Summary of Peer Reviewer Comments
Dr. Jacovides stated that the methodology is correct and can lead to correct results, as he had
familiarity with the approach and expected results from prior work. Given that familiarity, he felt
the report represented a superb implementation of the concept and that the analysis of the HEV
and internal combustion engine (ICE) equivalent was done very carefully, correctly, without any
obvious bias, and achieved results in agreement with his own. Mr. Bohn agreed generally, but
only for the baseline HEV.
However, two other reviewers expressed skepticism. Ms. Miller felt that the methodologies are
generally reasonable, but raised some specific concerns, including a lack of documentation
proving accurate results. Specifically, she noted that, while the paper references marketplace
validation, no examples were given. Dr. Ramaswamy agreed that the report does not
sufficiently document the validation of the methodology at a subsystem or a system level. The
implication was that, while the bottom-up approach was highly detailed, insufficient data was
given in the report to show that the resulting subsystem or system costs agreed with those
developed or published by other reasonable sources. Ms. Miller also noted that the
methodology only predicts absolute costs, and that a sensitivity analysis should be included and
documented.
Response 1: As summarized in Report section C.7, Marketplace Validation, marketplace
validation occurs at all stages of the cost analysis with special emphasis placed on cross-
checking in-process costs (e.g. material costs, material selection, labor costs, manufacturing
overhead costs, scrap rates, and individual component costs within an assembly).
For example, for the NiMH battery costs, over 80% of the battery component costs were
validated using data from subject matter experts (SME) and/or production supplier quotes. The
type of cost validation included component pricing for finished goods, raw material quotes and
manufacturing equipment pricing. Both active and passive electronics components, as well as
several mechanical type components including the D-cell battery canister, collectors, lid, and
vent button, were all validated using industry quotes. Also the majority of the battery raw
material costs (e.g. nickel-plated steel, positive and negative substrates, positive and negative
powders, electrode separator) were acquired through a formal quoting process. In several
instances, when prices were suspect, i.e. too low or high based on SME experience, more than
one supplier quote was obtained as a means of validation. For developing the manufacturing
equipment overhead rates (e.g. positive and negative coating lines, compaction lines, formation
equipment), general specifications were established and sent out for quote.
This same approach was taken for many of the main high-impact components evaluated in the
analysis including the traction motors, high-power electronic inverters, and high-voltage wiring.
Cross-checking on final assembly costs also occurs within the scope of the cost analysis, mainly
as a "big picture" check. In general cross-checking final assembly costs are typically achieved
through solicitation of industry experts. For example in the case of the traction motor analysis,
FEV consulted with a T1 traction motor supplier on the anticipated pricing of traction motors
A 5
-------
based on the established boundary conditions. The pricing estimate from the supplier was
within 5% of the calculated value.
In response to these comments concerning the potential impact of uncertainties within the cost
elements, FEV has updated the final report to include the following sensitivity analyses: 20%
increase in material costs, 20% decrease in material costs, 20% decrease in manufacturing
overhead rates, 20% decrease in labor rates, 20% increase in mark-up rates, 20% decrease in
mark-up rates.
Dr. Ramaswamy agreed that the methodology for determining costs is generally reasonable, but
highlights some significant exceptions. Specifically, engineering development costs and use of
indirect cost multipliers (ICMs) was not considered in sufficient detail and may be incorrect. An
example was given (see specific comment excerpt number 8 in Table 1) where ICM costs are
incorrectly applied to the OEMs. This would introduce bias to lower predicted costs beyond
reality, thus the engineering development costs for the subsystems should be revised.
Response 2: ICMs are used in EPA's rulemaking process to help determine the overall cost of
a technology. The specific values assigned to the ICMs are based on the complexity of the
technology. These multipliers were developed separately from this FEV costing methodology
and are not used by FEV, but rather are applied by EPA to the FEV cost results. They are,
therefore, beyond the scope of this Report. More information on ICMs can be found in the EPA
Report 420-R-09-003. See Responses #10 and #21 for discussion of engineering development
costs.
Ms. Miller also noted that the scaling methodology appeared to be overly simplified when it was
applied to labor and manufacturing overhead. Whereas the cost of direct labor is more a factor
of part complexity than one of size, certain elements of overhead costs were only minimally
affected by part size. This could introduce bias that should be explored through use of
sensitivity tests.
Response 3: Multiple scaling methodologies were applied in the analysis based on the
component type, the required change to the component for the new vehicle segment, and the
data available. For example, with the traction motor (estimated 60 kW) and generator
(estimated 30 kW), a ground-up cost calculation was developed for each assembly. By
developing a cost/kW factor based on these two data points, costs were estimated for
alternative size motors and generators.
In the scaling of the high-voltage battery, several different scaling considerations were applied.
For the various vehicle segments, the battery power capacity was increased/decreased by
altering the number of sub-modules (one sub-module equals 8 D-cell batteries which is
approximately equal to 10.6V). In the Fusion Hybrid analysis there were 26 modules connected
in series to provide 275V. To change the power capacity of the overall battery pack, sub-
modules were added or deleted to suit the system requirements. As the number of sub-
modules were added or subtracted from the analysis, so were the costs of the sub-modules. In
addition to the sub-module costs, there were nine other sub-subsystem categories which
contributed to the overall high-voltage traction battery cost. For a few of these sub-subsystem
categories (e.g. Vehicle Operations assembly, body wiring harness-low voltage) the change in
material, manufacturing overhead, and labor were considered insignificant so no scaling from
the Fusion Hybrid calculated costs was required. With some sub-subsystem (e.g. cooling,
battery covers, battery module assembly) the cost change was approximately proportional to the
number of modules added or deleted. Thus a cost scaling factor was developed based on the
A-6
-------
number of modules for these types of sub-subsystems. In selected cases (e.g. traction battery
sensing and control modules), where much of the hardware in the sub-subsystem remained
constant regardless the number of battery sub-modules added/deleted, only the hardware
changes within the sub-subsystem were accounted for, using one of two methods: (1) adding-in
or subtracting-out absolute component costs, (2) applying a scaling factor against the
components which would require change.
In the case of the high-voltage wire harnesses, a compensation factor was developed for each
vehicle segment based on the expected change in harness length, and a cost per harness unit
length. Other harness parameters would not significantly change (e.g. same connector count
and performance specification, same battery current for all applications, same number of
retention points). The same cost/unit length of harness used to develop the initial Fusion Hybrid
model was used in the scaling portion of the analysis.
The scaling methodology for many other general components, based on an increase or
decrease in size from the cost of the Fusion Hybrid components, utilized constant total
manufacturing cost (TMC) and mark-up ratios to scale up or down. For example, assume the
TMC cost of a Fusion Hybrid stamping was $2.50 and the breakdown in cost elements,
developed from a ground-up analysis, were as follows: 50% material costs ($1.25), 10% labor
costs ($0.25), and 40% in manufacturing overhead costs ($1.00). If the stamping doubled in
size (i.e., mass), maintaining the same TMC ratios, the material cost would go to $2.50, labor
costs to $0.50, and manufacturing overhead costs to $2.00. The manufacturing overhead
increase is rationalized by a larger press and longer cycle time where as the added labor cost is
the resultant of a longer cycle time.
In cases such as battery assembly, FEV and EPA agree that labor and manufacturing overhead
are not ratiometric to material mass/size changes. In these cases the methodology did employ
custom scaling factors for labor and manufacturing overhead based on input from subject matter
experts. FEV has added a sensitivity analysis in response to this comment.
The only key limitation Dr. Jacovides noted is that the methodology was limited to the two
architectures studied (split power hybrids used by Toyota and Ford and, to a limited extent,
Hyundai's P2 architecture).
Response 4: We agree that extension of the analysis to other hybrid architectures may be
useful, but this is beyond the scope of the work assignment and tasks assigned by EPA to FEV.
See also Response #9.
While he noted that the P2 battery was properly analyzed by tear down of an actual unit and
could be extended to other hybrids (GM [two mode and the Malibu ISG] and the Honda Insight),
Mr. Bohn expressed skepticism about the general subjectivity of the scaling assumptions,
particularly for P2 HEVs, but, while noting that bias was possible, he made no judgment on its
direction or magnitude. However, a general consensus seemed to be that the P2 HEV results
were more likely to be erroneous than the scaling to other vehicle types, which was, in turn,
likely to be more erroneous than the results for the baseline vehicles.
Response 5: Scaling was done on two levels in this analysis. First, component costs from the
powersplit hybrid large car (Fusion) were scaled to get costs for same or similar components on
vehicles of the same size but using a P2 hybrid design. Second, both powersplit and P2 hybrid
costs for the large car category were scaled to other vehicle sizes. We agree that there is
potentially greater uncertainty in the cost estimates as they move further away from direct
reliance on teardowns and toward more reliance on scaled results. However, cost and time
A-7
-------
constraints, and the lack of production P2 hardware at the time the work was initiated, made it
unworkable to conduct more teardowns and thereby reduce the need for scaling. The scaling
parameters and methodology were carefully developed to avoid reliance on subjective or
arbitrary assumptions, and to ensure high confidence in the whole range of results. Specific
comments about scaling for P2 components and across vehicle sizes are addressed in
Responses #3, #7, and #9; we have also added sensitivity analyses to the Report to help
address these comments.
b. in the
Mr. Bohn suggested that the scope is "just right" and offered no conclusive statements. He
noted that expanding the scope of the study would likely introduce more variability and that
reducing it would not necessarily increase its validity or accuracy.
Dr. Jacovides said that, although beyond the scope of this report, the study results would be
meaningless without knowledge of appropriate use of ICMs. This was a limitation of the study
the study may be sufficiently detailed exclusive of ICMs, but end results could vary by up to a
factor of two depending on the ICMs.
Response 6: See Response #2.
As introduced previously, more substantial concerns were raised over the scaling of results,
especially to P2 HEVs. Dr. Jacovides expanded on the comments from part (a), expressed
concerns about both the methods and results for the P2 system. While the results of scaling for
the P2 system may be in the right direction, the sizing of the electrical system (power electronics
and the electrical machine) were likely incorrect. Because the duty cycle of the electrical
system in a P2 HEV is very different than that of the powersplit HEV, the ratios of copper to iron
in magnets will likely be different. Further, if the electrical machine for the P2 was sized based
on power, this was incorrect. Instead, torque and duty cycle are the primary determinants of
size (and cost). Also P2 HEVs have a clutch to disconnect the engine so that regenerative
braking does not have to be reduced to provide for engine friction thus providing an all electric
range (AER). The resulting 32.4kW power of the electrical machine will not provide sufficient
required torque and power for AER. Further, since the size of electrical machine is determined
by torque, not power, a slower speed machine will be heavier which contradicts the assumed
20% vehicle curb weight reduction for the P2 architecture for all vehicle segments.
Response 7: We did not assume significant all electric range (AER) capability in the P2 hybrid
being costed and so did not evaluate the ability of the electric motor and associated
components to provide required torque and power for AER. We have revised the Report to note
this so that users of the Report results may be aware of it.
Since we were not considering significant all electric driving modes for the P2 application, we
believe motor sizing based on power is correct for this analysis. Furthermore, EPA engineers
involved with this project have evaluated the P2 designs now nearing production to help validate
our choices of machine specifications and found good correlation. The Report has been
clarified regarding how weight reduction is factored-in to the P2 architecture. Note that the mass
reductions considered in the P2 analysis are the result of non-hybrid-specific innovations, such
as the shift to lighter materials throughout the vehicle.
A-8
-------
Dr. Ramaswamy agreed that the assumption of a 20% power and weight reduction assumed
for the P2 hybrid may be unjustified. Further, she found that there is no justification in other
literature that the Lithium Polymer battery (as opposed to nickel metal hydride [NiMH]) would be
a better long term solution for the P2 hybrid.
Response 8: The Report has been revised to clarify the basis for the 20% weight reduction
(and resulting reduction in required power), and for the costing of a Lithium-polymer battery in
the P2 vehicle but not the powersplit vehicle. It is not the migration from powersplit to P2
architecture that enables these changes; we think it likely that the powersplit hybrid could
incorporate the weight reduction and use of Lithium-polymer battery as readily as the P2 hybrid,
though this was not evaluated. Rather, the weight reduction was specified by EPA as an
upfront assumption for the FEV analysis, and was for technology innovations not related to
hybridization such as a shift to lighter materials throughout the vehicle; it included consideration
by EPA of offsetting weight additions from hybridization.
Likewise the Report included a Lithium-polymer battery for the P2 architecture and a NiMH
battery for the powersplit architecture not because these are inherently the right choice for these
two types of HEVs, but because the EPA team expects Lithium-polymer batteries to have great
potential for use in hybrids and asked that it be included in the P2 costing. On the other hand,
for the powersplit costing, EPA directed FEV to maintain the Fusion characteristics (weight and
battery type) in order to keep that result focused on the teardown findings, with minimal
extrapolation to other hardware that might sensibly find its way into later-generation hybrids,
such as Li-polymer batteries and lightweight materials. For this reason, it would not be
appropriate to compare the powersplit and P2 cost results on an equivalent basis.
Dr. Jacovides argued that the study will be difficult to apply to other vehicles or architectures
without the detail provided by a similar tear-down. Ms. Miller agreed that extrapolation to other
vehicle sizes cannot be done without the basic underlying detailed studies, and that
extrapolation of costs for vehicles other than the Fusion relies on use of scaling and does not
have the same level of detail as the rest of the study. A different use of scaling factors, such as
by applying scaling factors to material cost and investment in equipment instead of for
manufacturing cost and burden could yield a very different result.
The general consensus was that the scaling portion of the study was the most dubious.
Response 9: We agree that extending the Report's findings to vehicle types or hybrid
architectures not specifically covered in the Report should not be undertaken without a properly
crafted methodology for doing so, although, depending on how far the extension goes, we do
not think it would necessarily require extensive new teardowns. See Responses #3, 5 and 7 for
additional responses on the scaling issues raised by these peer reviewers.
c, of
Dr. Jacovides reiterated his contention that the report's central analysis (comparison of a hybrid
and an ICE Fusion) was very well done. However, he raised concerns with estimation of the
following cost assumptions: 1) development of control software, 2) integration of the electrical
and mechanical parts, and 3) calibration. All are upfront engineering costs that should be
considered as part of the cost of the vehicle, although they may be insignificant by the time
production volume has reached 450,000 units.
Response 10: At the OEM level, all 3 of these concerns are accounted for by applying an
Indirect Cost Multiplier (ICM), to the FEV results (see Response #2). At the supplier level, the
cost of developing control software, integrating the electrical/mechanical parts, and calibration
A-9
-------
are accounted for with an ED&T cost for each component, which is a percentage of the
component cost, based on its complexity, in the context of the methodology assumptions and in
particular in the assumption of high volume production.
Ms. Miller was concerned that lack of communication [i.e. independent review of the study] with
the OEM's - while consistent with EPA policy - can lead to inappropriate validation of the
teardown costing. Dr. Ramaswamy agreed that insufficient independently determined
system/subsystem costs were used to validate the calculated costs. The Report does discuss
this, but specific examples of validation should be considered as additional inputs to the
process.
Response 11: The Agency does not typically ask the party that is potentially being regulated
for direct review of cost assumptions prior to publication of a rulemaking analysis. However
OEMs can and do provide extensive review of EPA cost analyses through the public process
required in every rulemaking. We have high confidence in the study validation methodology
because the subject matter experts (SMEs) utilized by FEV in the validation process are familiar
with OEM manufacturing practices and cost structures, and their input improves the veracity of
FEV final cost results. See also Response #1 for discussion of the extensive marketplace
validation performed by SMEs.
Dr. Ramaswamy also argued that the major flawed assumption of this study was that the high
voltage battery will be manufactured in the United States. NiMH batteries are not manufactured
in volume in the United States. Although several companies have plans to manufacture Li-ion
batteries, the cells typically come from Asia. This inaccurate assumption biased the cost results
high.
Response 12: The assumption of U.S.-based manufacturing was directed by EPA and is typical
of FEV's cost analyses for light-duty vehicle rules, based on direction from EPA. We agree that
some elements of vehicle component production are likely to occur elsewhere, and, because
this would most likely be for the purpose of cutting costs, the assumption serves to make FEV's
analysis conservative (i.e., biased toward higher costs). However, we believe trends toward
greater alignment of global wage structures and rising costs of long-distance shipping will tend
to mitigate this impact in the future. Furthermore this approach ensures a consistent framework
for costing performed by FEV as part of this work assignment.
We note that the question of future production locations for vehicle batteries and
subcomponents is far from settled. HEV systems represent a relatively new manufacturing
sector and the ultimate location of production facilities is yet to be determined for the domestic
market. Sizable new investments in the United States have been initiated within the past few
years. See also Response #14e.
d.
Reviewers noted concern about several assumptions included in the study. Dr. Jacovides again
noted his general conclusion that while the assumptions used are appropriate, the implicit
assumption that a downstream user without the same expertise as FEV would be able to run the
model is unlikely. Dr. Ramaswamy agreed that assumptions were generally reasonable, but
highlighted especially the unreasonable assumption incorporated in the scaling parameter for
the battery.
Response 13: All of the underlying information in the FEV cost analyses (including databases,
spreadsheet formulas, and process flow diagrams) are available to the public via the rulemaking
A-10
-------
docket. Any component supplier or OEM should be able to work with this information, and
understand exactly how the costs were determined. If suppliers, OEMs, or other stakeholders
discover any errors or incorrect assumptions in the FEV analysis, we encourage them to notify
EPA. The FEV methodology is not intended to be used by others as a standalone model. We
agree that people with expertise in this type of analysis would be needed to build on the FEV
analyses. See Response #3 regarding battery scaling assumptions.
Ms. Miller listed the following specific assumptions that should be re-considered:
The assumption that the technologies used may be considered mature should be
evaluated. The assumption of maturity impacted numerous underlying cost elements,
including lack of allowances for learning, scrap rates, non-recovered engineering,
design, and testing (ED&T) expense and capital costs, and equipment end of life costs.
o Response 14a: When conducting the cost analysis for each technology
configuration, a number of assumptions and boundary conditions are required
upfront in the analysis prior to the start of any costing work. The same
assumptions and boundary conditions are applied to both the new and baseline
technology configurations, establishing a consistent framework for all costing and
resulting in a level playing field for comparison. The user of the FEV cost
estimate must be aware of these boundary conditions when determining if the
FEV cost estimate is applicable to the user's specific situation.
o To account for cost differences arising from factors outside of this framework,
such as differing technology maturity levels in the timeframe of interest, EPA may
decide to apply adjustments, such as reverse learning factors.
o ED&T expenses are coved at the supplier manufacturing level within the mark-up
rate applied by FEV. ED&T at the OEM level is applied through the application of
an Indirect Cost Multiplier (ICM). See Responses #2, #10, and #21.
o Capital costs are included in the manufacturing overhead rate based on a
straight-line depreciation method with zero end-of-life value.
The assumption that no new or modified equipment maintenance is required is
inconsistent with equipment at the end of its life cycle, assumed above. Together, these
biased the cost estimates low.
o Response 14b: Wthin the manufacturing overhead rate/burden rate calculator
template, which calculates the overhead rates for various operations and
processes, there are allowances made for equipment "maintenance, repair and
other" (MRO) expenses. The contributions are calculated as a percentage of the
primary and process support equipment costs. The Report has been revised to
clarify this point.
The assumption that integration of new technology would be planned and phased in to
minimize non-recoverable expenses would be cost effective. In reality, new technology
requirements to achieve fuel economy improvements and emissions reductions will
preempt this consideration.
o Response 14c: The matching of new technology with timing of product cycles is
not within the scope of this study. However it is a key facet of the methodology
A-11
-------
used by EPA to ensure that new technology introduction due to standards does
not disrupt manufacturers' product cycles. This is accomplished through the
assumption of technology phase-in caps and specified product cycle intervals in
the agency modeling work.
The markup rate needs to vary dependent upon the part size and part complexity. If
tolerance limits are not considered part of part complexity, tolerances need to be
considered as another factor in determining scrap rates. Assumed scrap rates should
also be verified.
o Response 14d: Tolerance limits are considered to be one of the parameters
defining part complexity, which in turn affects costs. Through SME reviews,
which consider many variables including tolerance and performance
requirements, scrap rates are set accordingly. For example compressor wheels
in turbochargers have tight tolerance requirements and therefore were assigned
a higher scrap allowance in the FEV Pilot Study and additional case studies. In
addition, the cost of manufacturing tight-tolerance parts is accounted for in
manufacturing equipment selection and cycle times.
The assumption that all sourcing/manufacturing centers will be in the United States was
not valid and could bias the results high or low.
o Response 14e: See Response #12. To establish a consistent framework for
costing, a common cost structure was required for all cost analyses. This
included establishing a common manufacturing location, the United States.
Based on this boundary condition, most manufacturing operations and processes
were developed under the assumption of heavy automation with less manual
labor. This key assumption was directed by EPA at the start of this cost analysis
and, to the extent that it might bias the costs, it will do so in a conservative
direction, i.e., somewhat higher costs. Recent trends in manufacturing dynamics
for advanced technology components make it difficult to predict the degree and
direction of low-cost-country sourcing. We believe that trends toward greater
alignment of global wage structures and rising costs of long-distance shipping will
tend to mitigate this impact in the future.
Assumed labor rates may need to be adjusted to include overtime costs and other
premiums. It was unclear from the report if this was included and could bias the results,
depending on union agreements and/or operating expenses.
o Response 141: Labor rates do include allowances for overtime costs and other
shift premiums. These allowances are applied to the base wages as part of the
fringe calculation. Note that the rulemaking assumption that rollout of the new
technologies as a result of the standards will not disrupt the normal production
practices employed by OEMs and suppliers, helps to mitigate any need to
assume excessively large additional allocations for overtime pay. (See Response
#14c).
Packaging cost assumptions should be checked, based on the sample calculation (page
50, Figure C-6).
o Response 14g: FEV has made some simplifying assumptions in calculating
packaging costs because these costs are a very small part of the overall new
A-12
-------
technology costs. We believe a more thorough analysis of packaging costs
would not have a significant effect on the overall cost numbers.
Allowances for a percentage of pallets/racks out for cleaning/repair are not included and
biased the packaging cost low.
o Response 14h: The cost analysis includes a cost allowance in the
manufacturing overhead rate for general plant and dunnage cleaning. The
Report has been revised to clarify this point.
The assumed Cost of Complexity is inappropriate. Volumes of 450,000 per year
assumed that the major complex assemblies (engine, transmission, and complex
subsystems) are produced on dedicated lines. If they are not, then a cost of complexity
factor needs to be added. The 75% combined utilization/efficiency assumption was
reasonable unless hybrid components are assembled on the same lines as the baseline
products (as they will be), in which case this utilization/efficiency is over-stated. This
biases the results low; additional complexity should be factored into the
utilization/efficiency calculation.
o Response 14i: At the assumed 450K annual volume for hybrid technology
components, subsystem, and systems, it is expected that the technologies
costed are all produced on dedicated lines. In addition, given the high-
volume/mature-technology assumption, an 85% utilization is justified. If
manufacturers do in fact decide to mix other technologies into these lines, we
would expect the reason for doing so would be to reduce overall costs in
comparison to building on two lines, and so would not warrant assigning a cost
increase to the high-volume technologies covered in the report.
With respect to System Scaling Cost Analysis, ratios used to develop sizes and material
costs for HEV components (traction motors, high traction batteries, etc.) were
appropriate, but use of these ratios to determine other factors (especially labor and P2
HEV powertrain components) was less valid. These are more related to part complexity
than part size. Which costs are scalable should be reevaluated.
o Response 14j: See Response #3.
Mr. Bohn discussed some assumptions, particularly regarding the base vehicle and the P2
Hybrid having equivalent performance with increased fuel economy. He said associated
assumptions about the amount of engine blending and depth of battery discharge were
subjective and expressed concerns regarding the lack of electric machine rating standards.
However, he made no mention of their reasonableness or direction of influence on the study's
results.
Response 15: An objective of the technology cost analysis associated with the EPA rulemaking
effort is to generally assume equivalent performance (other than fuel economy) between
baseline and new technology configurations. For example, cost analysis that assumes
significantly reduced acceleration in new technology vehicles would not be expected to
adequately inform EPA rulemaking decisions, because of the added complexity it would
introduce over customer acceptance and safety. Similarly, because towing capability is
essential for large trucks, we assumed that downsizing of the engine would not be appropriate
for these vehicles. The blending and depth of discharge characteristics were selected by EPA,
consistent with the equivalent performance objective, as inputs to the study's P2 component
sizing determinations.
A-13
-------
Three of four reviewers generally considered the study results appropriate but commented on
the need for increased validation. Dr. Jacovides commented that the results were reasonable,
but noted that it would be useful to have Ford and Toyota review them before making the report
public. Dr. Ramaswamy agreed that the results were appropriate for the given scope,
assumptions, and inputs, but noted that the description of the validation of the costing
methodology was insufficient and that a sensitivity analyses and further analyses/correction of
some assumptions were warranted. Mr. Bohn, too, agreed that the results were reasonable
given the scope, assumptions, and inputs, but felt that reasonable validation was achieved,
although he considers the level of validation appropriate to be subjective.
Ms. Miller disagreed. She felt that, given the levels of assumptions made, at best cost
estimates are directionally correct, which is inconsistent with the stated goal of absolute costs.
In particular, she had concerns regarding validation. While the methods used were solid
(teardown analysis, process flow diagrams, analysis of comparable parts, etc.), numerous
methodological assumptions were used and their validation is insufficiently documented. She
recommended sensitivity testing, appropriate and correct accounting for component sourcing,
and reevaluating labor costs.
Response 16: See Responses #1, #3, and #11. FEV is adding sensitivity analyses to the
Report in response to the peer review comments.
Generally reviewers seemed to express more reservation about the scaled results than the
baseline vehicles, for a variety of reasons.
Dr. Jacovides noted that scaling for vehicles with identical architecture but different power
inappropriately account for labor. Similarly for P2 HEVs, costs for electrical machines should
not be scaled as power, but on torque and duty cycle. Ms. Miller agreed that the ratios used to
size HEV components was appropriate for material costs and investment in equipment, but that
using the size ratio scaling methodology for overhead cost, direct labor costs, and required
staffing was inappropriate. She had these same concerns with scaling for the P2 HEV
calculations. Dr. Ramaswamy also agreed that for most components, the scaling to other
vehicle classes was reasonable. Mr. Bohn added that while the approach used in scaling
appears reasonable, he had concerns that the actual values used in the scaling approach could
be off and lead to erroneous results. However, this was not supported by his general conclusion
above regarding the reasonableness of results.
Response 17: See Responses #3, #5, and #7.
Mr. Bohn and Dr. Jacovides commented that the NiMH battery scaling was done correctly,
although Dr. Jacovides noted that scaling did not consider an alternative approach of using a
larger number of smaller cells. He believed that the approach used for the P2 architecture was
directionally correct but the results would not be as accurate as those between the baseline
hybrid and ICE vehicles. Although he noted that the estimated cost of the cells seemed
reasonable, Dr. Jacovides raised two questions about the treatment of the Li-ion battery: 1) that
discussion should be added to explain preservation of battery life when scaling by nominal kWh,
and 2) that clarification should be made on what size battery is cost for the P2 HEV. Dr.
Ramaswamy agreed. She noted that scaling of parameters across different vehicle classes
-------
needed to be better explained and justified, given that this one component was responsible for
the bulk of the cost of the hybrid powertrain.
Response 18: Consistent with the core philosophy for all of the cost work performed by FEV
for EPA, we chose to determine Li-ion battery costs based on teardown of an actual production
battery. We agree that other configurations exist or may exist in the future, but these will likely
only serve to lower eventual battery costs to the extent that they prove more cost effective than
the design we chose, thus making the battery costs developed by FEV somewhat on the
conservative side. Li-ion battery sizes for the vehicle segments analyzed were provided by the
EPA team as an input to the P2 HEV cost analysis; depth of discharge constraints to ensure a
robust battery life was one of the considerations taken into account by the team. The battery
sizes costed for the P2 vehicle analysis are provided in Table H-2 of the Report. See Response
#3 for additional discussion on how the scaling of battery costs across the vehicle segments
was carried out.
1.2. on
General comments not included in the earlier sections are discussed in this section.
Ms. Miller complemented the detail and effort of the analysis and report and the use of
recognized methodologies. Dr. Ramaswamy noted a small number of omissions and
discrepancies. She noted that, while the report talked about the applicability of the powersplit
hybrid system to the sub-compact, small, large, and minivan vehicle segments, it should clarify
that this group also covers small SUVs such as the Ford Hybrid Escape, which is one platform
that already supports this architecture.
Response 19: The applicability of the powersplit hybrid system technology to the various
vehicle segments was established by EPA as part of the study inputs and so is within the scope
of the broader rulemaking analysis rather than the FEV Report. Note that the Ford Escape falls
into the large car segment, and therefore actually is included in the powersplit cost analysis.
Dr. Ramaswamy also noted several specific items of concern. She indicated that the study
seemed to omit a high-voltage DC/DC converter used by the traction motor and generator,
which is used in the Fusion Hybrid and should be included in the cost. Table E-2, compared to
those in Table D-3, showed inconsistencies that should be addressed. Dr. Ramaswamy also
noted that Table A-1 should have calculated the percentage increase as compared to the base
non-hybrid vehicle cost, instead of calculating the increase with respect to the mid/large size
vehicle segment cost. Also, in Figure B-1, she questioned why the bill of materials (BOM) was
not updated after step 6, when additional information has been gained about the component
after its disassembly. She also asked what the 19,149 parts stand for on page 50, first
paragraph.
Response 20: The DC/DC converter was included in the analysis and covered in the Report in
section D.7.1.2 "Voltage Converter/Inverter Subsystem". The costs of the DC-DC converter
were reported in Table D-10.
The discrepancies between Table D-3 and Table E-2 (now labeled "F-2" in the Final Report) are
based on the using an electric horsepower to kW conversion (0.746) versus mechanical
horsepower to kW conversion (0.736).
A-15
-------
The comparison bill of materials (C-BOM) referred to in Figure B-1 is a top level BOM used to
identify component/assembly differences at a high level. Every component or assembly which
is disassembled further for a cost analysis has a separate manufacturing indented BOM.
In the packaging calculator section of the Report, the 19,149 identifies the estimated number of
battery packs which would be shipped out in a two-week window. This number is directly used
to calculate the dunnage costs.
The percent increase in cost for each vehicle category (Table A-1) over the baseline vehicle is
not included in this study because we did not cost the base vehicle in each category.
Dr. Ramaswamy also believed the methods for determining the engineering design costs for
various components/systems were unclear. These included: 1) the Atkinson engine engineering
design cost, associated control system, and calibration; 2) the engineering design cost for the
electronics controllers, software for the battery system, and mechanical design of the battery
system (the numbers presented appear low); 3) the ED&T for the traction battery assembly (too
high relative to that for the control module, given the relative engineering efforts) (Table D-11).
Response 21: Within the automotive industry, there are several methods used to calculate
ED&T. These methods range from a detailed build-up of estimated R&D and ED&T costs (at all
levels, Tier 1, Tier 2, Tier 3), which are then amortized over a defined annual volume and life
cycle, to a percent allocation, based on component complexity, of the total manufacturing costs.
In between these two methods are hybrid approaches which utilize a combination of the two
methods. The exact method an OE or supplier may choose to use, or accept, may be based on
several factors including size of company, product portfolio, production volumes, product range
(i.e., custom, commodity or hybrid of the two), market risk, dual sourcing, and product
experience.
Performing a comprehensive work-up of ED&T costs, for every component, appears on the
surface to be the most robust method. However the probability of accomplishing this task
accurately, based on the amount of detailed information required, without the necessary
confidential business information, is very low.
For the EPA cost analyses, FEV chose to use the percent allocation of total manufacturing costs
as the method for calculating ED&T costs. This same methodology is used extensively by
automotive suppliers and OEs in today's industry. Also this is the same method chosen by
FEV, for all costing work, conducted for all customers.
FEV's ED&T rates are based on a constant recovery/payback rate. This constant rate is based
on the assumptions that the established automotive supplier, within a stable economy, will
maintain a certain size R&D department, product and design engineering department, and test
department. The cost to run this department will generally remain proportional to the value of
manufactured product leaving the facility, under a stable business climate. There are always
exceptions to these rules. For example, where the integration of the new technology
configuration is slow to reach financial planning volumes, or an existing technology remains in
production for an extended life, with no replacement technology planned, adjustments to rates
should be applied - based on the defined boundary conditions.
Based on the assumptions and boundary conditions established for the EPA cost analysis
project (i.e., well developed product designs, high production volumes, high first time
manufacturing yields, significant marketplace competition, low field warranty, etc) an ED&T
A-16
-------
mark-up range between 0% to 6% has been selected for the analysis. The selected range is
based on a combination of publications from OEMs and suppliers, input from FEV's subject
matter experts in manufacturing, examining ED&T mark-up rates in the automotive industry,
publicly-available financial data (i.e., 10K reports), and EPA's Indirect Cost Multiplier report.
Dr. Jacovides recommended specific companies that should be consulted to assess the
accuracy of results: Ford for the baseline vehicles and those scaled according to size and
Honda or GM for scaling to P2 HEVs. Also, individual component costs should be compared to
those used on the Volt and Leaf.
Response 22: See Responses #1 and #11. We agree that it may be instructive to compare
costs of hybrid components from this study with those from other studies that focus on other
hybrid and electric vehicle designs. However, we believe these vehicles are markedly different
in function and design and so such comparison would not directly inform the validity of the FEV
study results.
Ms. Miller also noted that validity testing of the Munro & Associates software, FEV databases,
and costing algorithms should be performed and documented. Hypothesis testing of
assumptions concerning burden rates, product maturity, etc. and sensitivity analysis to
demonstrate correlation to actual component costs should also be added to the study.
Response 23: As discussed in Response #1, cost validation occurs at multiple stages of the
analysis. During the build-up of cost models, much of the information acquired to construct the
models is based on supplier quotes. In general, subject matter experts (SMEs) construct all
cost models. Once the cost models are completed, the SME will review the models for
accuracy by running some pilot runs with components/assemblies with known costs as part of
the validation process.
Information loaded into the databases comes from various reputable sources. For example the
majority of high impact material pricing comes from supplier quotes. The data contained within
the labor databases is extracted from the Bureau of Labor Statistics. We have added sensitivity
analyses to the Report in response to the peer review comments.
A-17
------- |