-31-

4. VEHICLE PRODUCTION AND LIFECYCLE COST

This section examines the costs of producing and operating the advanced technology
vehicles that manufacturers may build to meet ZEV program requirements. First, we describe the
components of battery-powered electric vehicles (BPEVs), direct hydrogen fuel-cell vehicles
(DHFCVs), gasoline hybrid electric vehicles (GHEVs), and gasoline vehicles that meet the partial
zero emission vehicle (PZEV) emission requirements. Second, we provide an overview of the
factors that can cause component costs to fall over time. Third, we discuss the costs of the
various components. We estimate costs at the production volumes likely between 2003 and 2007
and at high production volumes. The estimates at high volume are based on well-grounded
projections of technological improvements that are plausible given what is known about the
technology and about the manufacturing processes possible at high volumes. They are thus the
costs that can currently be anticipated after an introductory period during which costs will be
higher.
Fourth, we combine the component costs to estimate the incremental cost of each type of
advanced technology vehicle over the cost of a gasoline internal combustion engine vehicle
(ICEV). The overall social cost of an advanced technology vehicle depends not only on the
initial vehicle cost, but also on the cost of operating and maintaining the vehicle over its lifetime.
We thus also estimate the incremental lifetime operating costs of each vehicle relative to an ICEV
and conclude the section by combining incremental operating cost with incremental vehicle cost
to calculate the incremental lifecycle vehicle cost.

4.1 KEY VEHICLE COMPONENTS

We begin by describing the primary components of BPEVs and then turn to the primary
components of DHFCVs, GHEVs, and PZEVs.

Battery-Powered Electric Vehicles

The cost increment of a BPEV over a gasoline ICEV comes from the addition of the
following basic components:
1. battery modules
2. battery auxiliaries such as tray and straps, harness, and cooling system (modules +
auxiliaries = battery pack)
3. electric motor
4. power electronics to control the electric motor
 -32-

5. electric vehicle transmission

6. on-board (integrated) charger1
7. other auxiliaries (e.g., electric power steering, electric brakes).
Cost savings accrue from the exclusion of some components common to conventional vehicles:
internal combustion engine and emission control system, fuel tank, and conventional
transmission.
Table 4.1 reviews the technical specifications of the BPEVs developed by major
manufacturers. Based on this review, we cost out a full-function EV with 30 kWh of battery
storage and a 50-kW electric motor between 2003 and 2007. We cost out a city EV with 10 kWh
of battery storage and a 25-kW electric motor during the same period. For our high-volume cost
estimates, we allow the full-function EV battery pack to vary between 24 and 30 kWh and the
city EV battery pack to vary between 8 and 10 kWh to account for plausible improvements in
vehicle efficiency. Manufacturers are currently using advanced batteries—nickel metal hydride
(NiMH), nickel cadmium (NiCd), and lithium ion (LiIon)—in their full-function EVs, rather than
the more conventional lead acid (PbA) batteries. Because NiMH batteries currently appear to be
the best suited for automotive use (see discussion in Anderman, Kalhammer, and MacArthur,
2000), we assume they are the advanced battery that auto manufacturers choose for their
vehicles.2 We also run scenarios with PbA batteries because of the possibility that they will be
used in some types of vehicles. Due to their low energy density, however, PbA batteries may be
difficult to use in vehicles with the ranges we assume here. Their size and weight also create
challenges in designing EVs with interior space and handling characteristics comparable to
vehicles currently on the road.

Direct Hydrogen Fuel-Cell Vehicles

The cost increment for DHFCVs is due to the following:
1. proton exchange membrane (PEM) fuel-cell system
2. compressed hydrogen tank
____________
1According to the Pacific Gas and Electric EV glossary, an off-board charger is “a charger with the
intelligence and control in the charger stand, not on the vehicle. [An] on-board charger is a charger with
the intelligence and control on the vehicle, not in the charger stand.” On-board chargers appear to be
superior to off-board chargers. According to Korthof (2000), “Integrated charging systems have been
developed for electric vehicles by Ford, GM, Renault, Toyota, and Volkswagen. AC propulsion drive
systems include an integrated charger rated at 20 kW. This system compares favorably in efficiency,
weight, cost, and power quality, and also supports several times more power than comparable isolated
conductive chargers or inductive charging systems.” An onboard charger also requires some off-board
equipment (described in Subection 4.2).
2LiIon batteries have the energy densities needed for vehicles with a 100-mile range, but this
technology has shortcomings (see Anderman, Kalhammer, and MacArthur, 2000).
 -33-

Table 4.1
Specifications of BPEVs

Peak Energy
Power Storage Battery Rangea Electric
Vehicle (kW) (kWh) Type (miles) Motor Type
Full-Function EVs
Toyota RAV4-EV 50 27 NiMH 84/78 DC permanent
magnet
Nissan Altra EV 62 32 LiIon 95/82 AC
Vehicle modeled in this study 50 24-30 NiMH 90-110 AC
& PbA
City EVs
Toyota e·com 19 ~8 NiMH 60 DC permanent
magnet
Ford Th!nk City 27 12 NiCd 60 AC
Vehicle modeled in this study 25 8-10 NiMH 50-60 AC
& PbA
aNumbers separated by a slash show city range with accessories followed by highway range
with accessories (see Table A.1 in Appendix A).

3. small (GHEV-sized) battery pack (or other energy storage device)

4. electric motor and transmission
5. power electronics.
Cost savings result from the exclusion of the engine, emissions, control system, transmission, and
fuel tank found in an ICEV.
The PEM system is in turn made up of the fuel-cell stack and auxiliary equipment
associated with the stack. The auxiliary equipment manages the air, water, and thermal issues
related to the system and includes compressors, heat exchangers, a humidification system, safety
devices, and a control system (Lipman, 1999b, p. 12). The fuel-cell stack is made up of the
following components:
1. membrane electrode assembly (MEA)
a. cathode
b. anode
c. electrolyte/membrane
d. cathode and anode catalyst (platinum); and
2. bipolar (separator) plates.
A fuel-cell stack consists of many fuel cells. The structure of a single fuel cell is illustrated in
Figure 4.1.
 -34-

SOURCE: Chalk, 2002.

Figure 4.1—Single PEM Fuel Cell

Ballard Power Systems, Incorporated, a leader in the development of PEM fuel cells,
describes the key components and principles of operation of a fuel cell as follows:

A single fuel cell consists of a membrane electrode assembly and two flow field
[bipolar] plates. Single cells are combined into a fuel cell stack to produce the
desired level of electrical power. Each membrane electrode assembly consists of
two electrodes (anode and cathode) with a thin layer of catalyst, bonded to either
side of a proton exchange membrane (PEM). Gases (hydrogen and air) are
supplied to the electrodes on either side of the PEM through channels formed in
the flow field [bipolar] plates. Hydrogen flows through the channels to the anode
where the platinum catalyst promotes its separation into protons and electrons.
On the opposite side of the PEM, air flows through the channels to the cathode
where oxygen in the air attracts the hydrogen protons through the PEM. The
electrons are captured as useful electricity through an external circuit and
combine with the protons and oxygen to produce water vapor on the cathode side
(Ballard, 2002).
The feasibility of compressed hydrogen fuel tanks has been shown in a number of
prototype vehicles. Table 4.2 lists the specifications of some of the prototype DHFCVs that have
been produced by various automakers. Manufacturers are experimenting with different energy
storage devices and different balances between battery and fuel-cell size. Some are using NiMH
 -35-

batteries to store energy and provide power; others are using supercapacitors. Some designs have
a large fuel cell (over 60 kW) and a modest battery (25 kW); others have the reverse. For our
Table 4.2
Specifications of Prototype DHFCVs

Electric
Motor
Top Power Battery Type and
Capacity and Pressure Speed Range Plant Power Power
Make/Model of Hydrogen Tank (mph) (miles) (kW) (kW) (kW)
Honda 5,000 psi, 130-liter tank 87 185+ 78 n.s 60 kW AC
Toyota 3,700 psi 95 155 90 n.s. 80 kW DC
Highlander
FCHV-4
Ford Focus FCV 5,000 psi, two 41-liter tanks 80 155 80 n.s. 67 kW AC
Hyundai 5,000 psi 77 100+ 75 n.s. 65 kW
Santa Fe FCEV
Vehicle modeled in this study
Option 1 5,000 psi n.s. 150+ 25 60 75 kW
Option 2 5,000 psi n.s. 150+ 60 25 75 kW
NOTE: n.s. = not specified.

near-term projections, we assume a 25-kW fuel cell because high fuel-cell costs in the near term
suggest that it will be cheaper to keep the fuel cell small. For volume production, we examine the
costs of both vehicles with relatively large and relatively small fuel cells (see options 1 and 2 in
Table 4.2). Range should be over 150 miles with the fuel tank we specify in our analysis.

Gasoline Hybrid Electric Vehicles

The cost increment associated with a GHEV comes from the addition of the following
components:
1. small battery pack
2. electric motor
3. power electronics (e.g., motor controller).
Cost savings over a conventional ICEV result from the smaller internal combustion engine
needed to power a GHEV.
Table 4.3 lists the specifications of GHEVs that have been introduced in the United States
or are scheduled to be introduced in the near future. Based on these figures, we cost out a GHEV
with an NiMH battery producing 25 kW of power, an electric motor of 33 kW, and an internal
combustion engine that is 33 kW smaller than it would need to be in a conventional ICEV.

Gasoline Partial Zero Emission Vehicles

A gasoline PZEV must meet very stringent emission, durability, and warranty standards.
The vehicles must be certified to CARB’s
 -36-

• 150,000-mile super ultra low emission vehicle (SULEV) exhaust emission

standard for light-duty vehicles (LDVs),

Table 4.3
Specifications of Typical GHEVs

ICE Battery Energy

Peak Motor Battery Power Storage Electric
Model Power (kW) Type (kW) (kWh) Motor
Honda Insight 50 NiMH ~20 0.94 10 kW DC
permanent
magnet

Honda Civic 63 NiMH 9.7 0.86 10 kW DC

GHEV permanent
magnet

Toyota Prius 52 NiMH 25 1.4 33 kW DC

permanent
magnet

Vehicle 52 NiMH 25 not 33 kW DC

modeled in specified permanent
this study magnet
NOTE: ICE = internal combustion engine.
• zero evaporative emission standard, and
• on-board diagnostics (OBD) system requirements. The OBD system must be able
to detect emissions that exceed the standards even very slightly.
In addition, the vehicle warranty on emission control equipment must be extended to 15
years or 150,000 miles, whichever occurs first.
As is discussed in Subsection 4.4, major progress has been made in designing vehicles that
meet PZEV standards. The required emission control system is similar to that in vehicles on the
road today.

4.2 OVERVIEW OF FACTORS THAT CAUSE COSTS TO DECLINE

Before examining the costs of vehicle components, it is useful to review the various factors
that cause costs to fall. Economists and engineers generally agree that three factors, or processes,
can lead to reductions in unit component cost: production scale economies, learning, and
technological advances.

Production Scale Economies

Higher rates of production per period usually cause the unit cost of production to drop.
They can mean discounts on bulk purchases of input materials and components and more efficient
 -37-

usage of machinery and labor. They can also justify the increased use of automated
manufacturing equipment, which can further reduce unit costs. In addition, higher production
volumes can justify changes in product designs that reduce production costs. For example, it may
not be cost-effective to design application-specific integrated circuits until production volumes
are high.
Predictions of the cost savings due to scale economies are based on what is known about
the product design and technological improvements that can be incorporated at high production
volumes. They are also based on what is known about the manufacturing processes possible at
high volumes.

Learning
Costs may drop over time as learning-by-doing improves the efficiency of the
manufacturing process. These reductions are in addition to those solely caused by an increase in
production volume. Refinements or wholesale changes in the production process may be
discovered based on accumulated production experience. For example, production experience
may lead to a better way to combine the various steps of the production process or to a more
efficient mix of capital and labor.

Technological Advances
Improvements in the technical design of a component can reduce its unit cost. For
example, advances that allow reductions in the amount of nickel per kilowatt-hour storage
capacity will reduce NiMH battery costs.
In our analysis of costs, we restrict our attention to component designs that appear feasible
given what is known about the technology. While a specific design does not need to currently
exist, there should be a clear path to it that appears feasible. Technology breakthroughs that
cannot currently be foreseen may indeed happen in the future, but it does not make sense to base
policy decisions on the presumption that they will.

How Cost Reduction Factors Enter Our Analysis

We carefully consider the effects of production scale on our estimates, and the high-
volume estimates are based on knowledge about the types of production processes that are
feasible at high volumes for the products considered. Manufacturing experience may result in
additional cost reductions, and experience in many settings suggests that costs do fall
substantially with manufacturing experience. But this is not guaranteed. It could be that the
manufacturing process envisioned when developing the high-volume estimate does not work out
as well as conceived. Component costs thus may indeed fall below those currently predicted at
 -38-

high production volume, but it is also possible that even with additional manufacturing
experience, unanticipated problems may mean that the high-volume predictions are not realized.
In the analysis below, we provide some evaluation of the opportunities for additional cost
reductions through learning. Generally, we believe that the high-volume cost estimates in the
studies we reviewed have squeezed costs down about as far as possible given the material
requirements of the product. Thus, we think it unlikely that costs will fall below the high-volume
cost estimates here absent technological advances that currently cannot be foreseen.
Our analysis combines estimates of the relation between production volume and cost and
expected production volume between 2003 and 2007 to predict component costs during the first
five years of the program. We think it likely that these cost projections are reasonably realistic
for what will be observed in the near term. We then predict costs at high production volumes.
Forecasting technological advances is very difficult beyond 10 years or so. We thus think it
appropriate to interpret our high-volume cost estimates as the lowest costs that can be expected
over the next 10 years or so given what is currently known about advanced vehicle technologies
and manufacturing processes.

4.3 COMPONENT COSTS OF BATTERY-POWERED ELECTRIC VEHICLES AND

GASOLINE HYBRID ELECTRIC VEHICLES
This subsection synthesizes the results of existing studies to estimate the costs of the
components in BPEVs and GHEVs. Subsequent subsections address the costs of DHFCV and
PZEV components.

Battery Modules and Auxiliaries

NiMH Battery Modules. CARB’s Battery Technology Advisory Panel (BTAP) assessed
the cost, performance, and availability of batteries for EVs and issued a report in 2000
(Anderman, Kalhammer, and MacArthur, 2000). It based its analysis on data from battery
manufacturers, automakers, existing research reports, and its own analysis. It concluded:

From the cost projections of manufacturers and some carmakers, battery module
specific costs of at least $350/kWh, $300/kWh and $225-250/kWh can be estimated
for production volumes of about 10k, 20k and 100k battery packs per year,
respectively. To the module costs, at least $1,200 per battery pack (perhaps half of
that sum in true mass production) has to be added for the other major components
of a complete EV-battery, which include the required electrical and thermal
management systems (Anderman, Kalhammer, and MacArthur, 2000, p. v).

The Panel assumed only incremental technological improvements (p. vi), so it based its
projections on what is known about the technology today. The declines in unit cost with
production volume are due to “economies of scale that result from discounts on bulk purchases of
 -39-

materials and components, higher efficiencies in the use of labor and equipment, and especially,
use of custom-designed automated manufacturing equipment with high production rates and
product yields” (p. 23). The Panel believes that further reductions in module costs below $225 to
$250 per kilowatt-hour are possible with sustained production on a true mass production scale
(100,000 or more battery packs per year).3 These reductions would be due to additional
automation, incremental improvements in battery design, and process technology refinement
based on accumulated production experience. The Panel did not predict how low costs could go,
but it concluded that reductions in costs below $225 to $250 per kilowatt-hour would be possible
only if material costs declined significantly (p. 24).4 We thus conclude that additional learning
by experience will produce only limited cost savings given what is currently known about the
physics of battery development.
Figure 4.2 displays a curve fit to approximate the BTAP estimates. Data presented by
Panasonic EV at the 1999 UC Davis conference are also included in the figure. In our analysis of
EV costs at high production volumes, we allow NiMH module costs to vary between $225 and
$250 per kilowatt-hour.
Lead Acid Battery Modules. The BTAP cites a range of $100 to $150 per kilowatt-hour
for the production costs of lead acid (PbA) batteries at volumes of 10,000 to 25,000 packs per
year (Anderman, Kalhammer, and MacArthur, 2000, p. iv). Lipman (1999b) estimates that PbA
battery costs are between $107 and $113 per kilowatt-hour for high volumes (>100,000 full-
function battery packs per year). Data provided to AC Propulsion and CARB by a major EV
battery manufacturer (the name is not publicly available) also provide a half-dozen data points.
We used these data to fit a relationship between cost and production volume (see Figure 4.3,
where the data represent the cost of a 60-Ah battery module).
In the early years of the ZEV program, manufacturers may produce 5,000 or fewer full-
function EVs. If the entire 5,000 full-function EVs were produced with PbA batteries, annual
battery production would be 150,000 kWh. The resulting battery cost from Figure 4.3 would be
roughly $150 per kilowatt-hour. Figure 4.3 suggests that $100 per kilowatt-hour is a plausible
high-volume cost for PbA battery modules.
Table 4.4 gives battery costs for two production scenarios between 2003 and 2007 and for
high-volume production. The high full-function EV scenario (2003-2007) accounts for an
average annual production of 16,600 vehicles between 2003 and 2007; the low full-function EV
____________
3Note that there are multiple battery modules in a battery pack.
4The Panel concluded that major advances or breakthroughs that could reduce battery costs were
unlikely for the next six to eight years—that is, through 2006 or 2008 (Anderman, Kalhammer, and
MacArthur, 2000, p. vi).
 -40-

600
BTAP estimates
Panasonic EV estimates

500

400
$ per kWh

300

200

100
0 1.5 3.0 4.5 6.0 7.5 9.0 10.5

kWh per year (millions)

Figure 4.2—NiMH Module Cost as a Function of Production Quantity

400

Estimates from a major

350 battery manufacturer
Lipman (1999b) estimates
300 BTAP estimates

250
$ per kWh

200

150

100

50

0
0 100,000 200,000 300,000 400,000 500,000 600,000 700,000 800,000

kWh per year

Figure 4.3—PbA Module Cost as a Function of Production Quantity

 -41-

Table 4.4
EV Battery Module Costs
($ per kWh)

Low Full-
High Full-Function Function EV Cost in High-
EV Scenario Scenario Volume
Battery Type 2003-2007 2003-2007 Production
PbA 103 174 100
NiMH 356 465 225-250
scenario accounts for an average annual production of 4,700 vehicles (see scenarios 1 and 2 in
Section 3).

NiMH Batteries for GHEVs. Batteries for GHEVs require a higher power density than
those for full-function EVs and city EVs. Consequently, these batteries are costed as a function
of power. NiMH GHEV batteries provide approximately four times the power density of NiMH
EV batteries. (Current designs from Sanyo and Ovonic claim to achieve a power density of 1000
watts per kilogram of battery weight.) According to a recent report, the cost to auto
manufacturers is $500 to $2,000 per GHEV battery pack, with a life expectancy of five to eight
years (PRNewswire, 2001). Given the power ratings on the current fleet of GHEVs, that equates
to roughly a range of $20 to $80 per kilowatt. Lipman, Delucchi, and Friedman (2000) develop a
range of $22 to $31 per kilowatt with future manufacturing experience taken into account. We
use $20 per kilowatt as a high-volume cost for a NiMH GHEV battery.
Battery Auxiliaries. Battery auxiliaries must be added to the battery modules to create a
complete pack. They include (1) a battery tray and straps, (2) an electrical wire harness, (3) bus
bars and terminals, and, perhaps, (4) a cooling/ventilation system. Delucchi et al. (2000) provide
information on the cost of battery auxiliaries. Based on this study, we put the costs of auxiliaries
at $14 to $19 per kilowatt-hour for a PbA battery pack and $8 to $11 per kilowatt-hour for a
NiMH pack between 2003 and 2007. As quoted above, the BTAP puts the cost for a 30-kWh
full-function EV pack at $1,200 but believes that this figure could be halved in high-volume
production. We therefore interpret the Panel estimates to be between $600 and $1,200 depending
on production volume, equivalent to $20 to $40 per kilowatt-hour of pack storage capacity. We
combine these findings to estimate the costs of battery auxiliaries as $8 to $40 per kilowatt-hour
for NiMH packs and $14 to $19 for PbA packs; these estimates hold for the different types of
EVs that may be used to satisfy ZEV program requirements. The lower ends of the ranges are
used to develop high-volume cost increments for the entire EV (see Table 4.5). While not
insignificant, the costs of battery auxiliaries are much less than the costs of battery modules.
 -42-

Table 4.5
EV Battery Auxiliary Costs
($ per kWh)

Volume
Type 2003-2007 Production
PbA auxiliaries 14-19 14
NiMH auxiliaries 8-40 8

Motors and Controllers

Two types of motors and motor controllers are used by major original equipment
manufacturers (OEMs) for BPEVs: AC and DC motor/motor controllers. It is accepted in the
literature that motor costs are a function of peak power requirements but that motor controller
costs are not as closely tied to the motor’s peak power. Generally, AC motors are less expensive
than DC motors, but the opposite holds for motor controllers (since DC controllers require fewer
power switching modules).
Our estimates of motor and motor controller costs, as well as the estimates discussed below
for EV transmission costs, are largely based on the work of Lipman (1999a) and Lipman,
Delucchi, and Friedman (2000). A brief description of the methods used in these analyses is
warranted.
Lipman estimates component costs for different production volumes by estimating
parameters for material costs, costs of adding values to materials (e.g., labor and equipment
costs), and manufacturer profit (Lipman, 1999a, p. 44). The parameters are estimated in part
using data provided by component manufacturers. His estimates are also based on work done at
Argonnne National Laboratory (ANL). ANL has extensively analyzed the relationships among
material, labor costs, production equipment, and other costs in conventional vehicles produced at
high volumes, and Lipman uses these multipliers to help estimate EV component costs at high
volumes. He develops high, central, and low estimates of cost at each production volume, which
in the case of motor controllers reflect “different assumptions about the degree to which costs of
key motor controller components will be reduced by production volume” (Lipman, 1999a, p. 44).
We use Lipman’s central cost estimates. He estimates production costs for 2,000, 20,000, and
200,000 units per year. We use curves fit to these data to predict costs between 2003 and 2007.
For high-volume production, we use the equations to predict costs when annual production
volume is 500,000 units.
Lipman’s estimates clearly capture technological scale effects, and his projections start
with detailed descriptions of the parts and costs of current technologies. Thus, his cost estimates
appear to be based on technological improvements that are reasonably likely. His high-volume
 -43-

cost projections consider ratios between materials cost, overhead, and other costs that are based
on volume production of conventional vehicles. Thus, he is incorporating reductions in
manufacturing costs that can be expected with increased volume based on past experience.
Additional cost reductions may be possible as manufacturers gain experience with the production
process. On the whole, however, Lipman’s description of his approach suggests to us that his
analysis contains enough optimistic assumptions that make it unlikely that learning will cause
costs to fall beyond our high-volume cost estimates for the next 10 years or so absent
unanticipated technological improvements.5
AC Induction Motor and Controller. Following Lipman, Delucchi, and Friedman
(2000), we assume that the cost of AC induction motors is not sensitive to further increases in
production volume and is $10.80 per kilowatt-hour of rated peak power from 2003 to 2007.
Lipman argues that existing production of AC motors is large enough that cost is steady across
production volumes: The tooling for these motors is in place, and they are produced on flexible-
flow production lines (Lipman, 1999a, p. 45).
We follow Delucchi et al. (2000), who cite the data from Lipman and SatCon Technology
Corporation in developing costs for AC controllers for several production quantities. The cost
estimates when production volume is 200,000 units a year are based on data from SatCon.
SatCon has recently received funding from the Department of Energy to develop EV motor and
controller components. SatCon seeks to reduce controller manufacturing costs by selecting low-
cost material, integrating subsystems to reduce parts counts, and utilizing low-cost production
techniques (Lipman, 1999a, p. 43). SatCon’s projections at 10,000 to 200,000 units per year
represent substantial cost reductions over costs today (from costs over $3,300 today to $500 to
$700 for a 50-kW-peak motor). In principle, additional cost reductions are possible with
manufacturing experience, but it seems likely that in developing its cost targets, SatCon has
squeezed margins above material costs very hard.
Delucchi et al. (2000) include a cost component that is a function of power rating and a
component that reflects internal elements of the controller that are not necessarily cost-sensitive
to the power rating but are cost-sensitive to production volume increases. Based on these data,
we fit the following equation for AC controller costs:

$7.10 to $9.00 per kWmotor-peak + y, where

-0.51
y = 149,255x and x is the annual unit production volume.

____________
5For example, his central-cost projections for controllers are based primarily on target data for
costs from SatCon Technology Corporation (Lipman, 1999a, p. 47).
 -44-

DC Permanent Magnet Motor and Controller. Lipman, Delucchi, and Friedman (2000)
surveyed the literature on DC permanent magnet motors (including Cuenca, Gaines, and Vyas,
1999) and determined costs on the basis of both peak kilowatt rating and production volume. His
estimates run from $12.51 to $29.69 per peak kilowatt depending on production volume, not
including additional power-invariant costs. Lipman adds $779 for production volumes as low as
2,000 per year and $88 for production volumes around 20,000 per year.6 His estimates are shown
in Figure 4.4 for three different production volumes; we can interpolate for production quantities
between these levels.
The material costs used in Lipman’s estimates of DC permanent magnet motors are
generally based on Cuenca, Gaines, and Vyas’s analyses (Lipman, 1999a, p. 45). Cuenca and his
colleagues at ANL have developed detailed material costs and high-volume manufacturing costs
for permanent magnet motors. At high volume, material costs account for 60 to 70 percent of
overall component cost per unit (Lipman, 1999a, p. 75). Thus, absent technological changes in
the design, there is not much room for reductions in manufacturing costs through manufacturing
experience.
Our estimates for DC permanent magnet controller costs are drawn from the same sources
as the AC controller costs discussed above. The resulting equation is

$4.80 to $6.90 per kWmotor-peak + y

-0.53
where y = 162,123x , and x is the annual unit production volume.

Motor and Controller Costs Used in Our Analysis. For 2003 to 2007, we combine the
above equations with estimates of production volume from Subsection 3.2 to predict motor and
controller costs.
Cuenca, Gaines, and Vyas (1999) provide a starting point for estimates of high-volume
production costs. As shown in Table 4.6, the materials in their DC system (i.e., motor and
controller) cost $1,190 to $1,360, and the materials in the AC system run from $1,316 to $1,516
for the size of the motor indicated. For the 50-kW motor used in our analysis, these estimates
translate to between roughly $1,200 and $1,500 for the motor and controller costs combined.
Since these estimates were made, the costs and sizes of motor controller parts have declined. GM
reported in 1999 that it had cut the cost of its EV1 drivetrain (motor and motor controller) in half,
reduced its size by half, and cut the number of parts by one-third as part of its “Generation II”
design. At one point, GM’s Advanced Technology Vehicle center was working on “Generation
____________
6These power-invariant costs can be approximated with the function 687,000x-0.90, where x is the
annual production volume.
 -45-

60

2,000 units per year

20,000 units per year
50
200,000 units per year

40
$ per kW

30

20

10

0
25 45 65 85 105 125
kW of peak power

SOURCE: Based on Lipman, Delucchi, and Friedman, 2000.

Figure 4.4—DC Motor Cost as a Function of Production Quantity

III” drive components that were to be half again the cost of the Generation II components (EV
World, 1999). CARB notes that for the GM Precept, equipped with Generation III drivetrain
components, a key component of the power inverter (motor controller) module is one-sixth the
size of the one used in the EV1. Honda has published similar statements in regard to electric
motors and controllers. Therefore, it seems likely that the material costs in Cuenca, Gaines, and
Vyas, 1999, are high given the technological improvements that have already occurred since their
study.
We extrapolate from the relationships between production volume and cost above to
predict motor and controller costs at high volume. The relationship between combined motor and
controller costs and production volume for a 50-kW motor is presented in Figure 4.5. The
combined cost at 500,000 units lies between $725 and $1,200 (compared to $1,200 to $1,500 in
Cuenca, Gaines, and Vyas, 1999). We use this range in our high-volume production scenarios.
As a city EV system is rated at half the power of a full-function EV system, we use a figure that is
roughly half these estimates for the city EV motor and controller in high-volume production. The
motor in the DHFCV that we cost out is rated at 75 kW. We thus increase the range estimated for
a 50-kW system by 50 percent: The resulting range is $1,100 to $1,800.
 -46-

Table 4.6
Estimate of High-Volume Costs for 53-kW DC Motor, 67-kW AC Motor,
and 70-kW DC and AC Motor Controllers
($)

Cost Driver DC System AC System

Motor
Magnets 175 0
Other materials 215 291
Assembly and test 97-167 125-194
Total mfg. cost 487-557 416-485
Gross margin (at 20) 97-111 82-97
Total cost 584-668 498-582
Cost per kW 11-12.6 7.4-8.7
Motor Controller
Switching modules 200-300 400-600
Other components 500 500
Assembly costs 100 100
Total mfg. cost 800-900 1000-1200
Supplier margin (at 20%) 160-180 200-240
Total OEM cost range 960-1080 1200-1440
Cost per kW 13.7-15.4 17.1-20.6
Total Motor and Controller ($/kW) 24.7-28.0 24.5-29.3
SOURCE: Cuenca, Gaines, and Vyas, 1999.
Summary of Motor and Controller Costs for BPEVs. Table 4.7 summarizes our
estimates of the average motor and motor controller costs. The second column predicts costs
between 2003 and 2007 for the high-volume scenarios (see Section 3); the third column predicts
cost for the same period for the low-volume scenarios. The final column lists high-volume
production costs.
Learning through manufacturing experience could conceivably reduce costs below the
volume production costs listed here. However, we have been aggressive about continuing to
reduce costs as volume increases (production scale economies) and thus think it unlikely that
learning will reduce costs further over the next 10 years or so given what is known about the
technology and manufacturing processes.
Motor and Controller Costs for GHEVs. Energy and Environmental Analysis estimates
the volume production costs for the Toyota Prius (Duleep, 1998). It put the motor controller at
$704 and its 33-kW electric motor at $147. Accordingly, we use the sum ($851) as our upper
bound for high-volume cost for a GHEV motor and motor controller. GHEV motor and
controller costs should not exceed those for a full-function EV in the long run. We thus set the
lower bound of the range at $725.
 -47-

3,000

High estimate
2,500
Low estimate

2,000
Cost ($)

1,500

1,000

500

0
0 100,000 200,000 300,000 400,000 500,000
California production volume of EVs

Figure 4.5—Combined Cost of 50-kW Motor and Motor Controller at

Different Production Volumes

Table 4.7
Motor and Motor Controller Costs for Full-Function EVs and City EVs
($ per motor and controller)

High-Volume Low-Volume
Scenarios Scenarios High-Volume
Type 2003-2007 2003-2007 Production
50-kW Full-Function EV
AC 2,000-2,100 3,000-3,100 725-1200
DC 2,100-2,500 3,500-3,850 725-1200
25-kW City EV
AC 1,550-1,600 3,000-3,100 450-550
DC 1,550-1,800 3,500-3,850 450-550

EV Transmission and Other EV Auxiliaries

Lipman (1999b) reports data from Unique Mobility on the cost of a single-speed EV
transaxle for volumes of 2,000, 20,000, and 200,000 units per year.7 Based on those data, we fit
the following equation:
____________
7Unique Mobility considers itself to be a “world recognized technology leader in the development
and manufacture of energy efficient, power dense electric motors, generators and electronic inverters aimed
at high growth and emerging markets” (http://www.uqm.com).
 -48-

-0.30
cost per kWmotor-peak = 315x ,
where x is the annual unit production volume.

For EV auxiliaries (including motors to drive compressors for steering and braking
systems), we use the data in Delucchi et al., 2000 (p. 40), to fit the function
-0.18 -0.20
cost per kWmotor-peak = 7.2x + 0.45x ,
where x is the annual unit production volume.

The costs of these auxiliaries are small, amounting to roughly $2 per kilowatt of peak motor
power when annual production is 2,000 and $0.85 per kilowatt when annual production is
200,000 units.
The high-volume estimate for the EV transmission and auxiliaries is based on predictions
of these equations when volume is 500,000 units per year. It comes to $7 per kilowatt of peak
motor power.

Integrated Charger
For integrated chargers, we use estimates obtained from AC Propulsion (Brooks and Gage,
2001): $1,338, $474, and under $300 for vehicle volumes of 100, 1,000, and 10,000,
respectively, for a conductive system.8 These cost figures fit the following curve:
-0.32
cost = 5417x ,
where x is the annual unit production volume.

This curve predicts a cost of $284 at a volume of 10,000, and we used it to project integrated
charger costs between 2003 and 2007.
AC Propulsion had earlier (in 2000) provided Delucchi et al. with integrated charger cost
estimates (Delucchi et al., 2000, p. 39). The costs were $800 per vehicle when production ranged
from 5,000 to 10,000 units per year. The 2001 estimates are substantially below this and may
reflect technical improvements or projected improvement in manufacturing processes. We have
not been able to investigate the reason for the decrease. Data are not available on high-volume
production costs. The substantial decrease in estimated costs between 2000 and 2001 may
indicate that costs will continue to decrease in the future. But they may also mean that the 2001
estimates have squeezed costs hard (AC Propulsion is a strong advocate of EVs). We conclude
that it is reasonable to use $250 to $300 for high-volume production costs.
____________
8In June 2001, CARB adopted on-board conductive charging as the standardized charging system
for EVs in California, with implementation to be in 2006.
 -49-

Savings From Exclusion of Internal Combustion Engine Vehicle Components

It is important to consider the savings that accrue because several key components of
ICEVs are not needed in EVs. In Cuenca, Gaines, and Vyas, 1999, manufacturing costs represent
50 percent of the manufacturer’s suggested retail price (MSRP). The engine, transmission, fuel,
and exhaust system represent 22 percent of the manufacturing cost (or 11 percent of the MSRP)
for a subcompact, and 20 percent (10 percent of MSRP) for a minivan.9 Based on these
percentages and a review of current MSRPs for subcompacts and minivans, we put costs uniquely
associated with ICEVs at $16 to $18 per kilowatt of rated engine power when nonmanufacturing
costs are excluded and $32 to $36 once nonmanufacturing costs are included.
The cost of ICEV components not required for an EV may well increase. Auxiliary power
requirements are expected to rise to the point that existing auxiliary power systems will have to
be improved for vehicles with internal combustion engines as the sole power plant. One
published estimate puts the cost increment (for more powerful batteries, alternators, and wire
harness) at $1,000 (Sullivan, 2001).
Currently, passenger cars operate between 1 and 2 kW of auxiliaries, but this amount is
rising as amenities such as cell phones, entertainment systems, navigation systems, electric
steering, electric brakes, electric suspension, and electric valves are added. Existing 14-volt
generators on ICEVs are limited to generating around 2 kW maximum. All together,
nonpropulsion-related power needs could rise to between 5 and 10 kW or more (see Table 4.8),
which is comparable to the power needs of a large home.
Cars with 42-volt electrical systems, designed for increased loads, may arrive in the next
several years. These systems entail added costs for a larger battery, more robust wiring, and
advanced switching components. EVs and hybrid EVs are inherently capable of producing large
amounts of electrical power, so it is possible that the cost differential between conventional
ICEVs and EVs and hybrid EVs may narrow as ICEV electrical systems are upgraded. We do
not make an adjustment for this effect in our analysis. First, we have not been able to examine
the basis for the $1,000 estimate above. Second, the extra power requirements will presumably
also require more battery storage (and higher costs) to maintain the same vehicle range.
Our discussion so far concerns the costs of ICEV components that are not required in EVs.
But our cost estimates of GHEVs must also include the cost savings due to the smaller internal
combustion engines that are feasible in hybrids. For example, the Ford Escape GHEV will have a
____________
9In Cuenca, Gaines, and Vyas, 1999, a subcompact has a 4-cylinder engine and a 5-speed
transmission; and a minivan has a 6-cylinder engine and a 4-speed transmission.
 -50-

4-cylinder internal combustion engine that performs like a conventional 6-cylinder power plant.
The increment in MSRP from the 4-cylinder to the 6-cylinder is $1,400 according to Ford’s price
Table 4.8
Breakdown of Current and Potential Power Needs of an ICEV

Load Peak (watts) Average

Electromechanical valves (6 cyl @ 6,000 rpm) 2,400 800
Water pump 300 300
Engine cooling fan 800 300
Power steering (all elec.) 1,000 100
Heated windshield 2,500 200
Catalytic converter pre-heat 3,000 60
Active suspension 12,000 360
Comm/nav/entertainment -- 100
Total -- 2,220
SOURCE: Kassakian, 2000.
guides. We assume that the MSRP increment is twice the increment of the engine cost, putting
the reduction in engine cost at $700.

4.4 COMPONENT COSTS OF DIRECT HYDROGEN FUEL-CELL VEHICLES

In this subsection, we first provide an overview of studies on projected proton exchange
membrane (PEM) fuel-cell system costs. We then examine the costs of key fuel-cell components
and their cost drivers.

Overview of Recent Studies on PEM Fuel-Cell Systems

The cost today of fuel cells is very high. Current estimates for a hydrogen/air fuel-cell
system run from $500 (Arthur D. Little, Inc. [ADL], 2001, p. 17) to $2,500 per kilowatt (U.S.
Department of Energy [DOE], 1998, as cited in Lipman, Delucchi, and Friedman, 2000, p. 18).
This translates into $30,000 to $150,000 for the 60-kW systems that will be needed for standard-
size vehicles. However, many analysts project that fuel-cell costs will drop dramatically because
of design and manufacturing improvements and volume production. For example, Ford Motor
Company and Directed Technologies, Inc. (DTI) detail a product specification and manufacturing
process that result in costs of $33 to $43 per kilowatt (fuel-cell stack and auxiliaries combined) at
high-production volumes (Oei et al., 2000). The following paragraphs describe the key studies;
Table 4.9 summarizes their findings.
Arthur D. Little (ADL) 2001 Study. This study employs the Delphi method to solicit
opinions from experts on the cost and performance of PEM fuel-cell vehicles over the next 30
years. It provides costs for a fuel-cell system that could use hydrogen generated from gasoline by
an on-board fuel reformer. Such fuel cells require higher loadings of precious metals because of
 -51-

impurities that remain after the reforming process. The report concludes that future costs for a
complete system (stack and auxiliaries) could drop to $111 per kilowatt. The key cost driver with
Table 4.9
Summary of Studies on Fuel-Cell Costs

Platinum
Loading
Study Sponsor (mg/cm2) High-Volume Cost
Arthur D. Little, 2001 California Energy Commission 0.8a $111/kW for stack plus
auxiliaries
Fuel Cell Technical CARB 0.25 $20/kW for stack;
Advisory Panel, 1998 $15/kW for auxiliaries
Oei et al. (Ford/DTI), 2000 Ford and U.S. DOE 0.25 $19-$28/kW for stack
Lipman, Delucchi, and Kirsch Foundation and Union of 0.25 $39-$43/kW for stack
Friedman, 2000 Concerned Scientists plus auxiliaries
aThe ADL design requires rhodium as well as platinum.

regard to fuel cells is the amount of precious materials required. As shown in Table 4.9, the ADL
study assumes 0.8 grams of platinum and rhodium per square centimeter of active fuel-cell area.
According to the ADL study (p. 17), experts from Ballard, Ford, and DaimlerChrysler
have predicted the cost of the Ballard Mark 900 fuel-cell module to be $50 to $60 per kilowatt in
large volumes (approximately 300,000 units per year). This range suggests that the platinum
loading in the Ballard design is substantially less than 0.8 gm/cm2.

Kalhammer et al. (Fuel Cell Technical Advisory Panel, FCTAP) 1998 Study. CARB
established the FCTAP in 1996 to assist it in assessing the status and prospect of fuel cells for
auto use. A study was completed, and a report was issued in July 1998. The research method
was to “acquire information from responses to a widely distributed questionnaire and in visits
with organizations considered leaders in key aspects of fuel cell electric engine and vehicle
development.” The panel concluded:
The Panel’s visits and discussions with leading developers and potential manufacturers of
PEM fuel cell components and stacks made clear that the fundamental technical barriers
to the development of automotive fuel cell stacks have been overcome by the advances
achieved over the past 5–7 years. The large increases in the specific performance of PEM
cells and stacks also have lowered the cost barriers to the point where future mass
production may be able to meet the stringent cost goals for critical cell components and
stacks intended for automotive applications (Kalhammer et al., 1998, p. III-21).
The FCTAP went on to predict:

[On] the basis of the performance already achieved with preprototype stacks, there appear to
be reasonable prospects for meeting the $20/kW ($1000/50kW) cost target for automotive
fuel stacks if production reaches about 100,000 to 200,000 units per year. At this level, the
 -52-

most critical stack components—membrane, MEA and separator plate—will reach production
volumes that justify true mass manufacturing methods (Kalhammer et al., 1998, p. III-54).
 -52-

Oei et al. (Ford/DTI) 2000 Study. This study considers four detailed designs and
assesses the cost based on high-volume production (500,000 units per year) with cost reductions
gained by applying DFMA (design for manufacturing and assembly) techniques. The DFMA
process has been formally adopted by Ford as a systematic means for the design and evaluation of
cost-optimized components and systems. It combines historical cost data and manufacturing
acumen accumulated by Ford since the earliest days of the company (Oei et al., 2000, p. 3-1).
Ford/DTI project that costs for PEM stacks will range from $19 to $28 per kilowatt in
high-volume production. The range is due to the overall stack power rating (e.g., 60 to 80 kW).
Costs include material, manufacturing, and assembly costs, as well as markups to reflect profit,
scrap, R&D, and administrative costs.
As we discuss in more detail below, the Ford design incorporates a number of well-
researched but still untested features. Learning through manufacturing experience could reduce
costs further, but Lipman notes that “the estimation methodology used by DTI was specifically
designed to identify the lowest cost PEM stack design configuration, and the choice of a
production volume of 300,000 units per year suggests that it would be difficult to construe a
lower cost case” (Lipman, Delucchi, and Friedman, 2000, p. 19). In Ford and DTI’s view, further
cost reductions will require important technological advances. They conclude that further cost
reductions will require a reduction in platinum catalyst loading or in gas diffusion electrode costs,
because the costs of other components are based on “mature” manufacturing technologies (Oei et
al., 2000, p. 3-47).
Lipman, Delucchi, and Friedman 2000 Study. To project fuel-cell costs in the near
term, Lipman, Delucchi, and Friedman use an experience curve, which is a way to relate unit cost
(in this case $/kW) to cumulative production. They start with a cost of $2,000 per kilowatt and
5,000 kW of cumulative production in 2006 (p. 19); costs then decline 22 percent for each
doubling of production (this corresponds to a 78 percent experience curve). They note that
experience curves typically range from 70 to 90 percent.10 Costs decline until they reach a lower
bound based on DTI’s estimates. Lipman, Delucchi, and Friedman reduce cell peak power
density roughly 15 percent from that assumed by DTI and arrive at a lower bound of $39 to $43
per kilowatt for the stack and auxiliaries combined.
Figure 4.6 illustrates the experience curve of Lipman, Delucchi, and Friedman for a 25-kW
PEM fuel-cell system. As cumulative production increases, costs fall rapidly and approach $60
per kilowatt.
____________
10Note that these experience curves are for the successful products. For many products, the
technology or costs did not allow commercial production that was long enough for an experience curve to
be observed.
 -53-

250

200

150
$ per kW

100

50

0
25,000 75,000 125,000 175,000 225,000 275,000
Worldwide cumulative production of DHFCVs (in thousands)

SOURCE: Based on Lipman, Delucchi, and Friedman, 2000.

Figure 4.6—PEM Fuel-Cell System Costs as a Function of

Accumulated Production Quantity

Fuel-Cell Components
Membrane Electrode Assembly Cost and Design. The biggest contributor to the cost of
a PEM fuel-cell stack is the platinum (which serves as a catalyst). Platinum accounts for a
significant portion of the MEA cost in any design. For the design specified in the Ford/DTI study
(Oei et al., 2000), the cost of the platinum represents 75 percent of the total cost of the MEA.
Table 4.10 compares the platinum loading in the MEA for the relevant studies. The ADL
2001 study assumes a design with a large amount of platinum (0.8 mg/cm2), which easily
accounts for half of the MEA cost. All the other studies assume far less for future-year designs.
DOE claims a breakthrough of 0.15 mg/cm2 for experimental MEAs in laboratory quantities from
a 3M design.11 The Ford/DTI study (Oei et al., 2000) assumes 0.25 mg/cm2, and the FCTAP
study (Kalhammer et al., 1998) cites designs that range from 0.25 to 0.35 mg/cm2. A Los Alamos
National Laboratory study (LANL, 2002) describes PEM fuel cells with low platinum loads (0.11
and 0.15 mg/cm2) that have been tested to 4,000 hours. One reason that the ADL study’s loading
____________
11In June 2001, DOE announced that 3M was selected for a $7.7 million cost-shared contract to
develop MEAs with improved cathodes, high-temperature membranes, and optimized gas diffusion layers.
Improved flow fields and MEA fabrication processes will also be developed.
 -54-

Table 4.10
Platinum Loading in Membrane Electrode Assembly

MEA Cost
Platinum in High
Loading Volume
Study (mg/cm2) ($/kW) Comment
Arthur D. Little, 2001 0.8 -- Assumes platinum is $13.50/gm
Oei et al. (Ford/DTI), 2000 0.25 10-11 Assumes platinum is $12.86/gm
Los Alamos National Laboratory, 2002 0.11-0.15 -- --
Kalhammer et al. (FCTAP); 1998. Designs reviewed:
Gore 0.3 -- --
Johnson Matthey 0.35 <15 --
3M 0.25 5-10 Assumes production of >1 million m2/yr
is much higher is that it is based on a fuel cell that can accept hydrogen reformed from gasoline
or methanol. Such fuel cells likely require more platinum and also rhodium to protect against
impurities that remain after the reforming process. A DHFCV is less susceptible to this type of
catalyst poisoning, so lower catalyst loadings are possible.
These studies suggest that it is highly likely that an MEA with 0.25 mg/cm2 can be built
for automotive use, and we have adopted this catalyst loading in our analysis. Information on the
durability of such cells is not publicly available. It is possible that the Ballard fuel-cell system
currently being tested by Toyota and GM and the fuel cell in Honda’s vehicles contain levels
approaching 0.25 mg/cm2, but such data are not publicly available. Even if the levels in these
cells are near 0.25, their real-world durability is still being investigated. We thus cannot be
entirely sure that fuel cells with 0.25 mg/cm2 of platinum will work for automotive use. Designs
with even lower loadings, however, suggest that 0.25 mg/cm2 is not a wildly optimistic
assumption.

The low MEA cost in the Ford/DTI study (Oei et al., 2000) is based on a number of other
design features that are well documented and researched but have not yet been produced. For
example, Ford/DTI propose a composite polymer electrolyte membrane rather than conventional
homogeneous membranes (p. 3-5). The projected cost savings are 98 percent of current
homogeneous membrane costs. The electrodes in the Ford/DTI study are constructed from an
advanced carbon paper that is still under development (p. 3-11). Highly automated
manufacturing techniques are used that might not pan out as well as envisioned. Ford/DTI did,
however, build some cushion into their estimates. For example, projected costs are inflated by 10
percent to ensure “appropriate conservatism” (p. 3-44), and the lowest projected advanced carbon
paper costs were not used (p. 3-14). Ford/DTI believe their projections are based on “accepted
manufacturing techniques and realistic technical parameters” (p. 3-19). In their view, the
projections would require a minimum of technological advancement (p. 3-46).
 -55-

Ford/DTI conclude that the MEA for a direct hydrogen fuel-cell stack can be built for $10
to $11 per kilowatt at high-volume production. CARB’s FCTAP concludes that costs could run
from $5 to $10 per kilowatt.
Bipolar Plate Cost and Design. The role of the bipolar plate (also called the separator
plate) is to provide support for the fuel-cell membrane as well as to uniformly distribute the
hydrogen on one side of the membrane and the oxygen on the other side. The plate also has a
role in conducting heat to allow for proper cooling. It must be corrosion resistant and able to
operate under pressure (Dayton, 1999; Bulk Molding Compounds, Inc., 2001).
The FCTAP 1998 study cites a variety of approaches and designs for bipolar or separator
plates, as well as uncertainty as to the cost of this component. A number of different types of
materials could be used for the bipolar plates, but the four main considerations are
1. corrosion-resistant metals
2. coated metals
3. carbon/graphite
4. compatible polymers.
The Ford/DTI study assumes a metallic bipolar plate made of 316 stainless steel (which
contains high percentages of chromium, nickel, and molybdenum). This study considers two
types of cell construction: a three-piece design, and a cheaper, unitized (one-piece) separator
plate. An earlier ADL study (ADL, 2000) assumes the use of a molded graphite/polymer
composite bipolar plate; the plate is molded into two pieces and bonded together.
The FCTAP study does conclude that at least one developer could provide a design that
allows the cost of the bipolar plate to be $5 per kilowatt:

No stack developer appears to have made a final choice, but several approaches
look promising, including embossing of impregnated porous carbon, molding of
commercially available carbon composites (Energy Partners), embossing of
coated metal plates (Allied Signal; Siemens); and perhaps bonding of
appropriately shaped metal sheets (H-Power). At present, confident estimates of
separator plate costs are still lacking. However, the approaches under
development were all selected for their potential to permit low-cost mass
manufacturing of plates from inexpensive materials, lending credibility to the
$5/kW cost projected by a leading developer (Kalhammer et al., 1998, III-24).

Oak Ridge National Laboratory reports that it is working on a design envisioned to meet a
cost goal of $10 per kilowatt. The design being considered is one with thin metallic plates
(thinner than a few millimeters) that uses carbon composites instead of graphite, which is more
costly to work with (Fuel Cell Catalyst, 2001).
 -56-

Fuel-Cell Auxiliaries. Auxiliaries are devices that manage air, water, and other thermal
issues. In addition, control electronics are required. The cost of the auxiliary equipment
supporting a fuel-cell stack has to be considered. Fewer studies project costs for this part of a
fuel-cell system. Auto industry goals are $20 per kilowatt. DTI (1998) assesses them to be $14
to $15 per kilowatt; ADL concludes that the balance of plant costs in high-volume production for
a gasoline-fueled PEM fuel cell would be $10 per kilowatt of stack power (ADL, 2000).
Summary of Fuel-Cell System Costs. It seems likely that DHFCVs will be available in
only very small numbers through 2005 (see Subsection 3.1). We base our fuel-cell system costs
between 2006 and 2010 on the manufacturing experience curve developed by Lipman, Delucchi,
and Friedman (2000). Our estimates of fuel-cell system costs for this period run from $100 to
$150 per kilowatt. For high-production volumes, we use $35 to $60 per kilowatt. The lower
bound reflects $20 per kilowatt for the fuel-cell stack and $15 for the auxiliaries. Ford/DTI (Oei
et al., 2000) predict stack costs between $18 and $28, and we use $20, which is the low cost for
the intermediate-size stacks likely in automotive applications. The upper bound is based on
estimates that the Ballard Mark 900 fuel cell will cost $50 to $60 per kilowatt in volume
production.
Our review of the literature led us to conclude that the projections that fuel-cell system
costs can fall to $35 to $60 per kilowatt in volume production are well grounded. Costs might be
even lower if the very low platinum designs currently being investigated pan out. However, it is
too early to tell if this will be the case. Manufacturing experience may also reduce costs,
although additional reductions may not be large, because most of the nonmaterial costs have been
squeezed out by mature manufacturing techniques. However, a number of uncertainties remain.
Unexpected technological or production problems may arise, and the durability of automotive
fuel cells has not been extensively tested in demanding real-world environments. We use this
range for volume production costs in our analysis but note that much remains to be done to
demonstrate that these low production costs can actually be realized.

Hydrogen Tank for a Fuel-Cell Vehicle

We considered the cost of the “novel polymer-lined cylinders” described by Thomas et al.
(1999). Lipman, Delucchi, and Friedman (2000) estimate their cost to be $510 per kilogram of
hydrogen stored at 350 atmospheres (5000 psi) based on current prices for compressed natural gas
tanks. Lipman cites a DTI study on high-volume production of such tanks to estimate a lower
bound cost of $84 to $163 per kilogram of stored hydrogen. This translates to $336 to $652 for a
tank that stores 4 kg. We use this cost range in our estimates of high-volume costs. For the near
 -57-

term, we use the relation between cost and cumulative volume postulated by Lipman, Delucchi,
and Friedman (2000).12 Figure 4.7 shows the resulting relationship for a 4-kg tank.

4.5 PZEV COSTS

CARB reviewed various technological approaches to PZEVs and concluded that the
incremental cost relative to a vehicle meeting the SULEV exhaust and near-zero evaporative
standards was minor:
[C]onfidential information from several manufacturers indicates that PZEVs soon
to be introduced for sale in California will use a simpler and much less costly
combined HC adsorber/catalyst rather than a separate adsorber and an attendant
switching valve as had been assumed earlier. Further, the new information
indicates that additional catalyst volume will not likely be required as was the
case in the first PZEV system certified for sale in California. . . . Manufacturers
will still face some increased cost to build increased durability into the emission
control components (e.g. increased catalyst loading) in order to avoid excessive
repair costs during the 150,000 mile emission warranty period. . . . Based on the
information acquired by staff, some additional carbon trap capability will be
added, along with improved seals and reconfiguration of some components that
do not add large cost. The cost of going to a zero evaporative system from the
near zero systems is now estimated to be $10 per vehicle. Taking all of these
factors into account, staff now estimates that the necessary hardware
modifications to meet PZEV requirements will range from $60 to $85 per vehicle
(CARB, 2001d, p. 5).
These estimates are for passenger cars and the smallest category of light-duty trucks (LDT1).13
The additional warranty requirements of PZEVs will range between $125 and $150 per
vehicle, according to CARB estimates. It is unlikely that all these costs represent incremental
costs of PZEVs. To some extent, the warranty causes a transfer of the costs of maintaining the
vehicle from consumers to producers and thus does not change the overall resource costs of
PZEVs. PZEVs may, however, cost more to maintain over their lifetimes because of the tighter
emission standards. The proportion of the warranty that represents a transfer of costs from
consumers to producers also depends on the stringency of California’s inspection and
maintenance (Smog Check) program: A less stringent program requires fewer problems to be
repaired in the absence of a warranty, so a larger fraction of the warranty costs are true resource
costs. Given these uncertainties, we vary resource costs due to the extended warranty from $0 to
$150 per vehicle. This range assumes from none to all of the increased warranty costs represent
transfers from consumers to automakers. Thus, our estimate of the incremental cost of a PZEV

____________
0.2318
12Tank cost = 321x- , where x is cumulative production volume.
13Paul Hughes, CARB, personal communication, May 2002.
 -58-

800

700

600

500
Tank cost ($)

400

300

200

100

0
0 1 2 3 4 5 6
Cumulative production volume (in millions)

SOURCE: Based on Lipman, Delucchi, and Friedman, 2000.

Figure 4.7—Cost of Tank That Stores 4 kg of Hydrogen as a Function

of Production Quantity

relative to a SULEV meeting the near-zero evaporative emission standards ranges from $60 to
$235.
We have not been able to examine in detail the basis for CARB’s PZEV cost estimates.
Earlier CARB estimates were higher. In August 2000, CARB put the incremental costs of a
PZEV (relative to a SULEV) at $500 (CARB, 2000b, p. 106). The design improvements outlined
above make a plausible case for the reduced estimates, but further analysis is warranted.

4.6 INCREMENTAL VEHICLE PRODUCTION COSTS

So far we have discussed the costs of individual vehicle components. To calculate the
incremental production costs of vehicles that manufacturers may use to meet ZEV program
requirements, we subtract the costs of components associated with the conventional ICEV and
add in the costs of components associated with each vehicle technology.14 Indirect costs
associated with each of the components are also dropped and added, and we account for changes
in these costs (e.g., corporate overhead, warranty, and normal profit). Here, we first detail the
____________
14As an alternative to this approach, one could also build costs of various vehicles from the ground
up and then compare the resulting total vehicle costs. Delucchi et al. (2000) use such an approach.
 -59-

indirect cost multipliers used in our analysis. We then present estimates of the incremental costs
of ZEVs, ATPZEVs, and PZEVs.

Multipliers Used to Calculate Indirect Costs

Analysis of indirect costs is usually framed in terms of the difference between
manufacturing cost and final manufacturer’s suggested retail price (MSRP). Vyas, Cuenca, and
Gaines (1998) break down the components of MSRP and show that the direct manufacturing cost
accounts for half of the price (see Table 4.11). A later report (Cuenca, Gaines, and Vyas, 1999)
proposes multipliers to translate vehicle cost to vehicle price. Depending on the component, the
multipliers range from 1.15 to 2.0: 1.15 to 1.3 for the battery pack, 1.5 for all other components
supplied to the auto manufacturer, and 2.0 for components developed solely by the auto
manufacturer (e.g., the internal combustion engine). We use these estimates in our analysis. For
battery packs and fuel-cell systems, 1.15 and 1.30 are used in calculations of the lower and upper
bounds, respectively, of the cost ranges projected for each production volume scenario.15

Incremental Vehicle Production Costs

Estimates of the incremental production costs of the various low-emission vehicles and
ZEVs are presented in Table 4.12. Costs are incremental to a standard ICEV that meets the
SULEV exhaust standard and the near-zero evaporative emission standard. The second column
presents estimates of the range into which incremental production costs are likely to fall on
average between 2003 and 2007 if annual production volumes are at the low end of the ranges
specified in Subsection 3.2. The third column presents the range if annual production volumes
are at the high end of the ranges specified in Subsection 3.2. The last column reports incremental
costs in volume production. Breakdowns of volume production costs for each vehicle technology
are presented in Appendix C, as is an analysis of the sensitivity of the incremental cost of full-
function EVs to changes in peak motor power and battery pack size.
As discussed above, the incremental costs of a PZEV are modest. GHEVs will likely cost
several thousand dollars more to produce during the first five years of the program. Costs come
down in high-volume production but are still expected to exceed standard ICEV costs by $500 to
$900 per vehicle. BPEVs will be much more expensive to produce during the initial years of the
program. Incremental production costs including the cost of the first battery pack for a full-
function EV using NiMH batteries range from $16,700 to $26,100. Incremental costs for PbA
____________
15CARB’s first estimate of PZEV costs included a 10 percent adjustment for indirect costs (CARB,
2000b, p. 106). We assume that CARB’s 2001 estimate also includes indirect costs, but the indirect cost
multiplier is unknown.
 -60-

Table 4.11
Breakdown of Manufacturer’s Suggested Retail Price for an ICEV

Share of MSRP (%)

Cost Category/Subcategory By Cost Category By Subcategory
Production 67.0
Manufacturing 50.0
Warranty 5.0
R&D and engineering 6.5
Depreciation/amortization 5.5
Selling 23.5
Distribution 20.0
Advertising and dealer support 3.5
Administration and profit 9.5
Corporate overhead 5.0
Retirement and health benefits 2.0
Profit 2.5
Total 100.0 100.0
SOURCE: Vyas, Cuenca, and Gaines, 1998.

Table 4.12
Incremental Vehicle Production Costs (relative to an ICEV meeting SULEV exhaust and
near-zero evaporative emission standards) (estimated ranges in brackets; $1000s per vehicle)

Near-Term Costs
Low-Volume High-Volume High-Volume
Vehicle Technology Scenario Scenario Production
2003-2007 2003-2007
PZEV [0.06, 0.23] [0.06, 0.23] [0.06, 0.23]
GHEV [2.4, 3.3] [2.3, 2.9] [0.5, 0.9]
Full-function EV
NiMH [22.5, 26.1] [16.7, 19.8] [5.1, 9.9]
PbA [12.7, 15.2] [8.0, 9.6] [1.8, 4.3]
City EV
NiMH [12.0, 13.4] [8.5, 9.7] [1.7, 3.4]
PbA [8.6, 9.9] [5.3, 6.0] [0.7, 1.5]
2006-2010 2006-2010
DHFCV [11.4, 15.2] [8.3, 11.0] [1.2, 3.9]

batteries and for city EVs are lower but remain very large between 2003 and 2007. Even in
volume production, production costs remain substantially above those for ICEVs. The lowest
incremental cost in high-volume production is $700 to $1,500 for city EVs with PbA batteries.
But these numbers should be used with caution because, as discussed above, there are drawbacks
to designing vehicles with PbA batteries. DHFCVs will also likely be much more expensive
when they are first introduced. Because of low production volumes and the developing
technology, we estimate that the incremental costs of DHFCVs will run from $8,300 to $15,200
between 2006 and 2010. But in the long run, DHFCV technology looks more promising than
BPEV technology does. Projections based on what is currently known about fuel cells show that
 -61-

incremental production costs may come down to $1,200. However, we cannot rule out the
possibility that production costs will remain nearly $4,000 above those of an ICEV.

4.7 INCREMENTAL LIFECYCLE COSTS

So far we have discussed the incremental production cost of advanced technology vehicles,
but their overall social cost also depends on their incremental lifetime operating costs, by which
we mean the cost of operating and maintaining the vehicle relative to the cost of operating and
maintaining a baseline ICEV. We add incremental lifetime operating costs to incremental
production costs to estimate incremental lifecycle cost.
We begin by presenting and justifying the parameters we use to estimate the incremental
lifecycle costs, starting with full-function EVs and city EVs and then moving to DHFCVs and
GHEVs. Any incremental lifetime operating costs of PZEVs (due, for example, to increased
warranty costs) are captured in the production costs estimated above. Thus, the PZEV
incremental lifecycle cost is only the incremental production cost. After presenting the
parameters, we discuss the method we use to predict incremental lifecycle costs using these
parameters. Finally, we present the results of the analysis.

Parameters Used to Estimate Incremental Lifecycle Costs of Full-Function and City EVs
The incremental lifecycle costs of BPEVs are determined by
• incremental initial vehicle costs (including first battery and off-board charger)
• vehicle life
• replacement costs and salvage value of batteries
• EV and ICEV fuel costs
• EV and ICEV repair and maintenance costs
• charger installation costs
• the discount rate.
Table 4.13 lists the parameters that determine the incremental costs for full-function EVs
in each of these areas, along with plausible ranges for their values between 2003 and 2007 and
when these vehicles are in high-volume production. Parameters for full-function EVs using both
NiMH and PbA batteries are shown. Table 4.14 lists parameter values for city EVs with both
types of batteries. We now discuss the rationale for each parameter range.
Incremental Vehicle Cost. The incremental vehicle costs between 2003 and 2007 and in
volume production are taken from Table 4.12. For both the full-function and the city EVs, they
include the cost of the first battery pack.
 -62-

Vehicle Life. Using data from the Transportation Energy Data Book and other sources,
Delucchi et al. (2000, p. 55) estimate an average vehicle life of 150,000 miles. In its analysis of
Table 4.13
Parameters Used to Calculate Lifetime Operating Cost of a Full-Function EV
Relative to a Comparable ICEVa

Full-Function EV with Full-Function EV with

NiMH Batteries PbA Batteries
Volume Volume
Parameter 2003-2007 Production 2003-2007 Production
Initial Vehicle Cost ($) (including 16,700- 5,100- 8,000- 1,800-
first battery and off-board charger) 26,100 9,900 15,200 4,300
Vehicle Life
Lifetime vehicle miles traveled 150,000 150,000 150,000 150,000
Vehicle life in years 15 15 15 15
Parameters Affecting Battery
Replacement Cost
Pack size (kWh) 30 24 30 24
Battery cost ($/kWh) 305-360 233-258 115-154 114
Cycle life (1000 cycles) 1.0-1.5 1.5 0.6-1.0 1.0
Range (miles/cycle) 90-110 90-110 90-110 90-110
Cost surcharge for installing 500 500 500 500
subsequent packs ($/installation)
Salvage value ($/kWh) 0-40 0-40 -15 to 5 -15 to 5
Parameters Affecting Fuel Costs
EV efficiency (mi/kWh AC) 2-3 2.5-3.5 2-3 2.5-3.5
ICEV efficiency (mi/gal) 25-30 25-30 25-30 25-30
Electricity price ($/kWh) 0.035-0.05 0.07-0.09 0.035-0.05 0.07-0.09
Gasoline price ($/gal) 0.56-1.49 0.56-1.49 0.56-1.49 0.56-1.49
Other Costs
EV repair and maintenance (% 65-70 65-70 65-70 65-70
of ICEV costs)
ICEV maintenance ($/mile) 0.06 0.06 0.06 0.06
Charger installation ($) 500-1,500 500 500-1,500 500
Discount Rate (%) 4 4 4 4
aThe comparable ICEV is mid-sized, meets the SULEV and near-zero evaporative emission
standards, and averages 25 to 30 miles per gallon.

lifecycle costs, CARB assumes that vehicles travel 117,000 miles over their lifetimes (CARB,
2000b, p. 98), which is close to CARB’s definition of a “useful vehicle” life, rather than actual
miles driven before a vehicle is retired. Thus, for a full-function EV, we set lifetime miles
traveled at 150,000 miles. CARB also assumes a vehicle life of 10 years; we set vehicle life to 15
years to correspond to the higher lifetime miles traveled that we assumed in our analysis.
Because of the limited top speed and range of city EVs, they are unlikely to be driven as
far on average as full-function EVs over their lifetimes. Based on the National Personal
Transportation Survey, Reuyl and Schuurmans (1996) estimate that vehicles with a daily range of
50 to 60 miles can be used for 50 percent of the miles driven by the average driver. We thus
 -63-

assume that city EVs will be driven 75,000 miles over their lifetimes; CARB assumes 88,000
(2000b, p. 113).
Table 4.14
Parameters Used to Calculate Lifetime Operating Cost of a City EV
Relative to a Comparable ICEVa

City EV with NiMH City EV with PbA

Batteries Batteries
Volume Volume
Parameter 2003-2007 Production 2003-2007 Production
Initial Vehicle Cost ($) (including 8,500-13,400 1,700–3,400 5,300- 700-1,500
first battery and off-board charger) 9,900
Vehicle Life
Lifetime vehicle miles traveled 75,000 75,000 75,000 75,000
Vehicle life in years 15 15 15 15
Parameters Affecting Battery Cost
Pack size (kWh) 10 8 10 8
Battery cost ($/kWh) 305-360 233-258 115-154 114
Cycle life (1,000 cycles) 1.0-1.5 1.5 0.6-1.0 1.0
Range (miles/cycle) 50-60 50-60 50-60 50-60
Cost surcharge for installing 250 250 250 250
subsequent packs ($/installation)
Salvage value ($/kWh) 0 to 40 0 to 40 -15 to 5 -15 to 5
Parameters Affecting Fuel Costs
EV efficiency (mi/kWh AC) 4-5 5-6 4-5 5-6
ICEV efficiency (mi/gal) 40 40 40 40
Electricity price ($/kWh) 0.035-0.05 0.07-0.09 0.035-0.05 0.07-0.09
Gasoline price ($/gal) 0.56-1.49 0.56-1.49 0.56-1.49 0.56-1.49
Other Costs
EV repair and maintenance (% 65-70 65-70 65-70 65-70
of ICEV costs)
ICEV maintenance ($/mile) 0.045 0.045 0.045 0.045
Charger installation ($) 0 0 0 0
Discount Rate (%) 4 4 4 4
aThe comparable ICEV is small, meets the SULEV and near-zero evaporative emission standards, and
averages 40 miles per gallon.

Parameters That Affect Battery Replacement Cost. The cost of the batteries (excluding
the first battery pack) needed over a vehicle’s lifetime is determined by
• the battery pack size
• the cost of the batteries in dollars per kilowatt-hour
• the cycle life of the battery
• the miles driven on a battery cycle
• the cost markup for buying and installing subsequent battery packs
• the salvage value of batteries.
We discuss each in turn.
 -64-

As described in Subsection 4.1, we set the 2003-through-2007 battery pack size to 30 kWh
for a full-function EV and to 10 kWh for a city EV. It seems plausible that the efficiency of EVs
(in terms of kilowatt-hours per mile) will increase in the future (see discussion below). We
increase efficiency by 20 percent for our estimates of lifecycle costs in volume production so that
the battery pack can be smaller and maintain the same range. Accordingly, we reduce the full-
function EV and city EV battery packs to 24 and 8 kWh, respectively, for the high-production
volume cost calculations.
The above analysis of battery costs led us to conclude that NiMH battery packs (including
battery auxiliaries and indirect costs) would cost from $425 to $460 per kilowatt-hour between
2003 and 2007. The range for PbA batteries is $130 to $170. The cost of replacement batteries
for vehicles sold between 2003 and 2007 may well be lower than it was when the vehicle was
built due to battery production volumes increasing over time. We assume that by the time the
battery is replaced, 50 percent of the difference between the 2003-2007 costs and the midpoint of
the high-volume cost range will have been eliminated. The result: NiMH replacement costs will
be $305 to $360, and PbA replacement costs, $115 to $154. In the volume production scenarios,
we allow NiMH battery pack costs to range from $233 to $258 per kilowatt-hour and set PbA
battery costs at $114 per kilowatt-hour.
The EV’s range (90 to 110 miles for a full-function EV and 50 to 60 miles for a city EV)
multiplied by the battery cycle life is a starting point for the miles driven over one pack’s
lifetime, which in turn determines the number of packs needed over a vehicle’s lifetime. We
reduce the resulting estimate of the miles driven using a single pack by 20 percent, because
vehicle range is determined by taking the battery down to empty, whereas cycle life is usually
determined by taking the battery down to a 20 percent charge.
We put NiMH cycle life at 1,000 to 1,500 cycles between 2003 and 2007 and at 1,500
cycles in the high-volume cost scenario. CARB’s Battery Technology Advisory Panel (BTAP)
report (Anderman, Kalhammer, and MacArthur, 2000, p. 37) cites one auto manufacturer’s
projection of 1,200 cycles and 100,000 miles at 77°F and 1,100 cycles at 95°F from bench tests
that have been partially corroborated by field tests of battery packs in vehicles. While cycle life
deteriorates when the pack is continuously exposed to extreme heat (113°F), the important
implication is that a NiMH battery pack that is properly thermally managed can now achieve
100,000 miles. In reference to a Panasonic NiMH battery,16 the BTAP report optimistically
____________
16At the November 1999 UC Davis Conference on EVs, Panasonic EV reported that the RAV4-EV
(NiMH) had already achieved 100,000 kilometers of driving on one pack in tests in Japan. Furthermore,
Panasonic EV suggested that longer life can be expected with “economy charging,” i.e., intermittent full
charging in order to reduce deterioration of the NiMH battery electrode.
 -65-

forecasts that “it seems quite possible that 1,000 to 2,000 cycles at 100 percent depth of discharge
can eventually be achieved, depending on the battery’s initial power versus the car’s
requirements” (Anderman, Kalhammer, and MacArthur, 2000, p. 31).
PbA batteries currently achieve roughly 600 cycles, but Panasonic has been developing
designs with longer cycle lives, and 1,000 cycles appears to be possible (Matsushita Electric,
2000).17 We set PbA cycle life to 600 to 1,000 cycles between 2003 and 2007, and to 1,000
cycles in the high-volume scenarios.
Following CARB (2000b, p. 100), we set the cost of handling and installing replacement
battery packs at $500 per pack.
Batteries may have salvage value at the end of their lives. Battery experts at Southern
California Edison told Delucchi et al. (2000, p. 45) that used NiMH batteries would be at least as
good for load-leveling, backup, and other utility applications as the new PbA packs they currently
buy for such uses. Thus, the salvage value for NiMH batteries might be as high as $100 per
kilowatt-hour (roughly the current cost of new PbA batteries), at least until the supply of retired
NiMH batteries saturates the market. The secondary market for NiMH batteries is quite
uncertain, however. CARB (2000b, p. 100) assumes that the salvage value is $40 per kilowatt-
hour for NiMH batteries, but Delucchi et al. (2000) believe the batteries will lose their remaining
value very quickly and that there thus will be little market for them. They believe the batteries
will have to be recycled at a small net cost at low volumes and will have a small net salvage value
at medium volumes. In our analysis, we vary the salvage value of NiMH batteries from $0 to $40
per kilowatt-hour.
CARB (2000b, p. 100) believes a secondary market for used PbA batteries at meaningful
volumes is speculative and assumes a salvage value of $3 per kilowatt-hour—the value of the
materials in the battery. Delucchi et al. (2000, p. 46) assumes that PbA batteries will have
negative salvage value—that is, they will cost between $5 and $15 to recycle. In our analysis, we
vary PbA salvage values from -$15 to $5 per kilowatt-hour.
The last battery pack installed in the vehicle might not be fully used at the end of the
vehicle life. Following Delucchi et al. (2000, p. 69), we set the salvage value of the unused
portion of the pack equal to 70 percent of the cost of batteries to the manufacturer.
____________
17“The newly developed EV battery lasts approximately twice as long and offers about twice as much
power as existing lead-acid EV batteries. It can be discharged and recharged approximately 1,000 times,
which is about twice as many cycles as the company’s existing lead-acid EV battery products” (Matsushita
Electric, 2000).
 -66-

Parameters That Affect Fuel Costs. The cost of electricity, the cost of gasoline, the
wall-to-wheels (or AC) efficiency of the EV, and the fuel efficiency of the ICEV determine the
lifetime fuel costs of a full-function EV relative to an ICEV.
The number of BPEVs is relatively small between 2003 and 2007, and we assume that
BPEVs can be charged off-peak using existing power generation, transmission, and distribution
facilities. As shown in Appendix D, the costs of power in such a situation might be expected to
run from $0.035 to $0.05 per kilowatt-hour. If EVs become a significant part of the
transportation system, they will use a significant amount of power, requiring additional
generation, transmission, and distribution facilities.18 For volume production, we thus put
electricity costs at $0.07 to $0.09 per kilowatt-hour, which covers the capital cost of the power
plant and the transmission and distribution system (see Appendix D). Because we are interested
in resource costs, we do not include taxes or charges imposed to recover costs during California’s
recent energy crisis. Likewise, CARB does not include taxes in its analysis of operating costs.
CARB assumes an electricity cost of $0.05 per kilowatt-hour; however, it also presents a scenario
in which costs are $0.075 per kilowatt-hour (CARB, 2000b, p. 107).
There is a good deal of uncertainty regarding gasoline prices over the lives of the vehicles.
We assume prices at the pump will vary between $1.00 and $2.00 per gallon, which is consistent
with prices observed in recent years in California. Federal and state fuel excise taxes currently
total $0.363 per gallon, and sales taxes on gasoline average roughly 8 percent in the state (CARB,
2000b, p. 107). We thus assume that gasoline costs vary from $0.56 to $1.49 per gallon, net of
taxes.
EV efficiency (in miles per kilowatt-hour AC) is based on data developed by Southern
California Edison for the EVs produced by the large manufacturers.19 Our figures for ICEV fuel
efficiency reflect a rough average of the fuel efficiencies of ICEVs that are comparable in size to
full-function EVs and city EVs (27.5 miles per gallon for mid-size ICEVs and 40 miles per gallon
for very small ICEVs).
The efficiency of EVs may improve over future design cycles and thus might be higher in
volume production than it will be between 2003 and 2007. CARB assumes that efficiency
increases from about 2.2 miles per kilowatt-hour in 2003 to roughly 4 miles per kilowatt-hour in
volume production for a four-passenger full-function EV (CARB, 2000b, p. 111). This
____________
18Forexample, to charge 100,000 full-function EVs overnight will require a 250-megawatt power
plant—a good-sized plant. (If each full-function EV requires 30 kWh per night, 100,000 full-function EVs
require 3,000 megawatt-hours. And 3,000 MWh/12 hours = 250 megawatts.)
19We determined the range using the efficiencies remaining after the vehicles with the highest
efficiency (the EV1) and lowest efficiency (the EPIC) were discarded (see http://ev.inel.gov/fop).
 -67-

improvement is driven by a substantial reduction in aerodynamic drag, lower-loss tires, a higher-

efficiency drive system, and substantial improvements in charging efficiency. While efficiency
may well improve, we think it likely that CARB is overly optimistic. For example, vehicles with
very low aerodynamic drag are technologically feasible but are not assured of success in the
marketplace. We increase EV efficiency roughly 20 percent from the levels used to estimate
costs between 2003 and 2007, yielding a battery pack that is 20 percent smaller but allows the
same range.
EV Repair and Maintenance Costs. It is often argued that EVs will have lower repair
and maintenance costs than ICEVs do, because EVs have fewer components and fewer moving
parts. While this may be so in the long term, we follow Dixon and Garber (1996, p. 244) in
assuming there are several reasons to be skeptical in the near term:
• EV technology is evolving rapidly, but projections of EV repair and maintenance
costs may be based implicitly on costs for stable designs. Components will
probably be less reliable in the early years of commercial scale production than in
later years.
• Several manufacturers have had problems with their battery packs and have had to
replace battery modules, an expensive process. Installing replacement packs may
also be time consuming and costly.
• Repair and maintenance costs of ICEVs appear to be declining.
Delucchi et al. (2000, p. 245) put EV repair and maintenance costs at approximately 70
percent of ICEV costs; CARB (2000b, p. 108) uses 66 percent in its analysis. We let the
proportion vary from 65 to 70 percent. The ICEV repair and maintenance cost we use for
reference for the full-function EV is $0.06 per mile (as in Delucchi et al., 2000, p. 245, and
CARB, 2000b, p. 108). We reduce the repair and maintenance cost by 25 percent for the smaller
ICEV used in the city EV incremental-cost calculation.
Charger Installation. We assume that the off-board charging equipment is included in
the cost of the EV, but that the off-board charger must be installed at the business or residence
where the vehicle is charged. Installation often requires significant extension or modification of
the business/residence electrical system and may require structural modifications. There appears
to be a great deal of variation across customers in charger installation costs. Nissan reports an
average cost (based on a fairly small number of installations) of $3,000 for installation of a full-
function EV charger inside the garage of a single-family home (Atkins et al., 1999); CARB sets
the cost of the charger plus installation at $1,500 in 2003 and $750 in volume production. Our
analysis assumes installation costs range from $500 to $1,500 between 2003 and 2007. Average
installation costs in volume production should be lower—in part because some households and
 -68-

businesses will already have installed the 220-volt line used by most full-function EV chargers.
We set installation costs at $500 for volume production.
It appears that city EVs will likely be offered with a number of different charging options,
ranging from simply plugging the vehicle into a standard 110-volt outlet to using a rapid charging
system that requires the same 220-volt, 30-ampere outlets used by full-function EVs. The 110-
volt option requires no charger installation costs, but it may take up to 8 hours to charge the 10-
kWh battery. The rapid-charge option will charge the battery very quickly but may require
installation costs as high as those for full-function EVs. We set charger installation costs to zero
in our analysis. The key requirement for any EV is that it be able to charge overnight, and the
city EV will be able to do so using a standard 110-volt circuit.
Discount Rate. The discount rate translates costs incurred in the future to present-day
costs. Because we are interested in the resource costs of advanced technology vehicles to society
as a whole, we set the discount rate to the social (as opposed to the private) discount rate.20 The
social discount rate is defined as the social opportunity cost of capital and is often based on the
long-run, risk-free (or low-risk) return on capital (Tietenberg, 1992, p. 88). Examples include the
return on U.S. Treasury bonds. Estimates of the long-run, risk-free return are typically 3 to 5
percent, net of inflation.

Parameters Used to Estimate Incremental Lifecycle Costs of DHFCVs

Table 4.15 lists the parameters used to estimate the incremental lifecycle costs of a
DHFCV relative to an ICEV that meets the SULEV exhaust and near-zero evaporative emission
exhaust standards. The key parameters in the incremental operating costs are the DHFCV’s fuel
economy in miles per kilogram of hydrogen and the cost of hydrogen in dollars per kilogram.
The Methanol Institute summarizes the findings of eight studies on the expected fuel economy of
DHFCVs.21 The efficiencies range from 55 to 106 miles per equivalent gallon of gasoline, and
the Methanol Institute uses 65 to 85 miles in its analysis. The two measures—miles per
equivalent gallon of gasoline and miles per kilogram of hydrogen—are very close, and we use 65
to 85 miles per kilogram for our cost projections for 2006 to 2010. For high-volume lifecycle
cost projections, we set the efficiency at 85 miles per kilogram.
Appendix D derives an estimate of the cost of hydrogen fuel. We assume that
hydrogen is sold at filling stations that generate hydrogen from natural gas piped to the site.
Assumptions on the cost of natural gas and the cost of the reformer that generates hydrogen
____________
20Consumers are thought to apply much higher discount rates when buying durable goods. See, for
example, Kavalec, 1996, p. 30.
21See http://www.methanol.org/fuelcell/special/contadini_pg1.html, Tables 1 and 2.
 -69-

from natural gas result in hydrogen costs of $2.32 to $3.16 per kilogram between 2006 and
2010. The costs fall to $1.85 to $2.33 per kilogram for high-volume production of
reformers. To put these numbers in perspective, $2.00 per kilogram corresponds to $0.027
per mile in a vehicle that gets 75 miles per kilogram. Fuel cost for a mid-size ICEV that
Table 4.15
Parameters Used to Calculate Lifetime Operating Costs of a DHFCV
Relative to a Comparable ICEVa

Parameter 2006-2010 Volume Production

Initial vehicle cost ($) 8,300-15,200 1,200 - 3,900
Vehicle efficiency
DHFCV (mi/kg of hydrogen) 65-85 85
ICEV (mi/gal) 25-30 25-30
Fuel cost
Gasoline ($/gal) 0.56-1.49 0.56-1.49
Hydrogen ($/kg) 2.32-3.16 1.85-2.33
Maintenance costs ($/mi)
DHFCV 0.06 0.06
ICEV 0.06 0.06
aThe comparable ICEV used in the calculations is mid-size, meets the SULEV and near-
zero evaporative emission standards, and averages 25 to 30 miles per gallon.

averages 27.5 miles per gallon when gasoline (net of taxes) is $1.00 per gallon is $0.036
per mile.
Little information is available on the maintenance costs of DHFCVs relative to
ICEVs. A DHFCV’s electric power train has fewer parts than an ICEV’s drivetrain and
may be cheaper to maintain. The reliability and maintenance costs of a PEM fuel cell,
however, are uncertain. Lacking better information, we set the maintenance costs of
DHFCVs equal to those for ICEVs.

Parameters Used to Estimate Incremental Lifecycle Costs of GHEVs

GHEVs may be cheaper to operate than standard ICEVs because of their higher gas
mileage. Following CARB (2000b, p. 114), we set the efficiency of a four-passenger GHEV to
45 miles per gallon between 2003 and 2007 and assume that improvements in vehicle design will
increase the fuel economy to 55 miles per gallon in volume production (see Table 4.16).22 These
fuel economies are 50 and 57 percent greater, respectively, than the 30 and 35 miles per gallon
that CARB (2000b, p. 115) assumes for a comparable ICEV.
In its initial analysis of hybrid costs, CARB concluded that hybrid maintenance costs may
be higher than those of standard gasoline vehicles because hybrids contain both a conventional
____________
22CARB’s estimate of 45 miles per gallon is based on published fuel mileage figures for currently
available hybrids.
 -70-

and an electric drive system. CARB noted that little was known about hybrid operating costs,
however, and “in the absence of more specific information,” set maintenance costs at $0.075 per
mile (CARB, 2000b, p. 108). CARB later reexamined the maintenance costs of hybrids and
 -71-

Table 4.16
Parameters Used to Calculate Incremental Lifecycle Cost of a GHEV
Relative to a Comparable ICEVa

High-Volume
Parameter 2003-2007 Production
Initial vehicle cost ($) (including first battery and 2,300-3,300 500-900
off-board charger)
Vehicle efficiency (mi/gal)
GHEV(ATPZEV) 45 55
ICEV 30 35
Fuel costs ($/gal) 0.56-1.49 0.56-1.49
Maintenance costs ($/mi)
GHEV(ATPZEV) 0.06-.078 0.06-.065
ICEV 0.06 0.06
aThe comparable ICEV used in the calculations is mid-size, meets the SULEV and near-zero
evaporative standards, and averages 30 miles per gallon between 2003 and 2007 and
35 miles per gallon when GHEVs are in volume production.

concluded that incremental maintenance costs would be zero. CARB noted that there may be
some incremental maintenance costs associated with the battery or electric drive components
(e.g., the 2002 Toyota Prius requires a coolant change for the electric power inverter at 34,000
miles) but that there may be less brake wear due to regenerative braking (CARB, 2001h, p. 167).
CARB never explicitly addresses the costs associated with maintaining and possibly
replacing the battery in hybrids. There is uncertainty over whether one battery will last the entire
life of the vehicle. As noted in Subsection 4.3, a recent report put the life expectancy of a NiMH
hybrid battery at five to eight years. Ford reports that the battery in its Escape hybrid will have to
be replaced at least once during a 15-year vehicle life. Honda implies that 15 years in service is
beyond today’s battery technology. Toyota believes that scheduled replacement of the battery
will not be required over a 15-year lifetime but that unscheduled maintenance is possible (CARB,
2001h, p. 129).
This uncertainty drives us to develop a range for hybrid maintenance costs. The lower end
of the range is $0.06 per mile, equal to the maintenance costs of standard ICEVs. The upper end
allows for the accumulation of enough money over the life of the vehicle to replace the battery
once. For 2003 to 2007, we take the upper estimate for the cost of the battery ($2,000—see
Subsection 4.3), add 30 percent for indirect costs (see Subsection 4.6), and add $125 for
installation (lower than the $500 assumed cost to replace the much larger full-function EV pack).
The sum translates into $0.018 per mile, resulting in a maintenance cost range that runs from
$0.06 to $0.07 per mile. In high-volume production, the pack costs $500, resulting in a range of
$0.06 to $0.065 per mile.
 -72-

Method for Calculating Incremental Lifecycle Costs

We assume that each of the parameters is statistically independent and uniformly
distributed on its specified range. We then randomly draw a value from each parameter range
and use the resulting set of values to calculate the incremental lifetime operating cost and
incremental lifecycle cost of each advanced technology vehicle relative to an ICEV. We repeat
this process 10,000 times and report the 5th and 95th percentiles of the distribution. Given
assumptions on the distributions of the parameters, if we repeated the process for constructing the
interval over and over, this interval would contain the true incremental cost 90 percent of the
time. We chose to repeat the process 10,000 times because the distribution is stable to two
significant digits with this number.

Findings on Incremental Lifecycle Costs

Results of our analysis of incremental lifetime operating and lifecycle costs are reported in
Table 4.17. The second and third columns show our findings on the incremental lifetime
operating cost in the near term and in volume production, respectively. The fourth and fifth
columns add in the incremental initial cost of the vehicle and show the incremental lifecycle cost.
The brackets in each cell contain the 90 percent probability interval for the estimate.
Full-Function EVs. During the first five years of the program, the lifetime operating cost
of full-function EVs with NiMH batteries may be higher or lower than that of an ICEV meeting
the SULEV exhaust and near-zero evaporative emission standards: As can be seen in the second
column of Table 4.17, the 90 percent probability interval runs from -$1,600 to $4,000. Once the
high incremental production cost for these vehicles is included, the incremental lifecycle cost
ranges from $17,300 to $27,800. Our analysis suggests that even in volume production, the
lifecycle cost of full-function EVs with NiMH batteries will remain substantially above that for
ICEVs: The incremental cost estimates range from $3,100 to $8,700.
PbA batteries are cheaper than NiMH batteries, but their short cycle life means that the
lifetime operating cost will likely be somewhat greater for full-function EVs with PbA batteries
than for their NiMH-powered counterparts. Incremental lifecycle cost is a good deal lower,
however, because the initial vehicle is cheaper. Between 2003 and 2007, there will still be a
substantial lifecycle cost premium for full-function EVs powered with PbA batteries ($9,000 to
$18,400). Incremental lifecycle cost may be small in volume production ($200), but there is an
equally good chance that incremental lifecycle cost will remain substantial (the upper end of the
90 percent probability interval is $5,100). Thus, even with PbA batteries, the lifecycle cost of
full-function EVs is likely to exceed that of ICEVs. What is more, questions remain about
whether PbA batteries are feasible for full-function EVs that have a range of roughly 100 miles,
 -73-

Table 4.17
Incremental Lifetime Operating and Lifecycle Costs of Advanced Technology Vehicles
(incremental over a vehicle meeting SULEV and near-zero evaporative emission standards)
($1000s per vehicle; 90 percent probability intervals in brackets)a

Incremental Lifetime Operating Cost Incremental Lifecycle Cost

Vehicle High-Volume High-Volume
Technology Near Term Production Near Term Production
2003-2007 2003-2007
Full-function EV
NiMH [-1.6, 4.0] [-3.1, 1.0] [17.3, 27.8] [3.1, 8.7]
PbA [-1.1, 5.2] [-2.5, 5.1] [9.0, 18.4] [0.2, 5.1]
City EV
NiMH [-1.3, 0.5] [-1.6, -0.3] [8.1, 12.9] [0.5, 2.4]
PbA [-1.1, 1.0] [-1.3, -0.1] [5.1, 9.9] [-0.4, 1.2]
GHEV [-1.4, 0.8] [-1.5, 0.3] [1.3, 3.7] [-0.9, 0.4]
PZEV 0 0 [0.06, 0.23] [0.06, 0.23]
2006-2010 2006-2010
DHFCV [-2.1, 2.0] [-3.4, 0.3] [7.8, 15.6] [-1.3, 3.4]
aDiscounted to the time of vehicle sale using a 4 percent discount rate.

since the size and weight of these batteries make it difficult to design such a vehicle with
acceptable space and handling.
City EVs. Mainly because of their smaller battery packs, city EVs have a much lower
incremental lifecycle cost than full-function EVs do. Between 2003 and 2007, the operating cost
of a city EV with either PbA or NiMH batteries may be above or below that of a small ICEV.
Substantial incremental vehicle production cost between 2003 and 2007 still means, however, that
incremental lifecycle cost for a city EV remains sizable with either type of battery.
For high-volume production, our analysis predicts that if battery cycle life reaches the
most optimistic expectations and vehicle efficiencies improve somewhat, the lifecycle cost of a
city EV with PbA batteries could be comparable to that of a small ICEV. As shown in Table
4.17, the 90 percent probability interval ranges from -$400 to $1,200. Again, however, note that
manufacturers are choosing to use NiMH batteries and NiCd batteries in their city EVs despite
the higher costs, which suggests that PbA batteries do not work well in a city EV design.
GHEVs. The cost of servicing and possibly replacing the battery in a GHEV more than
offsets the savings due to this vehicle’s higher gas mileage. Our estimates of incremental lifetime
operating cost range from -$1,400 to $800 per vehicle for vehicles produced between 2003 and
2007 and from -$1,500 to $300 in high-volume production. Once the incremental cost of the
vehicle is added in, the hybrid’s lifecycle cost is $1,300 to $3,700 greater than that of a standard
ICEV between 2003 and 2007. Incremental lifecycle cost runs from -$900 to $400 in volume
production.
 -74-

PZEVs. The greater incremental lifecycle cost of a PZEV relative to an ICEV that meets
the SULEV exhaust and near-zero evaporative emission standards is due only to the increased
initial cost of a PZEV. Our estimates range from $60 to $235 from 2003 on.
DHFCVs. Given our assumptions on the cost of hydrogen fuel, fuel efficiency, and
vehicle maintenance, the lifetime operating cost of a DHFCV may be several thousand dollars
less than that of an ICEV. The midpoint of the estimate for incremental operating cost between
2006 and 2010 is near zero, although there is enough uncertainty in the parameters that the upper
end of the 90 percent probability interval reaches $2,000. Very high initial vehicle cost means
that the incremental lifecycle cost of DHFCVs remains high through 2010. Plausible
improvements in fuel-cell technology and reductions in the cost of providing hydrogen fuel may
ultimately reduce the lifecycle cost for this vehicle to below that for an ICEV (the lower end of
the probability interval is -$1,300). However, this is by no means guaranteed: The upper end of
the interval is $3,400.