290 1

The Application of Super Capacitors to relieve

Battery-storage systems in Autonomous
Renewable Energy Systems
Arjan M. van Voorden, Laura M. Ramirez Elizondo, Gerard C. Paap, Senior Member, IEEE,
Jody Verboomen, Member, IEEE, Lou van der Sluis, Senior Member, IEEE

Abstract—The utilization of wind and sun as renewable sources

causes uncontrollable fluctuations in power generation.
Furthermore, the ratio between peak power and average power is
high for systems with a limited number of households. In small
autonomous renewable energy systems (ARES), energy storage is
needed; however, the use of Lead-acid batteries as energy buffers
is problematic, since it is not possible to cover fast power
fluctuations without dramatically reducing the batteries’ life-
time. In this paper Super Capacitors are applied to relieve fast
changes in the battery storage system. Batteries are used to meet
the energy requirements and Super Capacitors are used to meet
the instantaneous power demand. At the end of this paper, a
sizing method is proposed for the Super Capacitor system. Fig. 1. System lay-out of the renewable energy system

Index Terms—Autonomous Renewable Energy Systems, In this system, the generated power is delivered by 120 m2
Batteries, Control Strategies, Super Capacitors solar cells (12 kWp), a wind energy conversion system of
maximum 30 kW, and a combined heat and power set of 5.5
kWe. The maximum power demand in this system is 50 kVA.
The storage system was initially calculated to be a Lead-acid
I. INTRODUCTION battery arrangement of 100 kWh (720 V, 140 Ah).

S ome research and development sectors visualize that power

systems of the coming decades could exist of autonomous,
self supplying energy systems with a high penetration of
The laboratory model is designed to give a realistic
representation of the system. In this laboratory, the wind
turbine, the battery system, and the load demand part are
renewable sources [1,2,3]. The drive towards implementing replaced with a power electronic counterpart, which allows
these innovative energy systems is exerted by the liberalization more flexibility. The control system is programmed in such a
of the power market and the constraints of the greenhouse way that there is no difference in stationary or dynamic
problems, caused by the exhaust of environmental polluting behavior between the laboratory and a real situation. The
gasses. The biggest challenge in these systems is to maintain laboratory model is depicted in Fig. 2.
the energy and power balance, especially when the size of the
system is small.
In 1997, the Electrical Power Systems Laboratory at Delft
University of Technology started a research program to
investigate the behavior of autonomous renewable energy
systems. For this purpose, a renewable energy laboratory,
DENlab, was built to analyze the system integration of
renewable energy sources in a limited power scale of 50 kVA,
the size of a row of 10 houses. The system layout is depicted in
Fig. 1.

This work was partly supported by the Delft University of Technology

Fig. 2. DENlab model
research Center project: SENECU, dealing with the Sustainable Energy:
Extraction, Conversion and Use.
290 2

As shown in Fig. 2, the wind turbine is replaced with a B. Wind energy

motor-generator-set and controlled by a wind input signal from Buildings and other obstacles produce a lot of turbulences in
an anemometer. The battery system is replaced with a bi- the wind and lead to fluctuating magnitudes and directions
directional rectifier that includes the battery characteristics. [8-10]. Because of the fluctuations in the wind speed and due
The overall demand pattern is performed by a back-to-back to the characteristic of wind speed vs. power (the power curve
converter, which returns the demand into the public grid. The between 4 and 12 m/s almost depends on the speed to the
pattern of this load demand is made available from a stochastic power of 3), the power produced by a wind turbine in a built-
computer program. More details about this laboratory can be environment is very unstable. Fig. 3 shows an arbitrary
found in [4, 5]. measured day pattern of the wind power production, recorded
in the DENlab.

II. POWER MISMATCH

The power mismatch between generation and demand must
be stored or delivered. This section shows the power
fluctuations in generation and demand and concludes that the
power mismatch varies to a large extent. The power mismatch
must be met by the storage system, in other words, the battery
storage acts as the system’s slack-node, which means that the
surplus or shortage of power is equalized by the battery
storage. This slack-node approach is valid within the limited
power range of the battery capacity and the inverter.
A. Power demand
The rated power of the laboratory model is 50 kVA, which
is representative for the maximum load of ten households. For
Fig. 3. Wind speed and production
this small system, the ratio between the maximum load and the
average load is high. However, when the number of loads
increases, the coincidence in load behavior leads to a smoother
C. Solar energy
load curve and a smaller ratio between maximum and average
load [6]. Solar cells convert sun-irradiance into electrical power. The
intensity of the light depends mainly on the cloud cover and
the time of the day. The difference in power production
observed peak power in the group between direct and diffuse light is very large. In Fig. 4, two
C=
∑ (power pattern) (1) measured days are depicted. The large power dips are caused
by moving clouds. The power reduction can be very fast, as it
depends on the speed of the clouds. Days with few fast-moving
C = coincidence factor clouds are particularly bad for achieving a stable power
production.
To connect real loads in the DENlab, in order to represent
the power consumption, would be an unnecessary waste of
energy. Therefore, a back-to-back converter, which returns the
energy back into the public grid, is used to represent the total
demand.
A software program based on an inquiry of the demand of
many and various types of households, produces the energy
demand pattern for a specific number of households [7]. In
this inquiry, inhabitants were asked to enumerate the type of
equipment installed at their homes and to estimate the time
interval at which the different devices are used. The computer
program randomizes the large number of acquired data.
Finally, by entering the number of households and, when
applicable, the specific entries to select the type of households
and the number of inhabitants, a weekly load pattern is
generated. Fig. 4. PV power curves of two days in May 2004
290 3

D. Storage power IV. SUPER CAPACITORS

Like stated before, the difference between generated and In literature, the Super Capacitors appear under different
demand power is equal to the power exchange with the storage names: ultracapacitors (UC), electrochemical double layer
system. It can be concluded from section II. A. - C. that the capacitors (EDLC) or Super Capacitors (SC) [11]. A SC is an
fluctuations in the battery power result from the sum of the electrochemical capacitor that has an unusually large amount
individual solar, wind and demand power variations. of energy storage capability relative to its size, when compared
to common capacitors. These components are of particular
interest in automotive applications for hybrid vehicles and as
III. BATTERY CHARACTERISTICS supplementary storage for battery electric vehicles [12]. The
A battery is a widely known component that converts first Super Capacitor based on a double layer mechanism was
electrical energy into chemical energy or vice versa. In Lead- developed in 1957 by General Electronics in a patent using a
acid batteries, the charge process leads to a reform of the porous carbon electrode. It was believed that the energy was
sulfuric acid of the electrolyte as follows, stored in the carbon pores and it exhibited “exceptionally high
capacitance”, although the mechanism was unknown at that
time.
SO4 + 2 H + ⇒ H 2 SO4 (2)
Electrochemical capacitors (EC) store electrical energy in
The net (re)charge reaction on both plates and electrolyte is: the two series capacitors of the electric double layer (EDL),
which are formed between each of the electrodes and the
2 PbSO4 + 2 H 2O ⇒ PbO2 + 2 H 2 SO4 + Pb (3) electrolyte ions. The distance over which the charge separation
occurs is just a few angstroms (10-10 meter). The capacitance
Batteries may be required to supply long-duration energy- and energy density of these devices is thousands of times
type or short-duration power-type loads, depending on their larger than electrolytic capacitors.
application. Any battery type can be sized to supply short- Super Capacitors have several advantages in relation to
duration power loads, but high-power battery designs supply batteries:
them more effectively than others.
The main principle for designing high-power batteries is to
Very high rates of charge and discharge;
consider that these batteries can discharge a high percentage of
Little degradation over hundreds of thousands of
their stored energy in a short period of time. In contrast, lower- cycles;
power batteries can deliver more energy but only if they are
Good reversibility;
discharged during a long period of time. This phenomenon is
Light weight;
described with the Peukert law.
Low toxicity of materials used;

High cycle efficiency (95% or more).
Cp = I Nt (4)
Their disadvantages are:
where: Cpis the nominal capacity [Ah];
I is the discharge current [A];
The amount of energy stored per unit weight is
N is the Peukert constant; considerably lower than that of an electrochemical
t is the time of discharge [h]. battery (3-5 Wh/kg for a UC compared to 30-40
Wh/kg for a battery);
This law expresses the capacity of a battery in terms of the
The voltage varies with the energy stored. To
rate at which it is discharged. As the rate increases, the effectively store and recover energy requires
battery’s capacity decreases. For an ideal battery, the constant sophisticated electronic control and switching
N would equal one; in this case the actual capacity would be equipment.
independent of the current. However, for a Lead-acid battery,
the value of N is typically between 1.1 and 1.2. The energy content of a (Super) Capacitor is proportional to
Besides the reduction in remaining capacity, the reduction the square of the voltage, as shown in the following equation,
of the life-time is significant when the battery system operates
under high fluctuations. Therefore additional measures should 1
be taken in order to apply the battery in an efficient way to Wc = CVc 2 (5)
prolong its life. The best way is to split the energy and power 2
needs into two separate components. Lead-acid batteries could
be used as energy suppliers, whereas Flywheels and Super
Capacitors could be used as power suppliers. In this paper, the
use of Super Capacitors is considered to relieve the fast
changes in the battery storage.
290 4

V. CONFIGURATIONS AND CONTROL STRATEGIES The control is a simple integrator controller, where the time
The Lead-acid battery – Super Capacitor facility can be constant can be adjusted. In this way, the battery supplies the
arranged in different ways. Two basic configurations are the energy (average power), whereas the Super Capacitor supplies
DC-system and the AC-system. They are depicted in Fig. 5 and the power deviation from the mean value. The transfer
Fig. 6. function is experimentally derived.

DC
DC
DC
+
Battery SC
_
AC

Fig. 5. DC-system configuration

DC Fig. 7. Control system for battery current

SC
The DC/AC converter, connected to the Super Capacitor
AC
bank, acts as a Voltage Source Converter (VSC) that dictates
the frequency and AC voltage in the system. This converter is
the primary storage system, which mainly responds to sudden
DC energy unbalances.
+
The DC/AC converter, connected to the battery bank, acts
Battery
_ as a Current Source Converter (CSC). It receives the control
AC input from the Super Capacitor voltage difference and
mitigates the error-signal in order to fulfill the necessary
energy demand of the autonomous system.
Fig. 6. AC-system configuration

VI. SIZING SUPER CAPACITOR SYSTEM

The AC-system provides a better and simpler control of the
energy exchange and has a better performance than the DC- The calculation of the dimension of the Super Capacitor
system and will therefore be used here. system is always a matter of choice between costs aspects and
It should be noted that a DC/AC converter has a limited performance. The higher the capacity of the SC-system, the
voltage range at its DC side for a specific AC voltage. In more constant the battery current will be. Since the cost
standard converters of 400 V, the DC-voltage may vary aspects are ignored in this paper, the following procedure can
between 600 and 740 V. This limits the usable energy content be followed in order to select the size of the Super Capacitor
of the Super Capacitor. Because of the quadratic voltage system for the autonomous energy system application.
dependency, however, most of the energy is available at the
high voltage regions.
Collect the data of several successive days;
To define the properties of the SC-buffer for fast power
Calculate the mean value of the power of the total
fluctuations and the battery-buffer for slower energy changes, storage system using a moving average filter with a
the following control strategy is developed. As reference period of half a minute;
voltage for the SC-buffer, the average voltage of 670 V is
Determine the deviation of the instantaneous power
taken. When balancing power is needed, it will be delivered by from its mean value (output of the moving average
the SC, which causes the SC-buffer to drop. At the same time, filter);
the battery-buffer provides power proportional to the
Calculate the worst period of that day, in other
difference of the actual SC voltage and its reference. words, the highest average power in a certain
When the balancing power is negative i.e. the production is period (absolute value);
higher than the demand, both the SC and the battery will be
Calculate the required capacity, taking into account
charged in the same way. For fast power changes, the SC will the bandwidth of the DC inverter voltage.
take most of the fluctuations while the battery takes the longer
term changes in power. In Fig. 7 the control system is
depicted.
290 5

TABLE I
DATA OF BATTERY AND SUPER CAPACITOR
Size Resistance Voltage
Super Capacitor C = 10 F Rsc= 10 mΩ Usc =670V
Battery E = 100kWh Rbatt= 0.1 Ω Ubatt =720V

The simulation results have shown a 100 s period, where the

maximum energy exchange in the SC is 740.000 Ws. For the
calculation of the capacitor’s dimension, we assume that the
Super Capacitor voltage may vary between 740 and 600 V
(maximum to minimum operating voltage of the inverter).

1 1 1 1 
Fig. 8. Power at the storage energy system E = CVmax 2 − CVmin 2 = C  Vmax 2 − Vmin 2 
2 2 2 2  (6)
In Fig. 8, the power flowing into the storage system is 740.000
C= = 7,88 [F]
depicted. It can be seen that the storage power is fluctuating in 93.800
time (bottom fluctuating curve, dashed). The control system
with a time constant of 10 seconds and a gain of 5 delivers a From this calculation, it follows that the minimum capacity is
smoothed battery current (dotted line). The difference between 7.88 F. For security reasons, a higher value is chosen, for
mean and actual power (upper fluctuating curve) is delivered instance 10 F.
by the Super Capacitor. In time, the average power of the
Super Capacitor is almost zero. By changing the transfer
function, other system responses can be realized. VII. CONCLUSIONS
This paper proposes a Lead-acid battery - Super Capacitor
system as a solution for energy storage in autonomous
renewable energy systems (ARES). It describes and quantifies
the occurrence of power fluctuations as a consequence of the
usage of renewable energy sources in a small system of 10
households. Batteries are not appropriate to follow the sharp
fluctuations, since their life-time is negatively affected.
Therefore, the application of only Lead-acid batteries is
inadequate for ARES of this power scale.
By splitting the total energy demand, the energy and power
requests can be fulfilled by introducing two different types of
storage. Super Capacitors are applied to meet the fast changing
power demand, while the Lead-acid batteries balance the
average energy demand. The integration of the Lead-acid
batteries and the Super Capacitor in the storage system is
explained.
A way to dimension the Super Capacitor system is also
described. In general, the larger the Super Capacitor, the lower
is the impact on the life-time of the batteries. It can be
Fig. 9. Voltages and currents in the Battery - Super Capacitor system concluded that ARES with a storage system as proposed is
technically feasible.
Fig. 9 shows the individual voltages and currents of the
battery and Super Capacitor system. This curves show that the
battery current is smoothed and that the Super Capacitor REFERENCES
current is fluctuating around the zero-line. During periods of [1] CIGRE, Electric Power Systems 2020. 2003, Joint Advisory Group
SC15/E1-JAG 02 TC.2.
positive current, the voltage of the Super Capacitor decreases. [2] Technologies for Tomorrow, IEEE Power Engineering Review, 1998.
The rate of decrease depends on the size of the capacitor 18(9).
system and the actual current. The deviation of the Super [3] SmartGrids, vision and strategy for Europe’s Electricity Networks of
the Future. 2006, European Commission.
Capacitor voltage from the reference value controls the battery
current.
290 6

[4] van Voorden, A.M., Paap, G.C., van der Sluis, L. The set-up of a Jody Verboomen obtained his Master in Industrial
renewable energy laboratory. in IEEE young research conference. Sciences (Electronics) from Groep T Hogeschool in
2002. Leuven. Leuven, Belgium in 2001. He obtained his M.Sc. in
[5] van Voorden, A.M., Paap, G.C., van der Sluis, L. The Use of Batteries Electrical Engineering from the Catholic University
in Autonomous Renewable Energy Systems. in PowerTech 2005. 2005. of Leuven (KUL), Belgium in 2004.
St. Petersburg. He is currently working towards a Ph.D. on the
[6] Willis, H.L., Scott, W.G., Distributed Power Generation. 2000, New application of FACTS and phase shifters in
York: Marcel Dekker Inc. transmission systems in the Electrical Power System
[7] Uitzinger, J., Sepath, simulatie van energievraagpatronen van (EPS) laboratory of the Delft University of
huishoudens, SDE, Editor. 2003, IVAM: Amsterdam. Technology, The Netherlands. His research is
[8] Dutton, A.G., Halliday, J.A., Blanch, M.J., The Feasibility of Building funded by SenterNovem, an agency of the Dutch ministry of Economical
Mounted/Integrated Wind Turbines (BUWTs): Achieving their Affairs.
potential for carbon emission reduction. 2005, CLRC Rutherford
[9] Mertens, S., The energy yield of roof mounted wind turbines. Wind
Engineering, 2003. 27(6): p. 507-518. Lou van der Sluis was born in Geervliet, the
[10] Wit, M.H.d., Wind in de gebouwde omgeving. 2001, Technische Netherlands on July 10, 1950. He obtained his M.Sc.
Universiteit Eindhoven. in electrical engineering from the Delft University of
[11] Smith, T.A., Mars, J.P., Turner, G.A. Using Supercapacitors to Technology in 1974.
Improve Battery Performance. in 33th Power Electronics Specialists He joined the KEMA High Power Laboratory in
Conference. 2002. Cairns, Australia. 1977 as a test engineer and was involved in the
[12] Palma, L., Enjetti, P., Howze, J.W. An Approach to Improve Battery development of a data acquisition system for the
Run-time in Mobile Applications with Supercapacitors. in Power High Power Laboratory, computer calculations of
Electronics Specialists Conference. 2003. Acapulco, Mexico. test circuits and the analysis of test data by digital
computer. In 1990 he became a part-time professor
and since 1992 he has been employed as a full-time professor at the Delft
University of Technology in the Power System Department.
Prof. van der Sluis is a senior member of IEEE and head of the Power
Arjan M. Van Voorden was born in Rotterdam, the System Laboratory of the Delft University of Technology.
Netherlands, on April 11, 1972. He received his
B.Sc. degree in electrical power engineering from the
"Haagse Hogeschool" in June 1995. His thesis
project involved cable load calculations on
asymmetric cable routes in steady state and
emergency situations. This work was done in
cooperation with KEMA, The Netherlands. In April
1998, he received his M.Sc. degree in electrical
power engineering from the Delft University of
Technology. This master thesis project involved autonomous energy supply of
houses.
He is currently with the Power System Laboratory at the Delft University
of Technology, working on his PhD-thesis: "Power Balancing in renewable
energy systems".

Laura M. Ramirez Elizondo was born in San

José, Costa Rica on December 12, 1980. On 2003,
she received her Bachelor degree in Electrical
Engineering at the Universidad de Costa Rica. She
obtained a Bachelor degree in Music with a major
in piano on August 2005 at the same institution.
From January 2004 to July 2005 she worked as
instructor professor at the Department of Power
Systems and Electrical Machines of the
Universidad de Costa Rica. She is currently
working on her thesis as part of her M.Sc. studies
in Electrical Power Systems at Delft University of Technology.

Gerard Chr. Paap was born in Rotterdam, the

Netherlands, on February 2, 1946. He received his
M.Sc. degree from Delft University of Technology
in 1972 and his Ph.D. degree from the Technical
University of Lodz in 1988.
Since 1973 he has been with the Department of
Electrical Engineering at Delft University of
Technology. From 1973 to 1985, he was with the
Division of Electrical Machines and Drives, where
he lectured on the fundamentals and dynamics of
electrical machines; since 1985 he has been with the Power Systems
Laboratory where he is currently Associate Professor.
Dr. Paap's main research interests include power system transients,
dynamics and stability, the large-scale application of renewable energy as well
as the dynamics of electrical machines. He is senior member of IEEE.