Sizing Ultracapacitors For Hybrid Electric Vehicles

H. Douglas P Pillay
University of Cape Town Clarkson University
Department of Electrical Engineering Department of Electrical and Computer Engineering
Rondebosch 7701 Potsdam, NY,
Cape Town, South Africa USA 13699-5720
[email protected] [email protected]

Abstract – An efficient energy storage medium is essential in all prone to temperature effects as are batteries and can operate
hybrid electric vehicles. The advances in double layer in temperatures as low as –40oC [3].
electrolytic capacitor technology have opened new areas to Ultracapacitors cannot replace batteries completely,
complement batteries as a storage medium. In this paper we will however they can be used to complement each other. In the
review some of the present applications of ultracapacitors as
hybrid electric vehicle, the ultracapacitor can be used during
well as to provide guidelines for sizing ultracapacitors for
minimal mass in hybrid electric vehicles. Equations for both vehicle operation. During standstill a downsized battery can
constant current as well as constant power discharge are be used to power the auxiliary systems. A slow charge time
discussed. An iterative method for determining the minimum can used to charge the batteries and thus extending their
number of ultracapacitor cells is introduced. The effects of lifetime. Although batteries cannot deliver large currents at
ultracapacitor sizing on the rating of interface power electronics extremely low temperatures, they can deliver enough current
are examined. to slowly charge an ultracapacitor, which can be used to start
the vehicle. This could prompt the use of longer lasting
Keywords – Ultracapacitors, hybrid electric vehicles, batteries rather than large batteries designed to deliver
energy storage hundreds of amps during a cold start.
I. INTRODUCTION Available Performance Lead Acid Ultra Conventional
Battery Capacitor Capacitor
Ultracapacitor technology has been commercially
available for over the past decade. They can store much more Charge Time 1 to 5 hrs 0.3 to 30 s 10-3 to 10-6 s
energy than conventional capacitors and are available in sizes
Discharge Time 0.3 to 3 hrs 0.3 to 30 s 10-3 to 10-6 s
up to 4000F with voltage ratings of up to 3V per cell. They
can be discharged or charged faster than batteries and can Energy ( Wh/kg ) 10 to 100 1 to 10 < 0.1
deliver 10-20 times more power e.g. ultracapacitors typically
have 10 times the specific power (W/kg) as well as a much Cycle Life 1000 > 500 000 > 500 000
lower charge time when compared to lead acid batteries. Specific Power ( W/kg ) < 1000 < 10 000 < 100 000
They also offer 10 to 100 times the energy density (Wh/Kg)
of conventional capacitors. In terms of energy and power Charge / Discharge 0.7 to 0.85 0.85 to 0.98 > 0.95
density, ultra capacitors can therefore be placed between efficiency
batteries and conventional capacitors [1-2]. A comparison of Table 1. A Comparison of conventional storage technologies [1].
conventional storage technologies is shown in Table 1.
Ultracapacitors can be used as a temporary power source
for backup purposes. Batteries were first used for this II. THE DEFICIENCIES OF BATTERIES
application, however the large energy storage capability of
ultracapacitors makes them an alternative to batteries. In Although progress has been made in hybrid electrical
these applications the ultracapacitor is charged from the vehicle control, engine and motor design, very little has been
mains and provides temporary power during a power failure. done to improve the passive energy storage device, namely
[3] the battery. Cold weather adversely affects the operation of
The second application of ultracapacitors is important in batteries. Batteries have a very limited life cycle under
the field of energy storage in hybrid electric vehicles. extreme discharging conditions and need to be continuously
Ultracapacitors can be used to provide the short bursts of replaced throughout the lifetime of the vehicle. The cost
energy needed by hybrid electric vehicles during associated with the purchase of new batteries and the
acceleration. Ultracapacitors do not need regular replacement disposal of old ones is cumulative. Batteries are not
like batteries because they are not as adversely affected environmentally friendly and thus cannot be easily disposed
during repetitive deep charging and discharging. This also [3].
implies that ultracapacitors are more environmentally Batteries are the secondary power source in hybrid electric
friendly since they don’t need to be frequently discarded. vehicles. They provide short bursts of supplementary power
Ultracapacitors can be fully charged from a total discharge during acceleration, cold starting and even braking.
within a few seconds. This makes them ideal for use in However, not one battery type delivers all the power
regenerative breaking systems. Ultracapacitors are not a requirements needed by hybrid electric vehicles. Lithium

0-7803-9252-3/05/$20.00 ©2005 IEEE 1599

iron batteries can provide a small amount of current over a ultracapacitors are connected in series to form a string of
long time but cannot provide large bursts of power over a cells. Ultracapacitors, like conventional electrolytic
short time. They also have difficulty in absorbing energy capacitors, suffer from large deviations in their capacitance.
during regenerative breaking [3]. These deviations can be as high as 20% from the nominal
Lead acid and Ni-MH batteries are capable of delivering cell capacitance value. When connected in series, a common
the short bursts of power, however operating them under the current flows through all the cells and since the cell
extreme discharging conditions dramatically reduces their capacitances differ, the cell voltages will vary. This could
lifetime. As a solution to this problem, manufacturers parallel result in individual cells being overcharged and thus
many batteries to increase the power characteristics of the exceeding their voltage limits. For this reason, an active or
battery pack. This practice increases the weight and cost of passive voltage balancing circuit is employed to regulate the
the vehicle. Since the weight has been increased, the vehicle cell voltage.
will demand higher currents from the batteries and therefore The discharge profile for an ultracapacitor under constant
once again more batteries are needed. Clearly this is not the current is shown in Figure 2. A constant current discharge is
particularly useful when determining the parameters of the
best solution to the energy storage problem.
ultracapacitor. These parameters include the equivalent series
resistance, specific power and specific energy.
II. MODELING OF ULTRACAPACITORS

When sizing the ultracapacitor, it is necessary to

understand the implications of the various factors that not
only affect the capacitor, but will also affect the design of the
interface power electronics.

These factors affecting the choice of capacitor include:

• the peak capacitor voltage,
• allowable maximum percentage discharge,
• peak current flowing through the capacitor,
• capacitor time constant (τ),
• capacitance per cell, Figure 2. The voltage discharge profile for an ultracapacitor under
constant current discharge [6].
• cell voltage,
• number of cells needed, The equation relating the capacitance to the voltage,
• mass of the cell array, current and discharge time can be expressed as follows [6]:
• cost of the cell array.
I
An ultracapacitor can be modeled in a similar manner to C = (t + τ )
Vmax − Vmin (1)
conventional capacitors. The circuit schematic in Figure 1
represents the first-order model for an ultracapacitor. It’s
comprised of four ideal circuit elements: a capacitance C, a where,
series resistor Rs, a parallel resistor Rp, and a series inductor C is the capacitance,
L. Rs is called the equivalent series resistance (ESR). This I is the average current,
resistance contributes to the energy loss component of the Vmax is peak capacitor voltage,
ultracapacitor during charging or discharging. The parallel Vmin is the lowest voltage after discharge,
resistor, Rp, models the leakage current found in all and τ is the product of the equivalent series resistance and the
ultracapacitors. This leakage current varies from a few capacitance.
milliamps to tens of milliamps in large ultracapacitors. The
series inductance, L, is usually very small and is neglected Although equation (1) gives an insight to the properties of
for constant current charging / discharging applications. the ultracapacitor, it should be noted that this equation should
However, in many applications where there is a repetitive not be used to size ultracapacitors for constant power
charging / discharging, this inductance plays a significant applications. The capacitor voltage (under constant power
role especially at high frequencies. discharge) is described by equation (2). For simplicity the
equivalent series resistance is neglected but will be analyzed
later.

Figure 1. First order model of an ultracapacitor [5]. 2 2pt

V (t ) = Vmax −
C (2)
The voltage rating of ultracapacitors is typically less than where
4V. When higher voltage ratings are required, many p is the constant power,

1600
 t is the time,
C is the total capacitance of a string of capacitors, Vmax V(t)
I(t)
Vmax is the initial voltage,
V(t) is the capacitor voltage at time t. Vmin

Since the power being delivered is constant, the minimum

voltage can be determined based on the current conducting Imax
capabilities of the ultracapacitor. Equation (2) can then be
rewritten as
Imin
2 2pT T t
Vmin = V max −
C (3) Figure 3. The voltage and current waveforms for a constant power
where, discharge.
p is the constant power,
T is discharge the time,
C cell
C is the total capacitance of a string of capacitors, C =
Vmax is the initial voltage, n (7)
Vmin is the final voltage. where
n is the number of cells,
Similarly, the maximum current occurs at the minimum Vcell is the maximum cell voltage,
voltage and the expression for the current can be derived as Ccell is the individual cell capacitance.
follows.
Substituting these equations into (3) and (5) results in the
p following expression.
i(t ) =
2 2pt
V max −
C (4) 2pT p2
2
Vcell n2 − n− 2 =0
C cell I max
p (8)
I max = 2 2pT
V max −
C (5) In equation (8), the only variable is n since the cell
where, voltage, Vcell, and cell capacitance Ccell, is known. The
p is the constant power, maximum current, Imax is chosen based on the rating of the
T is discharge the time, individual cells. The variable n therefore represents the
C is the total capacitance of a string of capacitors, amount of capacitors required to deliver the constant power
over the time period.
Vmax is the initial voltage
Solving for n thus yields,
Imax is the final current at time T.

Equation (3) describes the voltage discharge profile for a 2pT 4p 2T 2 2 p2

± 2 + 4Vcell 2
specified constant power, p and time, T. The variables in the Ccell C cell I max
n = 2
equations are therefore the initial voltage Vmax and the 2Vcell (9)
capacitance C. It is common to make the mistake of choosing
a specific voltage and thus calculating the capacitance
The voltage discharge ratio is defined as the ratio of the
required, which often results in an over sizing of the
final voltage over the initial voltage. It is common to choose
ultracapacitors. The aim when sizing any string of capacitors
a value of 50%, resulting in 75% of the energy being utilized.
is to minimize the mass, which implies using the least
Equation (9) does not assume that the ratio is 50%, in fact it
amount of capacitors. The voltage function, V(t), and current
can be greater than or less than 50% depending on the
waveform, I(t), is shown in Figure 3.
maximum cell voltage and cell capacitance. This equation
It is important to note that the variables Vmax and C are in
holds for applications where constant power is needed at a
fact related by the common variable n, which is the number
very low current as well as capacitors with extremely low
of cells. The assumption is that the capacitors will never be
equivalent series resistances, because the voltage drop across
charged above the combined maximum voltage rating of all the equivalent series resistance is negligible. However in
the cells. We can therefore introduce this relationship with applications where high currents are drawn, the effect of the
the following equations. equivalent series resistance has to be taken into account. The
energy dissipated in the equivalent series resistance as well
Vmax = nVcell (6) as the cabling and connectors could result in an under sizing
of the number of capacitors required. For this reason an

1601
alternative approach to the capacitor sizing is necessary. cells, and the final value of each terminal voltage trajectory is
i(t) RESR compared to the target voltage. If the final value is lower
V(t)
than the target voltage then the number of cells is
incremented until it is greater than or equal to the target
voltage. In doing so, the minimum amount of cells can be
Constant
found.
E(t) Power
When the number of cells to meet the requirements has
Application
been found, it can be used to determine the initial current.
The initial current can then be used in (17) to show the
current waveform. The instantaneous power dissipation in
the equivalent series resistance can then be determined.
Figure 4. Equivalent circuit including the equivalent series
resistance. di(t ) i(t )
=−
dt ⎛ p ⎟⎞
C ⎜⎜ R − ⎟
The system depicted in Figure 11 is governed by equations ⎜⎝ i(t )2 ⎟⎠ (17)
(10-15).

dE (t ) The above numerical analysis can be applied to a hybrid

i(t ) = −C electric vehicle where 25kW is needed for 18s. The
dt (10) specifications of the capacitor used is listed in Table 2. The
maximum current for the capacitor is 400A. Based on this
1 maximum current and a power of 25kW, the minimum
E (t ) = −
C ∫ i(t )dt (11) terminal voltage, v(t), can be calculated to be 62.5V.
In the iterative procedure, the number of cells in the string
E (t ) = i(t )RESR + V (t ) is incremented by a single cell and a trajectory of the
(12) terminal voltage is determined. It is assumed that the voltage
across the individual cells will not exceed the rated cell
V (t )i(t ) = p voltage. If the final terminal voltage falls below the target of
(13)
62.5V then the number of cells is increased. This increases
the initial voltage and also decreases the overall capacitance
E (t ) = nEcell (t )
(14) of the string. The number of cells is increased until the final
voltage of the terminal trajectory is at least equal to the
Ccell required target. Since the number of cells is now known it
C = can be used to determine the current, instantaneous power
n (15)
and efficiency waveforms etc, as shown in Figures 5-7. In
where,
this example the umber of cells has been found to be at least
E(t) is the internal capacitor voltage,
74 and the mass of the string of the capacitors will be about
V(t) is the terminal voltage, 37.44kg.
I(t) is the current,
p is the constant power,
Cell Weight ESR Specific Specific Cell Time
RESR is the equivalent series resistance.
Capacitance Per cell (mΩ) Energy Power Voltage Constant
(F) (kg) (Wh/kg) (W/kg) (V) (τ=RC)
The solution to the set of equations is
2700 0.52 0.6 2.49 8929 2.5 1.6

dE (t ) −CE (t ) ± C 2E (t )2 − 4RESRC 2 p
= Table 2. Specifications for the ultracapacitor used in the example.
dt 2RC 2 . (16)

An iterative numerical method is needed in order to find

the number of capacitors to deliver the power in the required
time. The numerical method requires an initial value for the
voltage E(t) and the iterative procedure needs a final target
minimum voltage. The maximum current is chosen based on
the rating of the capacitor. The terminal voltage, V(t), will be
a minimum when the current is a maximum. By adding the
voltage drop across the equivalent series resistance, the final
value of E(t) can be found. This final value is then used as
the target for the numerical method employed. The Figure 5. The terminal and internal voltage of the ultracapacitor
magnitude of the initial voltage is iterated by the number of string.

1602
 Jeol Ltd, Japan. These capacitors are not yet available
commercially, however prototype specifications have been
made available. The specifications for the ultracapacitors are
shown in Table 3.
The assumption made in the calculations is that there is no
limit on the current conducting capability of the
ultracapacitors. Although this is not practical, knowing that
all capacitors have a practical limit, it gives an insight into
the sizing if manufacturers are able to design them with
higher current ratings. The current limit for large
Figure 6. The capacitor current. ultracapacitor cells is about 400A-600A.

Jeol 45000F

Maxwell 2700F

Ness 2600F
Figure 7. The total instantaneous power delivered by the
ultracapacitors including the losses in the ESR.

Montena 1800F
IV. THE EFFECT OF ULTRACAPACITOR SIZING ON Panasonic 1200F Jeol 2800F
POWER ELECTRONICS.

In the previous section we illustrated how the number of

capacitors, and consequently the mass, can be determined
using an iterative numerical method. This reduction in Figure 8. The mass of the capacitors versus the maximum allowable
capacitor mass has implications on the rating of the power current.
electronics used to interface between the capacitors and the
constant power application. The power electronic converter
needs to be rated at the maximum voltage as well as the
maximum current as indicated by (18).

Pdc − dc = Vmax I max (18)

where, Panasonic 1200F

Pdc-dc is the peak power rating of the power electronics, Montena 1800F
Vmax is the peak capacitor voltage,
Imax is the peak capacitor current. Maxwell 2700F
Ness 2600F
In the case of the previous example the peak power rating
of the converter will be 72kW. It is well documented that the
current power density of automotive electronics is 5kW/kg
Jeol 45000F
[4]. The mass of the power electronics can therefore be Jeol 2800F

estimated as 14.4kg. The total mass of the system will

therefore be approximately 51.84kg. This value does not
include the mass of the cabling and the auxiliary cell voltage
balancing circuitry.
In the next illustration we will examine how the choice of
Figure 9. The rating of the power electronics versus the maximum
capacitor and maximum operating current affects the sizing allowable current.
of the overall system. The commercially available capacitors
that will be analyzed are from Maxwell, Panasonic, Montena
and Ness. The latest technology in ultracapacitors are from

1603
 Weight ESR Specific Specific Cell Time rated current of the ultracapacitor. However, the maximum
Per cell (mΩ) Energy Power Voltage Constant
current affects the sizing of the power electronics and when
(kg) (Wh/kg) (W/kg) (V) (τ=RC)
the masses of both the ultracapacitor string and power
Ness 2600F 0.65 0.25 2.31 7284 3 0.65 electronics are combined, the minimum system mass can
Panasonic1200F 0.34 1 1.3 6618 3 1.2 occur at currents less than the rated ultracapacitor current.
Maxwell 2700F 0.52 0.6 2.5 8929 2.5 1.6 It is imperative to minimize the mass of any energy storage
Montena 1800F 0.4 0.6 2.49 7812 2.7 1.8
system in hybrid electric vehicles. The results have shown
Jeol 2800F 0.21 2.1 27 8000 3.8 6
that there are reductions in mass when comparing the newer
Jeol 45000F 1.5 4.4 60 540 3.8 200
Jeol ultracapacitors to the older ultracapacitors.
Table 3. Specifications for the ultracapacitors. The rating of the power electronic interface is influenced
by the choice of operating current and voltage of the
ultracapacitors. However, using different topologies for the
power electronics could reduce the power rating.

Panasonic 1200F REFERENCES

Montena 1800F [1] Burke, A., “Ultracapacitors: why, how, and where is the technology”,
Journal of power sources, vol. 91, pp. 37-50, 2000.
[2] Burke, A.F. and Miller, M., “ Characteristics of Advanced Carbon-
Maxwell 2700F
based Ultracapacitors”, Proceedings of the 10th International Seminar
Ness 2600F on Double-layer Capacitors and Similar Energy Storage Devices,
Deerfield Beach, Florida, December 2000
[3] Burke, A.F., "Prospects for ultracapacitors in electric and hybrid
vehicles," 11th Annual Battery Conference on Applications and
Jeol 45000F Jeol 2800F Advances, Long Beach, CA, USA, 1996
[4] Miller J, “Propulsion Systems for Hybr`id Electric Vehicles”, IEE
Power and Energy Series, 2004.
[5] Youngho K, “Ultracapacitor Technology Powers Electronic Circuits”,
Power Electronics Technology, October 2003.
[6] Maxwell Technologies, “How To Determine The Appropriate Size
Figure 10. The mass of the power electronics versus the maximum Ultracapacitor For Your Application.”
allowable current. [7] Schneuwly, M. Bärtchi, V. Hermann, G. Sartorelli, R. Gallay, R.
Koetz, “Boostcap Double-Layer Capacitors for Peak Power
Automotive Applications,” Procedings of the 2nd AABC Conference,
Las Vegas (USA), 2002.
[8] Michel, A. Schwake, B. Staib, “Recent Progress in Ultracaps for
Maxwell 2700F Automotive Applications,” Procedings of the 2nd AABC Conference,
Las Vegas (USA), 2002.
[9] Von Jouanne, A., Enjeti, P.N., & Banerjee, B., “Assessment of
Panasonic 1200F ridethrough alternatives for adjustable-speed drives,” IEEE
transactions on Industry Applications, vol. 35, pp. 908-916, 1999.
[10] Jordan, B.A., & Spyker, R.L., "Integrated capacitor and converter
package," 15th Annual IEEE Applied Power Electronics Conference
and Exposition APEC 2000, New Orleans, LA, USA, 2000.

Jeol 45000F
Jeol 2800F Ness 2600F
Montena 1800F

Figure 11. The mass of the system versus the maximum allowable
current.

V. CONCLUSIONS

It has been shown that the overall mass of the

ultracapacitor and power electronics can be minimized by an
appropriate choice of the maximum operating capacitor
current. The minimum amount of capacitors and therefore the
minimum mass of the ultracapacitor string can be achieved
when specifying the maximum operating current to be the

1604