ZEBRA1.pdf

(1492 KB) Pobierz

IEEE TRANSACTIONS ON INDUSTRIAL ELECTRONICS, VOL. 57, NO. 3, MARCH 2010

943

Electric Vehicle Using a Combination of

Ultracapacitors and ZEBRA Battery

Juan Dixon,

Senior Member, IEEE,

Ian Nakashima, Eduardo F. Arcos, and Micah Ortúzar

Abstract—The

sodium–nickel chloride battery, commonly

known as ZEBRA, has been used for an experimental electric

vehicle (EV). These batteries are cheaper than Li-ion cells and

have a comparable speciﬁc energy (in watt–hours per kilogram),

but one important limitation is their poor speciﬁc power (in watts

per kilogram). The main objective of this paper is to demonstrate

experimentally that the combination of ZEBRA batteries and

ultracapacitors (UCAPs) can solve the lack of speciﬁc power,

allowing an excellent performance in both acceleration and regen-

erative braking in an EV. The UCAP system was connected to the

ZEBRA battery and to the traction inverter through a buck–boost-

type dc–dc converter, which manages the energy ﬂow with the help

of DSP controllers. The vehicle uses a brushless dc motor with a

nominal power of 32 kW and a peak power of 53 kW. The control

system measures and stores the following parameters: battery

voltage, car speed to adjust the energy stored in the UCAPs, in-

stantaneous currents in both terminals (battery and UCAPs), and

present voltage of the UCAP. The increase in range with UCAPs

results in more than 16% in city tests, where the application of

this type of vehicle is being oriented. The results also show that

this alternative is cheaper than Li-ion powered electric cars.

Index Terms—Energy

management, energy storage, road vehi-

cle electric propulsion.

Fig. 1. Speciﬁc power of the most common secondary batteries for EVs.

I. I

NTRODUCTION

ODIUM–NICKEL chloride batteries (ZEBRA) are a good

choice for electric vehicles (EVs) [1], [2]. They are safe

and low cost and can endure more than 1000 cycles without

signiﬁcant degradation [3]. Moreover, they can be discharged

almost to 100% of its total capacity without degradation in

its cycle life. Its speciﬁc energy (in watt–hours per kilo-

gram) is comparable with high-quality batteries, like Li-ion

(120 Wh/kg). However, speciﬁc power (in watts per kilogram)

of ZEBRA batteries is rather low when compared with other

batteries as shown in Fig. 1 [4].

For example, Li-ion batteries have almost three times more

speciﬁc power than ZEBRA (around 400 W/kg), but its price in

terms of dollars per kilowatt hour is around three times higher.

Manuscript received January 11, 2009; revised July 6, 2009. First published

July 28, 2009; current version published February 10, 2010. This work was

supported in part by Fondecyt Project 1070751 and in part by Millenium Project

P-04-048-F.

J. Dixon is with the Department of Electrical Engineering, Pontiﬁcia Univer-

sidad Católica de Chile, 6904411 Santiago, Chile (e-mail: jdixon@ing.puc.cl.).

I. Nakashima is with CGE Transmisión, 8340434 Santiago, Chile (e-mail:

inakashi@gmail.com).

E. F. Arcos is with Woodtech S.A., 7550130 Santiago, Chile (e-mail:

farcos@gmail.com).

M. Ortúzar is with CAM-Endesa, 8330287 Santiago, Chile (e-mail:

mortuzar@gmail.com).

Color versions of one or more of the ﬁgures in this paper are available online

at http://ieeexplore.ieee.org.

Digital Object Identiﬁer 10.1109/TIE.2009.2027920

Today, it is quite difﬁcult to get a Li-ion battery at an affordable

price. The cheapest Li-ion batteries available today in the mar-

ket are the small-cell-type 18650 (3.6 V, 2.4 Ah), from which

a battery car, using around 4000 units, can be implemented [5].

However, the lowest market price per unit, buying more than

1000 of these cells, is US$4.75 per unit [6], which means that

for 4000 units, the cost is US$19 000. This value does not

consider the large amount of electronic circuits for voltage

balancing, temperature control, and current balancing, which

may increase the total cost to more than US$40 000. Besides,

this battery pack will need the container, reinforcements, me-

chanical protections and time to construct, which will increase

the cost of the battery, probably to more than US$50 000.

The maker of electric cars, AC Propulsion Systems, sells the

“TZero” EV, made with 6800-cell-type 18650, at a cost of

US$220 000. They also sell the same EV with lead-acid bat-

teries at US$80 000 [7], [8]. This difference in price means

that this battery with 6800 cells, with all the electronic sup-

port for safe operation, costs more than US$140 000. Then,

for 4000 cells, the cost is higher than US$82 000 (around

US$2400/kWh). AC propulsion decided to make the car with

those small units because this solution demonstrated to be

cheaper than a large Li-ion module (which is not easily avail-

able today). Another example is the Chevrolet “Volt” from GM.

This hybrid plug-in EV (also deﬁned as range-extended EV) is

going to be commercialized by the end of 2009. It uses a Li-

ion battery pack of 16 kWh and GM estimates that the battery

alone will cost around 20 000 (US$800/kWh). However, this is

a very unreal estimation because, at present, the price of Li-ion

batteries for EVs is more than US$2000/kWh [9]. The price of

“TZero,” from AC Propulsion Systems, is a good example of

cost of Li-ion batteries today. The experimental vehicle under

study uses a ZEBRA battery of 28.2 kWh with a cost on the

order of US$19 000 [10] (real cost in March 2009), which

includes auxiliary control circuits (US$680/kWh). The ZEBRA

Authorized licensed use limited to: Pontificia University Catolica de Chile. Downloaded on March 26,2010 at 09:32:14 EDT from IEEE Xplore. Restrictions apply.

944

IEEE TRANSACTIONS ON INDUSTRIAL ELECTRONICS, VOL. 57, NO. 3, MARCH 2010

Fig. 2. ZEBRA-type Z36-371-ML3X-76.

Fig. 4.

UCAP bank in front of the vehicle.

Fig. 5. Block diagram of the power system to manage the energy ﬂow into the

vehicle.

Fig. 3. EV prototype.

being used in the experimental EV comes from factory in only

one pack, easy to install, which weights only 245 kg. It stores

28 kWh of useful energy, which gives a speciﬁc energy of

115 Wh/kg (the speciﬁc energy of the battery used by the

Chevrolet “Volt” is only 90 Wh/kg).

Fig. 2 shows the ZEBRA battery (type Z36-371-ML3X-76)

that is being used in the experimental EV [2]. This EV is shown

in Fig. 3 and was implemented at the Department of Electrical

Engineering of the Pontiﬁcia Universidad Católica de Chile

[11]. The car is powered by a 53-kW brushless-dc traction

motor and has a gross weight of 1700 kg [12].

To solve the lack of power problem, an ultracapacitor

(UCAP) bank was installed [13]–[15]. This bank, with a total

capacity of 20 F and 300 Vdc, stores a practical amount

of 200 Wh of energy. However, with this small amount of

energy, but with the big amount of speciﬁc power (more than

1000 W/kg), the UCAP can easily deliver 40 kW of power

during 20 s (more than enough to solve the lack of power of the

ZEBRA during the acceleration). In a similar way, during

regenerative braking, the UCAPs can receive, in a short period

of time, a high amount of energy. This is quite important be-

cause sudden peaks of negative power can increase the ZEBRA

battery voltage dangerously [16]–[18]. The actual cost of the

UCAP bank installed on the vehicle is around US$9000. How-

ever, road tests have demonstrated that a small capacitor bank

(80 Wh), with a cost of US$3000, is adequate for the accelera-

tion and regenerative braking (the weight of this small bank is

around 30 kg with frame included). Fig. 4 shows one of the

packs of UCAPs installed in front of the vehicle. The com-

bination of ZEBRA (high amount of energy but low speciﬁc

power) with UCAPs (low amount of energy but high speciﬁc

power) allows making an electric car [19], with better range,

good acceleration, and full regenerative braking capability [20].

II. P

OWER

IRCUIT

Fig. 5 shows the block diagram of the power circuit used in

the EV to manage the energy ﬂow between ZEBRA, UCAPs,

and the traction system [21]. The battery feeds the power

inverter, and when the battery voltage goes low, the UCAPs

inject energy to both ZEBRA and inverter through the dc–dc

converter. This process keeps the battery voltage at normal val-

ues and helps the EV during acceleration. During regenerative

braking, the UCAPs recover the energy to avoid overvoltages at

the battery terminals. The dc–dc converter is water cooled and

weighs only 15 kg. It was designed and implemented at the De-

partment of Electrical Engineering, and the cost of implemen-

tation was less than US$2000. Taking into account the prices of

ZEBRA (US$19 000), UCAPs (US$3000), and dc–dc converter

(US$2000), the total cost of this system is US$851/kWh, which

is far lower than Li-ion price (today around US$2000/kWh).

Special control algorithms keep the capacitor voltage at

the required level, according to particular driving conditions

(mainly speed, battery state of charge (SOC), and UCAP

voltage).

Authorized licensed use limited to: Pontificia University Catolica de Chile. Downloaded on March 26,2010 at 09:32:14 EDT from IEEE Xplore. Restrictions apply.

DIXON

et al.:

ELECTRIC VEHICLE USING A COMBINATION OF ULTRACAPACITORS AND ZEBRA BATTERY

945

UCAPs is needed. In this way, all the energy recovered from

regenerative braking will go to the UCAPs. By contrast, if the

battery SOC is poor, the UCAP should keep an amount of

energy higher than under normal conditions. The charge curves

of the UCAP for those operating conditions are shown in Fig. 8.

These curves were estimated, taking in account the time needed

by the control to store the right amount of energy into the

UCAP. This control, using the same TMS320F241, was also

implemented using neural networks strategy [22], [23].

The SOC is estimated by time integration of the battery

current (positive or negative). The system also recognizes a

fully charged battery when its voltage goes up rapidly under

a regenerative braking condition.

IV. S

IMULATION

ESULTS

Fig. 9 shows a simulation of the ZEBRA battery at 80%

depth of discharge (DOD) during acceleration of the vehicle

from 40 to 60 km/h in 4 s. The UCAPs are not connected, and

it can be seen that the battery voltage is strongly affected by the

current variations. A serious problem arises when battery has

to supply more than 150 A, because, in this case, the voltage

at the ZEBRA terminals drops to values smaller than 250 Vdc.

Under these conditions, the control of the traction motor reset

the PWM signals of the inverter due to undervoltage operation,

and the current cannot go higher. The simulation shows, in

dot lines, the unreachable current when voltage goes below

250 V, because, in fact, the vehicle cannot work under these

conditions.

The deep variation of the ZEBRA voltage is due to the poor

speciﬁc power of this kind of battery. The only way to avoid

this problem without power assistance is to keep a very low

acceleration value.

Fig. 10 shows the ZEBRA battery at the same 80% DOD,

but with UCAPs, accelerating the vehicle from 40 to 60 km/h

in 4 s. In this case, the battery voltage is not seriously affected

by the current variations because UCAPs are injecting power to

the system to keep the ZEBRA current limited to 70 A. Under

this condition, the voltage cannot go lower than 320 Vdc unless

UCAPs become fully discharged. After acceleration comes to

an end, the control adjusts the UCAP voltage according to

car speed and battery SOC. During constant speed, the battery

voltage recovers its normal operating value (around 370 V,

depending on current released by the ZEBRA battery at that

speed). As was already mentioned, at higher speeds, UCAPs

are fully discharged (at factory limits). When a vehicle travels

at medium speeds, the UCAPs keep some charge to have energy

for future acceleration. Similarly, the UCAPs must have some

room to receive energy during regenerative braking. When

the car is not moving, UCAPs are fully charged for good

acceleration, with minimum support from the battery.

The simulation in Fig. 11 shows the ZEBRA during regenera-

tive braking from 40 to 0 km/h in 2 s. UCAPs are not connected,

and it can be seen that the battery voltage goes higher than

400 Vdc when the current reaches around

−40

Adc.

However, under real conditions, the control of the inverter

would disrupt the PWM signals to avoid overvoltage condi-

tions. This operation is necessary to avoid a permanent damage

Fig. 6.

Control scheme of the system.

III. C

ONTROL

CHEME AND

ONTROL

IRCUIT

Fig. 6 shows the control circuit in block diagrams, and

Fig. 7 shows the control board implemented for the ZEBRA–

UCAP system. This control scheme was implemented in a

TMS320F241 DSP from Texas Instruments. In addition, a

monitoring feature was implemented in the DSP, which com-

municates with a portable PC. The monitoring program at the

PC allows real-time plotting and storing of all valuable data. In

addition, the control program at the DSP can be commanded

from the PC to work in slave (user-controlled currents) or auto-

mated mode.

The signals required to perform calculations are the follow-

ing: ZEBRA battery voltage, battery current, drive current,

UCAP voltage, input and output currents of the dc–dc con-

verter, UCAP current, battery SOC, and also vehicle speed.

Some of these signals are taken from the main microprocessor

that controls the power inverter. The rest of the signals are

acquired from specially installed sensors and an ampere–hour

counter installed in the vehicle. The control system outputs are

two pulsewidth-modulation (PWM) signals, which commutate

the two insulated-gate bipolar transistors in the buck–boost

dc–dc converter. The PWM is calculated as part of a closed-

loop PI control, comparing a preset current reference and the

measured current from the dc–dc converter. The power transfer

algorithm calculates the preset current value for the current

control, considering the battery SOC, the battery voltage, the

UCAP charge, the vehicle speed, and the power drive system

current.

The transfer algorithm adjusts the amount of energy stored in

the UCAPs according to car speed and battery SOC. The lower

the speed, the higher the UCAP charge. The speed information

is necessary because, at low speeds, more energy is required

for acceleration, and for high speeds, more space for storing

regenerative energy is needed. Similarly, if the batteries are

fully charged, only a small amount of energy stored in the

Authorized licensed use limited to: Pontificia University Catolica de Chile. Downloaded on March 26,2010 at 09:32:14 EDT from IEEE Xplore. Restrictions apply.

946

IEEE TRANSACTIONS ON INDUSTRIAL ELECTRONICS, VOL. 57, NO. 3, MARCH 2010

Fig. 7. DSP control circuit.

Fig. 11. Regenerative braking simulation from 40 km/h to stop in 2 s, without

UCAPs.

Fig. 8. Transfer algorithm charging curves for UCAP as a function of speed

and battery charge.

Fig. 9. Acceleration simulation from 40 to 60 km/h in 4 s, without UCAPs.

Fig. 12. Regenerative braking simulation from 40 km/h to stop in 2 s, with

UCAPs.

Fig. 10. Acceleration simulation from 40 to 60 km/h in 4 s, with UCAPs.

on the power inverter. The car will be unable to recover an

important part of kinetic energy, and the life of mechanical

brakes will be reduced.

Finally, the simulation in Fig. 12 shows again the ZEBRA

during regenerative braking from 40 to 0 km/h in 2 s. UCAPs

are now connected, and it can be seen that the battery voltage

never goes higher than 400 Vdc because most of the current

is now absorbed by the UCAPs. The UCAP current

CAP

proportional to

COMP

(they are related through the modulation

index of the dc–dc converter). In this case, the regenerative

braking works well because UCAPs take care of the currents,

avoiding battery voltage to go higher than 400 Vdc. It is im-

portant to mention that the control system can be programmed

according to the characteristics of the system. In this case,

two important limits have to be respected: the undervoltage

of 250 Vdc and the overvoltage of 400 Vdc. It is also worth

to mention that regenerative braking with only UCAPs does

not work in large downhills, because they ﬁnally will reach

their maximum voltage, being unable to receive more energy.

Under these conditions, the regenerative braking has to be

limited to the capacity of the ZEBRA to receive power without

reaching the voltage limit (400 Vdc). Other solution is to add

power resistors for downhill operation, but this option is very

inefﬁcient because the braking energy is not recovered.

Authorized licensed use limited to: Pontificia University Catolica de Chile. Downloaded on March 26,2010 at 09:32:14 EDT from IEEE Xplore. Restrictions apply.

DIXON

et al.:

ELECTRIC VEHICLE USING A COMBINATION OF ULTRACAPACITORS AND ZEBRA BATTERY

947

Fig. 15. Acceleration tests with UCAPs.

TABLE I

CCELERATION

ITH AND

ITHOUT

UCAP

Fig. 13. Urban circuit test course. In red: fast track. In blue: slow track.

Fig. 14. Acceleration tests without UCAPs.

V. E

XPERIMENTAL

ESULTS

The following oscillograms and tables show the performance

of the EV with the ZEBRA battery, when it operates with-

out and with the help of UCAPs. All the experiments were

performed with the full equipment installed (UCAPs are dis-

connected electronically). Tests road where made in a circuit

around the university campus, as shown in Fig. 13. The approx-

imate average speed was of 18 km/h, with maximum speeds of

80 km/h in the fast track showed in the external (red) rows.

A. Acceleration Tests

The oscillograms in Figs. 14 and 15 show the acceleration

tests without the UCAP bank and with the UCAP bank, re-

spectively. In Fig. 14, the EV needs around 25 s to reach the

80 km/h at full battery power capability

MAX

= 43

kW).

When

MAX

is reached, the battery voltage drops to 250 Vdc.

Under this situation, the control circuit disconnects the power

inverter due to undervoltage protection. However, when the

inverter is disconnected, the ZEBRA voltage goes up, and the

power inverter is connected again. A cyclical off–on operation

is created that can destroy the power inverter (very dangerous

repetitive operation).

When UCAPs are connected, the total power needed by the

power inverter is shared between the ZEBRA (P

MAX

= 29

now) and the UCAPs

MAX

= 45

kW), as shown in Fig. 15.

In this case, the power disruptions are eliminated because the

voltage never drops below 250 V. Moreover, as the traction

power has increased from 43 to 64 kW (29 kW

45 kW), the

time to reach 80 km/h has been reduced from almost 25 s to a

little more than 15 s. The total energy delivered by the UCAP is

around 160 Wh, but the stored energy can be restricted to reduce

the size of the UCAPs, because 29 kW delivered by the ZEBRA

is smaller than its

MAX

(43 kW). Moreover, if the acceleration

time is increased a little more (20 s), then the necessary energy

stored into the UCAPs will be around 70 Wh. This value is

around 33% of the total energy stored in the actual capacitor

bank (200 Wh). This reduction permits having an UCAP bank

that weighs only 30 kg (frame included) and with a cost of

around US$3000.

The regenerative power is now safer and more efﬁcient be-

cause overvoltages (over 400 Vdc) are avoided with the UCAP

system. During the regenerative braking, the UCAPs recover

part of the kinetic energy. In the cyclic process (stop–start–

stop), some energy is lost and has to be supplied by the battery

to recover the full charge into UCAPs. It is important to say

that the UCAPs do not supply energy during an acceleration–

deceleration cycle (this average energy is zero). They only

ﬂatten the energy delivered by the battery pack.

Table I shows the acceleration time, from zero to different

ﬁnal speeds (0–40, 0–60, and 0–80 km/h). It can be noticed that

the acceleration time is reduced and UCAP assistance becomes

more signiﬁcant for higher ﬁnal speeds.

The information about acceleration shown in Table I was

obtained with the ZEBRA battery almost fully charged (95%

SOC). When the ZEBRA SOC is low, results without UCAPs

become less effective, because battery voltage goes down more

Authorized licensed use limited to: Pontificia University Catolica de Chile. Downloaded on March 26,2010 at 09:32:14 EDT from IEEE Xplore. Restrictions apply.

ZEBRA1.pdf

Plik z chomika:

Inne pliki z tego folderu:

Inne foldery tego chomika: