Journal of Power Sources 168 (2007) 459–468

Using fleets of electric-drive vehicles for grid support

Jasna Tomić ∗ , Willett Kempton
University of Delaware, Newark, DE 19716, USA
Received 15 January 2007; received in revised form 6 March 2007; accepted 6 March 2007
Available online 12 March 2007

Abstract
Electric-drive vehicles can provide power to the electric grid when they are parked (vehicle-to-grid power). We evaluated the economic potential
of two utility-owned fleets of battery-electric vehicles to provide power for a specific electricity market, regulation, in four US regional regulation
services markets. The two battery-electric fleet cases are: (a) 100 Th!nk City vehicle and (b) 252 Toyota RAV4. Important variables are: (a) the
market value of regulation services, (b) the power capacity (kW) of the electrical connections and wiring, and (c) the energy capacity (kWh) of
the vehicle’s battery. With a few exceptions when the annual market value of regulation was low, we find that vehicle-to-grid power for regulation
services is profitable across all four markets analyzed. Assuming now more than current Level 2 charging infrastructure (6.6 kW) the annual net
profit for the Th!nk City fleet is from US$ 7000 to 70,000 providing regulation down only. For the RAV4 fleet the annual net profit ranges from
US$ 24,000 to 260,000 providing regulation down and up. Vehicle-to-grid power could provide a significant revenue stream that would improve
the economics of grid-connected electric-drive vehicles and further encourage their adoption. It would also improve the stability of the electrical
grid.
© 2007 Published by Elsevier B.V.

Keywords: Electric-drive vehicles; Energy storage; Vehicle-to-grid power; Ancillary services; V2G

1. Introduction compared to individual vehicles, fleets are more easily accom-

modated within existing electric market rules, which typically
Several recent studies [1–6] show that electric-drive vehicles require power blocks of 1 MW. We selected battery-electric vehi-
(EDVs) may profitably provide power to the grid when they are cles over plug-in hybrids and fuel cell vehicles because battery
parked and connected to an electrical outlet. We call this vehicle- vehicles already must be grid connected (in order to recharge
to-grid power (V2G). In other studies [7,8], we analyzed V2G the batteries) and because such fleets already exist. Among the
from three types of electric-drive vehicles – battery, fuel cell, ancillary services, we analyze regulation because: (a) it has the
and plug-in hybrid – and analyzed the economic potential for highest market value for V2G among the different forms of elec-
individual vehicles to provide power for baseload, peak power, tric power (much higher than peak power, for example), (b) it
and for the electric grid services known as ancillary services minimally stresses the vehicle power storage system, and (c)
(A/S), as well as storage for renewable energy sources [9]. The because battery-electric vehicles are especially well suited to
focus of the current paper is a more near-term opportunity, using provide regulation services.
fleets of battery-electric vehicles to provide ancillary services. We begin the paper with a section that explains why battery-
We focus here specifically on one type of ancillary service- electric vehicles are a good source of power for ancillary
regulation. We decided to analyze utility fleets because they have services. First, we describe ancillary services in general and, in
in-company expertise in, and need for, ancillary services. Also, more detail regulation. Then we describe the main principles and
components of vehicle-to-grid power as well as the advantages
of using EDVs for regulation services.
∗ Corresponding author. Present address: WestStart-CALSTART, 48 S Chester
Section 2 provides the general equations used in the calcu-
Ave, Pasadena, CA 91106, United States. Tel.: +1 626 744 5695; lations of the value and cost of V2G for regulation. Cost and
fax: +1 626 744 5610.
revenue calculations are introduced in this section as well. Then
E-mail addresses: [email protected] (J. Tomić), [email protected]
(W. Kempton). the electrical power capacity for V2G and the costs of providing
URL: http://www.udel.edu/V2G (W. Kempton). V2G power are quantitatively defined.

0378-7753/$ – see front matter © 2007 Published by Elsevier B.V.

doi:10.1016/j.jpowsour.2007.03.010
460 J. Tomić, W. Kempton / Journal of Power Sources 168 (2007) 459–468

In Section 3, we present two case studies of battery EDV fleets an amount commensurate with their loads. The important char-
providing V2G power for regulation. The results of the economic acteristic of the ancillary services market price is that it has
value created by each fleet are presented based on the general two parts—a capacity price and an energy price. The capacity
equations described. We then expand these two cases to calculate price is the price paid to have a unit available for a specified
the economic value that these two fleets costs are presented in service while the energy price is the price paid for the energy
US$ would offer in different A/S markets across the country. All output when a unit is called in real time to supply incremental
cost and revenue calculations in this article are in US$. or decremental energy.

2. Vehicle-to-grid power for ancillary services 2.2. Vehicle-to-grid power: an improved power source for
ancillary services
2.1. Ancillary services
The basic concept of vehicle-to-grid power is that EDVs pro-
In the electric power system, ancillary services are neces- vide power to the grid while they are parked. The EDV can be a
sary for maintaining grid reliability, balancing the supply and battery-electric vehicle, hybrid, or a fuel cell vehicle connected
demand, and supporting the transmission of electric power from to the grid. Details on the economic analysis for all three types
seller to purchaser. They are not widely known because prior of EDVs can be found elsewhere [7,8]. Battery EDVs provide
to restructuring of the power sector, they were bundled with the their stored electricity for V2G power.
energy supply and the cost of ancillary services was hidden in Each vehicle must have three required elements for V2G: (a)
the overall energy rates and operating expenses. With deregula- a power connection to the grid for electrical energy flow, (b)
tion however, some jurisdictions have created separate markets control or logical connection necessary for communication with
for ancillary services, making their costs more apparent. grid operators, and (c) precision metering on-board the vehicle.
We are concerned here with regulation ancillary service. The Fig. 1 is an illustration of connections between vehicles and the
main purpose of regulation is to adjust the grid, specifically the electric power grid. The control signal from the grid operator is
local control area, to the target frequency and voltage. Regu- shown schematically as a radio signal, but this might be through
lation helps maintain interconnection frequency, balance actual the medium of a cell phone network, direct internet connection,
and scheduled power flow among control areas, and match gen- or other media. In any case, the system operator (ISO or util-
eration to load within the control area [10]. The required amount ity) sends requests for ancillary services to a large number of
of regulation service is determined as a percentage of aggre- vehicles. The signal may go directly to each individual vehicle,
gate scheduled demand. In California for example, regulation schematically in the upper right of Fig. 1, or via a fleet’s home
requirements range between 5 and 10% of the scheduled load office to vehicles centralized in a fleet parking lot, schematically
[11]. shown in the lower right of Fig. 1.
Generators providing regulation are operated differently from To understand why V2G makes sense, one must under-
generators providing just bulk power. For regulation, generators stand the scheduling and economic value of V2G. In order to
ramp up and down to match the needs of fluctuation in the grid. schedule dispatch of power, a grid operator needs to rely that
Regulation is provided continuously (24 h a day) by genera- enough vehicles are parked and potentially plugged in at any
tors that are online, equipped with automatic generation control minute during the day. In the US, an average personal vehi-
(AGC) and will respond quickly (within minutes) to control cen- cle is on the road only 4–5% of the day, which means that a
ter requests to increase or decrease power output. In states with great majority of the day the vehicles are parked. Our prior
Independent System Operators (ISOs), the ISO may purchase analysis estimates that at least 90% of personal vehicles are
ancillary services and/or require individual utilities to provide parked even during peak traffic hours [7]. For fleet vehicles pre-

Fig. 1. Illustrative schematic of power lines and wireless control connections between vehicles and the electric power grid.
 J. Tomić, W. Kempton / Journal of Power Sources 168 (2007) 459–468 461

dictability of using V2G is excellent because they follow a daily a distribution company, and cost and revenue are calculated on
schedule. an annual basis. Payments for regulation are based on two com-
The economics question is second. While electricity from ponents: (a) a contract payment for availability (in US$/MWh)
V2G is not cheap when compared to bulk electricity from large plus (b) an energy payment per kWh when power is produced.
power plants (e.g. US$ 0.30/kWh versus $ 0.05/kWh), it can Yearly revenue from regulation up is calculated using Eq. (1)
be competitively used for ancillary services because of the two
parts that make up the price of power in the ancillary service rReg−up = (pcap P tplug ) + (pel P tplug Rd-c ) (1)
market—capacity price and energy price. When a generator, in
this case a battery-vehicle, provides ancillary services it is paid where pcap is the capacity price in US$/kW-h, tplug the time
a capacity price for being available to respond on a minute’s in hours the EDV is plugged in, pel the market (selling) price
notice, and an energy price for the actual energy output. The of electricity (US$/kWh) and P is the power of the vehicle or
energy output may be quite small, making the cost to produce power of the line in kW (described later in a separate section).
each kWh of little consequence for the overall economics. More The capacity prices for regulation up and regulation down (pcap )
important factors than cost per kWh are: (a) the capital cost are obtained from system operator data for each region being
of generation or storage equipment, (b) ability to vary output analyzed. We use the market clearing prices in the day-ahead
quickly, and (c) ability to operate in these modes without serious markets to derive the average price in a year (in US$/MW-h).
maintenance penalties. Vehicles are better than central genera- Note that the unit US$/MW-h refers to a power capacity con-
tors on all three counts, as we detail elsewhere [8]. The capital tracted for 1 h and should not be confused with MWh, a unit
cost of vehicles can be attributed to their transportation function, of energy produced. This contract payment value is determined
since our proposed operating modes for ancillary services do not by the particular power market or ISO region and varies from
affect vehicle operation. region to region.
To add V2G capability to a battery EDV, two capabilities The term tplug is determined directly as the time that the vehi-
must be added. First, the on-board (vehicle) power electron- cle is plugged-in, or potentially plugged, and available for V2G.
ics designed for V2G and second, real-time control so that the The term (Rd-c ) is the dispatch to contract ratio, which in com-
ISO or grid system operator can request power exactly when bination with tplug defines the dispatch of V2G power. The Rd-c
needed. Electric system operator control is essential because is defined by Eq. (2)
V2G has value greater than its cost only if the buyer (the elec-
tric system operator) can determine the precise timing of the Edisp Edisp
Rd-c = = (2)
dispatch. Of course the dispatch would be within limits set by Pcontr tcontr P tplug
the driver or fleet operator and such that the driver or fleet oper-
ator would have always sufficient power left in the battery for This ratio is defined by the energy dispatched for reg-
driving. ulation as a proportion of contracted power and contracted
Unlike large generators, battery EDV’s energy storage and time (kWh/kW*h). In the case of regulation, tcontr = tplug and
power electronics are already designed to provide large and fre- Pcontr = P, where P is power of vehicle or power of line (dis-
quent power fluctuations over short time periods, due to the cussed later). We requested the data for the ratio from multiple
nature of driving. This makes these vehicles especially well engi- utilities and grid operators but found that this ratio is not tracked
neered for regulation. Once a signal is received, the vehicle can or recorded [7–9]. Therefore, the Rd-c ratio was calculated based
respond in less than a second to change its power output. A “reg- on a signal available for frequency regulation from California
ulation up” signal would cause the vehicle to provide power to ISO (CAISO) during a course of a day [13] and modeling the
the grid (V2G) and a “regulation down” signal would cause a response of one EDV. The result is a value of 0.10 for Rd-c which
decrease in the power output or even draw power from the grid we use in the present analysis.
(the regular battery charging mode). Brooks [12] successfully Battery EDVs are best suited to provide both regulation up
demonstrated use of a single battery electric vehicle to respond and down, as the result is no net change in battery charge. How-
to a regulation signal. ever, a near-term approach would be to simplify controls and
approval by providing regulation down only, so power flows
3. Value of V2G power for regulation only from grid to vehicle. Yearly revenue from regulation down
only is defined by Eq. (3):
This section develops the equations used to calculate the value
of V2G for regulation. In separate subsections, we introduce the rReg-down = (pcap P tplug ) (3)
calculation for revenue, cost, and electrical power capacity for
V2G. These general equations are subsequently used to calculate If the EDV is providing only regulation down, (battery is only
the values of V2G for our case study fleets. charging) the battery may become fully charged and therefore
tplug will be shorter than when the EDV is performing both reg-
3.1. Revenue of V2G power for regulation ulation up and down. In our calculations we assumed that the
battery is at 50% state of charge at the start of regulation down
Calculations for revenue and cost for regulation services mode. More details on calculating tplug in this case are shown
make the following assumptions. Regulation is purchased by later in Section 3.3.
462 J. Tomić, W. Kempton / Journal of Power Sources 168 (2007) 459–468

3.2. Cost of providing V2G power for regulation (see Eq. (8), LET = LC Es DoD), then at 3% DoD the throughput
is 10 times greater than the throughput at 100% DoD. A sim-
The cost to produce regulation up is calculated as the cost to ilar relationship of throughput and DoD is also suggested by
produce each kWh times the number of kWh produced per year. Miller and Brost [15]. Their Fig. 8 suggests that at 3% DoD the
Cost for regulation down is considered zero because regulation throughput is about 28 times greater than at 80%. The relation-
down is the same as charging the battery, thus it is “free charging” ship of DoD and throughput depends on the electrochemistry of
at times when the vehicle is providing regulation down. Yearly the battery. In lack of data specific to the battery types consid-
cost for regulation up is: ered in this paper, and to be conservative, we choose to use here
a factor of 3 greater throughput at shallow cycling compared to
cReg-up = (cen P tplug Rd-c ) + cac (4) deep cycling.
Yearly cost from regulation down is: The other cost component of delivering V2G power is the
fixed cost, expressed as annualized capital cost cac for additional
cReg−down = 0 (for regulation up and regulation down) (5) equipment required for V2G. A simple way to annualize a single
or capital cost is to multiply the cost by the capital recovery factor
(CRF) as in Eq. (9)
cReg-down = cac (for only regulation down) (6) d
cac = cc × CRF = cc × (9)
where cReg-up and cReg-down stand for total cost of regulation up 1 − (1 + d)−n
or regulation down, cen the cost per energy unit in US$/kWh and where cc is the capital cost (the one-time investment) in US$, d
includes cost of electricity, losses, plus battery degradation costs, the discount rate, and n is the time during which the investment
and cac is the annualized capital cost for additional equipment is amortized in years.
needed for V2G. Fixed costs can be incurred on the vehicle’s power electron-
The economic viability of V2G depends critically on the cost ics and connectors, and off-board due to charging station or
to the vehicle owner to produce V2G power. Eq. (7) is used wiring upgrades. Battery EDVs already must have electrical
to calculate the per kWh cost to the battery EDV owner for connections for recharging their batteries. To add V2G capa-
providing power to the grid and Eq. (8) is used to calculate cost bilities requires little modification to the charging station and
of battery degradation no modification to the cables or connectors, but the on-board
cpe power electronics must be designed for this purpose. AC Propul-
cen = + cd (7)
ηconv sion, Inc. has designed and built a power electronics system that
allows charging from and discharging to the grid and includes
cbat (Es cb ) + (cl tl )
cd = = (8) extensive control and safety to ensure no back feeding of power
LET LC Es DoD onto the grid during an outage [16]. The incremental cost of the
where cpe is the cost of purchased electricity for recharging in power electronics system is reported to be US$ 400, assuming
US$/kWh (in most cases equal to $0.05/kWh), cd cost of battery moderate production runs [17,13]. Another fixed cost is that we
degradation in US$/kWh calculated as shown in Eq. (8), ηconv assume the necessity of on-board metering of electrical flow for
the conversion efficiency of fuel or electricity—in this case it is billing purposes. We assume use of a chip available from Analog
the two-way electrical efficiency (electricity to battery storage Devices, Inc with original equipment manufacturer (OEM) cost
and back to electricity), which for a more efficient than aver- of US$ 3.00 [18]. With additional parts and labor, we estimate
age battery EDV is 0.73, cbat the battery replacement cost in that the total incremental cost for an on-board electric metering
US$ (capital and labor costs), LET the battery lifetime energy system is US$ 50. A wireless communication system would be
throughput for a particular cycling regime in kWh, Es the total necessary to allow communication with the ISO. The cost of a
energy storage of the battery in kWh, cb cost of battery replace- wireless system installed in production scale is estimated around
ment in US$/kWh, cl the cost of labor in US$/h, tl labor time US$ 100 [17]. Thus, for battery EDVs, the total capital cost is
required for battery replacement, and LC is battery lifetime in equal to US$ 550. This capital cost annualized according to Eq.
cycles. We assume here that battery replacement is determined (10) using a discount rate of 10% over a period of 10 years,
by its cycle life, not calendar life. (For some batteries and driv- amounts to US$ 90 per year, per vehicle.
ing cycles, calendar life would be reached first, in which case cd When the vehicle is providing only regulation down, the capi-
would be a zero cost rather than the values we calculate here.) tal cost is lower. In this case, power flows only from the grid to the
Regulation requires a modification regarding the battery vehicle and the vehicle would require only the on-board meter-
degradation costs which will be lower as a result of the shallow ing device (US$ 50) and the wireless interconnection (US$ 100).
type of cycling for regulation rather than deep charge/discharge The incremental capital cost is only US$ 150 and the annualized
cycling that battery degradation tests usually assume. It has been cost (using Eq. (9)) is US$ 25 per vehicle.
shown that shallow cycling has much less impact on battery
energy throughput than more common deep cycling. For exam- 3.3. Electrical power capacity
ple, test data on a Saft lithium-ion battery show a 3000-cycle
lifetime at 100% discharge, and a 1,000,000-cycle lifetime at The electrical power capacity available for V2G is deter-
3% discharge [14]. If we use these data to calculate throughput mined by two factors: (a) the limitation of the electrical circuit
 J. Tomić, W. Kempton / Journal of Power Sources 168 (2007) 459–468 463

where the vehicle is connected, and (b) the stored energy in Table 1
the battery divided by the time it is used. The electrical cir- Vehicle characteristics of Th!nk City
cuit limit is computed from the circuit’s ampere capacity (A), Vehicle characteristics Th!nk City
multiplied by the circuit’s voltage (V). This term we call the Battery type NiCd, 100 Ah 19 modules 6 V
power capacity of the line or Pline . For example, with home Energy stored (kWh) 11.5
wiring at 240 V AC, and a 50 A circuit rating typical for a large- Maximum depth of discharge (%) 80
current appliance such as an electric range, the power capacity Maximum power to motor (kW) 27
is 50 A × 240 V, or 12 kW. Based on practical limits on typical Effveh (miles/kWh) 5.71
Max range (miles) 53
home and commercial circuits, here we use 15 kW as the Pline Battery cycle life (cycles)a 1500
limit. Battery cost OEM ($/kWh) 300b (600c )
The limit imposed on the electrical power capacity for V2G Replacement labor (h) 8
by the vehicle (Pvehicle ) is a function of the energy stored onboard a At 80% depth of discharge.
(i.e. in the batteries), the dispatch time needed, and the driver’s b OEM cost, from verbal communication Lipman [21].
requirement for driving range. The formula for calculating Pveh c Retail cost that individual customers pay for replacing the battery pack [22].
for battery EDVs is shown in Eq. (10):
  Table 2
Es DoD − ddη+d veh
rb
ηinv Comparison of Pline of Th!nk City depending on the limit of the vehicle and
Pveh = (10) station electrical connections
tdisp
Types of electrical connections Ampere Voltage (V) Pline (kW)
capacity (A)
where Pveh is power capacity in kW, Es the stored energy avail-
able in kWh, DoD the maximum depth of discharge of the Th!nk City connected at a station 14 208 2.9
Station electrical connection 30 208 6.2
battery, usually 80% for NiMH and 100% for Li-Ion batteries,
dd the distance driven in miles since the battery was full (we use
16 miles as half of the US average daily vehicle miles traveled 4. Value of V2G power using utility fleets
[19], drb the range buffer required by the driver in miles and is
equal to 20 miles based on the minimal range required by US Using the general equations defined in the previous section,
drivers [20], ηveh the vehicle driving efficiency in miles kWh−1 , we use two actual utility fleets as case studies and calculate the
ηinv the efficiency of the inverter and other power electronics net revenue from those fleets selling regulation from V2G.
(dimensionless) with a value of 0.93, and tdisp is the dispatch
time in h. The dispatch time will be a fraction of the plugged-in 4.1. Fleet Case A
time.
The electrical power capacity for regulation is determined This fleet consists of 100 Th!nk City cars leased by Ford
by the limits imposed by Pline rather than the Pveh . When V2G Company to commuters in New York State, under management
is used for regulation Pveh is a much higher value than Pline of New York Power Authority (NYPA).1 Participants drive the
due to short instantaneous dispatch time (usually on the order vehicles from their home to the commuter station in the morning,
of 1–4 min). More details on this can be found in our recent charge up at the station using charging stations there, and com-
paper [8]. In the calculations in the present paper, we use sev- mute home in the afternoon. Chargers are also being installed
eral values for power capacity. We use 15 kW as the upper limit at the homes. The specifications of the Th!nk City are listed in
of typical wiring circuits, 6.6 kW as the limit given by Level Table 1.
2 chargers, and any lower limits imposed by the electronics The electrical power capacity available for regulation for this
on the vehicle itself (as we will see in the example of Th!nk vehicle fleet is limited by the electrical connections on the vehi-
City vehicles). When the vehicle is providing regulation down cle and at the station, in other words by Pline . The Th!nk City has
only (power flowing from grid to vehicle), the power capacity on-board electrical connections of 14 A while the station elec-
will be defined by wiring and the electronics (Pline ), but stor- trical connections are at 30 A. Table 2 lists the different Pline
age capacity of the battery and DoD will determine how long capacities based on the limit of the vehicle’s electrical connec-
the vehicle will be plugged-in (tplug ) before the battery is full. tions or the limit of the station’s electrical connections. In our
Substituting Edisp = Es DoDηcharger and P = Pline into Eq. (2) and calculations, we use both power capacities of 2.9 and 6.2 kW.
then rearranging it, we arrive at Eq. (11) Using Eqs. (7)–(9) and cost of purchased electricity of US$
Es DoD ηcharger 0.05/kWh we calculate the cost of energy for regulation (cen ).
tplug = (11) Cost of energy for regulation up and down is US$ 0.16/kWh
Pline Rd-c
and the annualized capital cost per vehicle is US$ 171. For reg-
where ηcharger is the efficiency of the charger, or efficiency of line
AC to battery charge, with a value of 0.93. In regulation down 1 The program was launched in 2002 with the actual number of vehicles in
only mode, we assume that DoD is 50% at the start so after operation varying with the gradual phase-in and phasing-out. Although the cur-
the battery is fully charged, the vehicle will not be available to rent number of vehicles in operation may be smaller, we base the analysis on
provide regulation down. the 100 vehicles.
464 J. Tomić, W. Kempton / Journal of Power Sources 168 (2007) 459–468

Table 3 power capacity of the vehicle. The net profits span from a low
Cost of energy and annualized capital cost for V2G power using Th!nk City of US$ 19,000 to a high of US$ 51,000 reflecting the different
Ancillary service cen (US$ kWh−1 ) cac (US$) market prices of regulation in the 4 years analyzed and are not
Regulation up and down 0.16 171
sensitive to the power capacity of the vehicle. The reason that the
Regulation down only 0 25 profits are not sensitive to power capacity is because the number
of hours that the vehicle can be plugged in each day decreases
with the increase in power capacity. Based on the above results
ulation only, the cost of energy is 0 and annualized capital cost and the values from Table 5 for regulation up and down, we
per vehicle is much less, only US$ 25. The results are listed in conclude that for this EDV fleet it is more lucrative to provide
Table 3. only regulation down.
To calculate the value of V2G power for regulation we use
market data from New York ISO. Table 4 lists the average market 4.2. Fleet Case B
values for regulation for 4 consecutive years. The values shown
are average yearly values in the day-ahead market expressed Our second case study is an investor-owned utility with a
in US$/MW-h, and apply to both regulation up and regulation substantial fleet of battery EDVs. This utility’s EDV fleet con-
down. In some markets, there are separate prices for regulation sists of 252 Toyota RAV4 EDVs.2 The fleet vehicles are in use
up and for regulation down, as we will see later. mostly for meter reading during the day and are parked after
The market price for regulation in the NYISO ranges from 3 p.m. when their state of charge is between 30 and 50%. They
US$ 11–27/MW-h with 1 year having the lowest value (i.e. US$ are thus available for V2G from 3 p.m. to 8 a.m., or a total 17 h
11/MW-h in 2001). As we will see later, this will affect the per day. The specifications of Toyota RAV4 EDVs are listed in
results. The use of the commuter vehicles is estimated at 1 h Table 7.
each day and thus their availability for V2G, or tplug , is 23 h The power capacity for regulation is limited by the capacity
per day. Using the market value for regulation and Eqs. (1)–(4) of the electrical connections or in this case primarily it is limited
we calculate the annual profit for the Th!nk City fleet providing by the charger or limits of home or commercial circuits. We
regulation up and down. The cost and net profit values for the 4 use two capacities: 6.6 kW reflecting today’s (Level 2) vehicle
years are shown in Table 5. The costs include annualized capital chargers and 15 kW based on practical limits on a home or typical
costs for providing regulation. commercial circuit. At the 6.6 kW the total power capacity of
Regulation up and down from this fleet is profitable in all this fleet is 1.67 MW and at 15 kW it is 3.78 MW. The cost of
years with the exception of 2001 when the market clearing price energy and annualized capital costs for RAV4 providing V2G
was very low causing negative profits. The market price of regu- power are listed in Table 8.
lation in 2001 was too low making V2G power for regulation not The historical market clearing prices for regulation up and
profitable. On the other hand this vehicle fleet at 2003 market down for the California ISO (CAISO) are listed in Table 9. The
prices would net over US$ 20,000 at 2.9 kW, and around US$ prices are the average yearly values of the day-ahead market
70,000 if upgraded to 6.2 kW. The profits for this particular fleet clearing prices in US$/MW-h with separate prices for regulation
are more interesting at higher power capacity of 6.2 kW. up and regulation down.
We can use the same fleet for regulation down only. The Separation of the regulation up and regulation down also
availability of the vehicle to provide regulation down is smaller allows for the EDVs to bid into both markets simultaneously.
and restricted by the state of charge of the battery at the point Based on the above data and Eqs. (1) and (3), we first calculate
of starting to provide regulation down. We assumed in our cal- the net annual profit for providing regulation up and down from
culations that the battery is 50% charged at the initial point of this battery EDV fleet. The results are shown in Table 10.
connecting for V2G. Using Eq. (11), we obtain tplug of 18 h (per The profits range between roughly US$ 150,000 and 2.1 mil-
day) at 2.9 kW and 8.6 h at 6.2 kW. This is the maximum num- lion. This large span reflects fluctuations in the market price for
ber of hours this vehicle can provide regulation down before the regulation, but even in the year with the lowest market prices,
battery is fully charged. Using Eqs. (3) and (6), we obtain the the calculated profit for the fleet is considerable—around US$
annual cost and net profit for this fleet providing only regulation 150,000 at 6.6 kW and around US$ 350,000 at 15 kW. Provid-
down. The results are listed in Table 6 for 4 different years. ing V2G power for regulation services (up and down) is very
The cost amount comes directly from the annualize capital profitable for this particular case.
cost per vehicle (US$ 25, see Table 3) and is independent of the Another option is for the vehicle fleet to provide only regula-
tion down service. The tplug will depend on the power capacity
Table 4 and using Eq. (11) we calculate that tplug is 19 h for 6.6 kW (same
New York ISO average market prices for regulation as the total available time of the vehicles) and 8.5 h for 15 kW.
Year Regulation price (US$/MW-h) Using Eqs. (3) and (6) we calculate the value of providing reg-
ulation down only. Table 11 lists the calculated annual cost and
2000 20.9
2001 10.9
2002 19.7
2 The actual number of vehicles may vary from year to year depending on
2003 27.5
maintenance and repair issues.
 J. Tomić, W. Kempton / Journal of Power Sources 168 (2007) 459–468 465

Table 5
Annual profit for Th!nk City fleet of 100 vehicles providing regulation up and down
Power per vehicle (kW) Cost fleet (US$) Fleet annual net profit (US$)

Year 2000 Year 2001 Year 2002 Year 2003

2.9 55,500 7, 300 −17,000 4,300 23,200

6.2 99,500 35, 300 −16,800 28,700 69,400

Table 6
Annual net profit for Th!nk City fleet of 100 vehicles providing regulation down only
Power per vehicle (kW) Cost (US$) Annual net profit (US$)

Year 2000 Year 2001 Year 2002 Year 2002

2.9 2,500 38, 600 18,900 36,100 51,500

6.2 2,500 38, 500 18,800 36,000 51,300

Table 7 Table 11
Technical characteristics of the Toyota RAV4 EDV Value of V2G power from the RAV4 EDV fleet for regulation down only in
CAISO market
Vehicle characteristics Toyota RAV4
Fleet power Cost Annual net profit (US$)
Battery type NiMH (kW) (US$)
Energy stored (kWh) 27.4
Year Year Year Year
Maximum depth of discharge (%) 80
2000 2001 2002 2003
Effveh (miles/kWh) 3.65
Maximum range (miles) 80 At 6.6 kW
Battery cycle life (cycles)a 1750 1,683 6,300 150,800 397,700 135,700 200,900
Battery cost OEM (US$/kWh)b 350
At 15 kW
Replacement labor (h) 10
4,233 6,300 172,300 452,800 155,100 229,200
a From Battery Panel Report 2000 [23].
b More recent Ni–metal hydride models.
annual net profit for the fleet of 250 RAV4 providing regulation
Table 8
Cost of energy and annualized capital cost for V2G power using Toyota RAV4 down only.
The range of profits is from US$ 135,000 to 450,000 per
Ancillary service cen (US$/kWh) cac (US$)
year. The profits are much lower when providing only regulation
Regulation up and down 0.15 90 down, but nonetheless may be attractive given that equipment
Regulation down only 0 25 changes and certification would be reduced.
In summary, this vehicle fleet could profitably provide V2G
Table 9 power for regulation up and down or for regulation down only.
Average annual CAISO market prices for regulation Most profitable though is regulation up and down because it pro-
Year Regulation up Regulation down vides twice as much regulation due to the battery not filling up.
(US$/MW-h) (US$/MW-h)

2000 54.5 15.4 5. Value of V2G power from fleets in other A/S markets
2001 62.5 39.7
2002 12.9 14.0 The two cases evaluated these two fleets in their home ISOs,
2003 19.5 20.3 that is, in the ancillary service markets they actually would par-
ticipate in. In order to evaluate the potential and profitability of
Table 10 V2G power for regulation services in other A/S markets (other
Value of V2G power from the RAV4 EDV fleet for regulation up and down in ISO areas), we take the same two fleet cases and calculate the
CAISO market net profits if these fleets were providing regulation in other A/S
Fleet power Cost Annual net profit (US$) market regions. The four ISO markets we examined are NYISO,
(kW) (US$) CAISO, ERCOT (Texas), and PJM (Pennsylvania, New Jersey,
Year 2000 Year Year Year and Maryland ISO)3 . The market clearing prices for regula-
2001 2002 2003 tion for 4 years in these four different A/S markets are listed
At 6.6 kW in Table 12.
1,683 180,000 584,900 912,000 144,800 277,600
At 15 kW 3 In the US currently there are two other ISOs: New England ISO and Mid
4,233 380,000 1,358,000 2,102,000 358,000 659,700
West ISO.
466 J. Tomić, W. Kempton / Journal of Power Sources 168 (2007) 459–468

Table 12
Market clearing prices for regulation in several A/S markets (in US$/MW-h)

Year NYISO regulation CAISO ERCOT PJM regulationa

Regulation up Regulation down Regulation up Regulation down

2000 20.9 54.5 15.4 –b –b 35.9

2001 10.9 62.5 39.7 7.7 7.2 34.3
2002 19.7 12.9 14.0 6.5 5.1 31.7
2003 27.5 19.5 20.3 22.1c 7.9c 38.2
a The data are for the PJM East market. Regulation in PJM West is not yet provided via a competitive market.
b The ERCOT market started in summer of 2001.
c Includes data January–March 2003.

We examine a fleet of 100 Th!nk City vehicle providing V2G markets seem relatively similar, with a range from around US$
power for regulation services in the four different A/S markets. 40,000 and the high around US$ 450,000. The CAISO market
All the costs and revenue calculations are the same as in Case showed very high net profits in 2000 which was not a typical
A with the exception of changing the market clearing price for year for this market. The 2003 prices are more typical and the
regulation depending on the specific A/S market. The results profits more similar among the four markets. As expected, the
are presented in Fig. 2 with case (a) for power level of 2.9 kW, profits are larger at higher power levels.
if limited by vehicle’s electrical connections, and case (b) for Encouraged by the results for providing regulation down only,
power level of 6.2 kW, if limited by station electrical connections we analyzed this option in different A/S markets. The results are
(for details on electrical connections see Table 2) (Fig. 3). summarized in Fig. 4 with case (a) power level of 2.9 kW and
The results show that a fleet of 100 Th!nk City EDVs could case (b) power level 6.2 kW.
provide regulation services with a net profit in three of the four The earlier conclusions for providing regulation down only
markets. The ERCOT and the NYISO showed lower profits, are supported in other A/S markets as well. While the maximum
and in few instances negative profits, when the market price of profits are lower in this case, they are consistently positive across
regulation was relatively low in a particular year. The other three the different markets and years making this a clearly interesting

Fig. 2. Annual net profits in different ISO markets for a 100 Th!nk City fleet Fig. 3. Annual net profits in different ISO markets for a 100 Th!nk City fleet
providing regulation up and down at (a) 2.9 kW and (b) at 6.2 kW. providing regulation down at (a) 2.9 kW and (b) at 6.2 kW.
 J. Tomić, W. Kempton / Journal of Power Sources 168 (2007) 459–468 467

in battery development and address the abovementioned

challenges.
A number of institutional barriers should be addressed as
well. These include: (1) lack of vehicles aggregators to manage
multiple fleets and individual vehicles, (2) regulation signal is
not broadcast by all ISOs, (3) rates for regulation services are
not available at the retail level, (4) no mass production of V2G
capable vehicles, and (5) need for standards for V2G provision
quality.

7. Conclusions

We have analyzed the use of V2G power from battery-electric

fleets to provide regulation, which is a short-duration but high-
value power market. The results vary across fleets and A/S
markets which demonstrates the importance of fleet and region-
specific analysis of economic attractiveness. Factors that emerge
as important variables are: (a) the value of ancillary services in
the area, (b) the power capacity (kW) of the electrical connec-
tions and wiring, and (c) the kWh capacity of the vehicle battery.
The amount of time the vehicles were on the road or discharged
did not turn out to be a major variable. The results show that
battery EDV fleets have significant potential revenue streams
from V2G. In general, larger profits come from providing V2G
power for regulation up and down but regulation down only
option can be more attractive for certain vehicles and/or A/S
Fig. 4. Annual net profits in different ISO markets for a fleet of 252 RAV4 markets. This should be evaluated for the specific fleet and A/S
providing regulation up and down at (a) 6.6 kW and (b) at 15 kW. market combination.
EDVs can provide regulation of higher quality than cur-
option for this particular fleet. However, each market should be rently available—fast response to a signal, available in small
evaluated on its own to determine if it is economically more inter- increments, and distributed. From the perspective of the electric
esting to provide both regulation up and down or only regulation power sector, this is a new source of high quality grid regula-
down. tion. For the EDV owners this is a significant revenue stream that
We also calculated the net profits of a fleet of 252 Toyota would improve the economics of grid-connected EDVs and fur-
RAV4 EDVs in these four different A/S markets. The results are ther encourage their adoption. The additional use of clean EDV
presented in Fig. 4 with case (a) at 6.6 kW power level and case vehicles not only for transportation but as a source of power has
(b) at 15 kW power level. benefits for both the electric grid sector and the transportation
The results are very positive with high net profits in most sector.
of the examined A/S markets. For the years we examined,
the weakest market for V2G power seems to be ERCOT and Acknowledgements
NYISO. However, the other two A/S markets (CAISO and PJM)
Sections of the paper are drawn from a paper presented at the
show very high profits for V2G power (e.g. US$ 2 million and
20th International Electric Vehicle Symposium and Exposition,
600,000). Overall these results are very encouraging for the
(EVS2); this article has been greatly modified and expanded.
prospects of V2G power from fleets of EDVs for regulation
This work was funded in part by a grant from Steven and Michele
services.
Kirsch Foundation. We are grateful to the fleet managers and util-
ity staff who provided data. Prior analysis of individual vehicles,
6. Making V2G possible upon which part of this article is based, was funded by the Cal-
ifornia Air Resources Board and the Los Angeles Department
Our calculations show that V2G power from EDV fleets is of Water and Power. We also would like to acknowledge sup-
economically feasible. To allow implementation of V2G sev- port from Conectiv (now Delmarva Power). The statements and
eral barriers should be addressed. On the technical side these conclusions are those of the authors and not necessarily those of
are related to batteries. First, the current batteries are not specif- the funders.
ically designed and optimized for EVs. Second is the issue of
battery cycle life which needs to be higher than in current bat- References
tery designs to support a greater number of charge/discharge
cycles. The recent increased interest in hybrid and even plug- [1] W. Kempton, S. Letendre, Transportation Research D 2 (1997) 157–175.
in hybrid vehicles will likely increase the rate of progress [2] W. Kempton, T. Kubo, Energy Policy 28 (2000) 9–18.
468 J. Tomić, W. Kempton / Journal of Power Sources 168 (2007) 459–468

[3] S. Letendre, W. Kempton, Public Utilities Fortnightly 140 (2002) 16–26. California Environmental Protection Agency, December 2002, available at:
[4] D. Hawkins, Presentation at the EVAA Electric Transportation Industry http://www.acpropulsion.com/white papers.htm.
Conference, Sacramento, CA, 13 December, 2001, Presentation slides [13] A. Brooks, unpublished data.
available at: http://www.acpropulsion.com/reports/Hawkins ETI.pdf. [14] N. Raman, K. Chagnon, K. Nechev, A. Romero, T. Sack, M. Saft, Proceed-
[5] A. Brooks, Presentation at the EVAA Electric Transportation Industry ings of the 20Th International Electric Vehicle Symposium and Exposition,
Conference, Sacramento, CA, 13 December, 2001, Presentation EVS 20, Long Beach, CA, 15–19 November, 2002.
slides available at: http://www.acpropulsion.com/reports/A%20Brooks% [15] J. Miller, R. Brost, Presented at the Advanced Automotive Battery Confer-
20ETI%20conf.pdf. ence, Las Vegas, NV, February, 2001.
[6] T. Lipman, J. Edwards, D. Kammen, Economic Implications of Net [16] AC Propulsion AC-150 Gen-2 EV Power System: Integrated Drive
Metering for Stationary and Motor Vehicle Fuel Cell Systems in and Charging for Electric Vehicles, available January 2007 at:
California, University of California, Berkeley, 2002, Available at: http://www.acpropulsion.com/technology/gen2.htm.
http://socrates.berkeley.edu/∼rael/papers.html#fuelcells. [17] A. Brooks, Personal Communications (2001).
[7] W. Kempton, J. Tomić, S. Letendre, A. Brooks, T. Lipman, Vehicle-to-Grid [18] A. Coolins, W. Koon, A temper-resistant watt hour energy meter
Power: Battery, Hybrid, and Fuel Cell Vehicles as Resources for Distributed based on the AD7751 and two current sensors, 2002, available at:
Electric Power in California, Davis, CA. Institute for Transportation Stud- http://www.analog.com.
ies Report # UCD-ITS-RR-01-03, 77+xiv pages, June, 2001. PDF also [19] P. Hu, J. Young, 1995 Nationwide Personal Transportation Survey,
available at http://www.udel.edu/V2G. US Department of Transportation, 1999, available at: http://www-
[8] W. Kempton, J. Tomić, J. Power Sources 144 (2005) 279–285. cta.ornl.gov/npts/1995/Doc/trends-report.pdf.
[9] W. Kempton, J. Tomić, J. Power Sources 144 (2005) 280–294. [20] K. Kurani, T. Turrentine, D. Sperling, Transport Policy 1 (1994) 244–256.
[10] K. Brendan, E. Hirst, The Electricity Journal 14 (2001) 48–55. [21] T. Lipman, Personal Communications (2000).
[11] CAISO, California Independent System Operator, Weekly Market Watch [22] H. Schon, Personal Communications (2001).
(2001), available at http://www.caiso.com/marketanalysis. [23] M. Anderman, F. Kalhammer, D. MacArthur, Report: Advanced Bat-
[12] A. Brooks, Final Report: Vehicle-To-Grid Demonstration Project: Grid teries for Electric Vehicles: An Assessment of Performance, Cost, and
Regulation Ancillary Service with a Battery Electric Vehicle, Contract Availability, Prepared for California Air Resources Board Battery Panel,
number 01-313, Prepared for the California Air Resources Board and the Sacramento, CA, 2000.