IEEE TRANSACTIONS ON POWER SYSTEMS, VOL. 22, NO.

3, AUGUST 2007 1259

Optimizing a Battery Energy Storage

System for Primary Frequency Control
Alexandre Oudalov, Daniel Chartouni, and Christian Ohler, Member, IEEE

Abstract—This paper presents a method for the dimensioning

of a battery energy storage system (BESS) to provide a primary
frequency reserve. Numerical simulations based on historic fre-
quency measurements are used to determine the minimum possible
capacity, i.e., the lowest possible cost, which fulfills the technical
requirements of the grid code. We implement a novel control al-
gorithm with adjustable state of charge limits and the application
of emergency resistors. At current European market prices, an op-
timized lead-acid BESS can be a profitable utility solution for the
primary frequency control.
Index Terms—Ancillary service market, battery energy storage,
net present value, primary frequency control.

I. INTRODUCTION
Fig. 1. Principal diagram of the BESS.

P OWER systems currently undergo considerable change

in operating requirements—mainly as a result of dereg-
ulation and due to an increasing amount of discontinuous
minimizing the BESS capacity and as a result for minimizing
the BESS installation cost. The method is based on historic fre-
distributed generation. Additionally, continuing load growth quency data, and the BESS capacity is minimized with a novel
and increasing regional power transfer in interconnected net- control algorithm that has adjustable state of charge limits.
works lead to stressed and less secure power system operation. Emergency resistors are an essential component of an op-
This has triggered interest both from transmission system op- timal BESS for primary frequency control. They dissipate en-
erators and power utilities in large-scale battery energy storage ergy during rare events when an extreme over-frequency ex-
systems. While battery energy storage technologies can cover a cursion occurs while the BESS happens to be fully charged: it
wide spectrum of applications, ranging from short time power would be uneconomic to size the batteries for such a rare event
quality support to hours-long energy management, the supply of extreme over-frequency excursion. Because there is a con-
of primary control reserve has been identified as the application flict between the objective to minimize the BESS cost and the
with the highest value for the owner of the battery energy objective to minimize the use of the emergency resistors, the
storage system (BESS) [1]. The profitability of this application latter condition enters as a side constraint (sometimes called a
was established by comparing frequency control reserve prices budget constraint) in the optimization procedure.
on ancillary service markets with realistic installation and The outline of this paper is as follows: Section II gives a
maintenance costs of BESS units. brief overview of commercially available BESS technologies;
Previous research and practical installations [2]–[8] have Section III describes principles of primary frequency control in
technically shown that a BESS can provide frequency regula- Europe; Section IV explains the economic valuation of a BESS
tion indeed. The BESS unit absorbs energy when the system unit for such application. Section V presents the BESS capacity
frequency is above a nominal value and discharges this energy minimization and corresponding BESS operating strategy.
back into the grid when the frequency is below the nominal
value. The main driving forces for these BESS prototypes,
however, were either special system reliability requirements II. BATTERY ENERGY STORAGE SYSTEM
or the testing of new technology. The batteries were usually All commercially available BESS have a similar system
over-dimensioned, so that the total BESS cost was too high to design: batteries are connected to a power conversion system
allow a monetary payback for the installations. In contrast to (PCS) that converts a variable dc voltage of the battery to a
this, we treat here BESS units that are commercially viable at three-phase ac voltage of the utility (see Fig. 1). Four types of
current market conditions. This paper provides a method for batteries have been considered in this study (see Table I).
The cost of BESS units per power unit is a strong function
Manuscript received January 17, 2007. Paper no. TPWRS-00029-2007. of their (energy) capacity, i.e., the maximum discharge time. To
The authors are with ABB Switzerland Ltd., Corporate Research Center, calculate the total BESS cost (over the complete lifetime of 20
CH-5405 Daettwil-Baden, Switzerland (e-mail: [email protected].
com). years), we add the cost of PCS to the net present value (NPV)
Digital Object Identifier 10.1109/TPWRS.2007.901459 of costs of each BESS type, including needed cell replacements.
0885-8950/$25.00 © 2007 IEEE
1260 IEEE TRANSACTIONS ON POWER SYSTEMS, VOL. 22, NO. 3, AUGUST 2007

TABLE I
CHARACTERISTICS OF DIFFERENT BATTERIES

Fig. 3. Deployment of frequency control reserves in UCTE.

TABLE II
REQUIREMENTS FOR PRIMARY FREQUENCY CONTROL RESERVE IN UCTE

Fig. 2. NPV of cost for different types of 10-MW BESS and PHS versus vari-
able capacity and NPV of profit (dashed horizontal line) obtained by selling
primary reserve in Germany.

Fig. 2 shows the NPV of cost for 10-MW BESS units as a func- system operators (TSOs) members of the Union for the Co-or-
tion of BESS capacity. dination of Transmission of Electricity (UCTE) (synchronous
The NPV of cost for a pumped hydro storage (PHS) plant is interconnection of central European countries) to achieve the
plotted on the same diagram for comparison. PHS is the unbeat- required generation-load balance (see Fig. 3).
able storage solution for a discharge time higher than 0.75 h in A deviation of system frequency will cause primary con-
comparison to any BESS technology; however, its application is trollers of all generators subject to primary control to respond
very limited due to geographical and environmental constraints. within a few seconds to stop frequency drop/rise. Requirements
Among batteries, lead-acid BESS is the most economic solution for primary reserve in UCTE are shown in Table II [13]. Since
for a discharge time up to 1.25 h. the early days of UCTE, the primary frequency control was a
Fig. 2 anticipates results for the market price for the primary mandatory service organized through the vertically integrated
reserve service and results for the capacity minimization to be electrical utilities making all large thermal and hydro units
discussed below. ( MW) available for the provision of primary reserve.
During the last decade, markets for ancillary services have
III. PRIMARY FREQUENCY CONTROL RESERVE been established in many European countries. Under free
The electric power system is unique in that power production market conditions, the control philosophy is still respected, but
and consumption must be matched instantaneously and contin- the TSOs do not exercise direct authority over the power plants.
uously. Disturbances in this balance cause a deviation of system Other entities such as loads, distributed generation, and energy
frequency from a set-point value and reduce the quality of power storage have the right, not an obligation, to offer frequency
supply. Therefore, each power system operator is obliged to control reserves as long as they fulfill technical and commer-
maintain a sufficient amount of active power in reserve to com- cial requirements set by the TSO. Currently, primary reserve
pensate for the worst credible contingency (loss of the largest markets are local; cross border reserve markets, however, are
generation or transmission facility). being discussed and will be established in the near future [14].

A. Frequency Control Reserves in UCTE B. Primary Frequency Control Market Mechanism

Three types of reserves performed in different successive In case of primary frequency control, the TSO pays for the
steps are maintained to help Central European transmission mere availability of control power. There is no utilization pay-
OUDALOV et al.: OPTIMIZING A BATTERY ENERGY STORAGE SYSTEM FOR PRIMARY FREQUENCY CONTROL 1261

years in Germany is slightly less than 14 million (Euro). Com-

paring this figure to the total cost over 20 years, of the
battery technologies, we note that the 10-MW lead acid BESS
is a profitable solution if—and only if—the BESS capacity can
stay smaller than 9 MWh.
The owner of the BESS unit would attempt to minimize the
BESS capacity in order to maximize his profit. It is the main
object of this paper, and in particular the next section, to cal-
culate the minimum size needed—such that the BESS is never
completely discharged and 100% of the time able to provide the
contracted primary reserve power.

V. BESS DIMENSIONING
Fig. 4. Prices for primary frequency control reserve in Germany in 2001-2007. To determine the minimal BESS capacity, we have modeled
the actual BESS operation and adapted the operating algorithm
to measured frequency data of the UCTE grid. The minimal con-
ment proportional to the actual amount of energy supplied or ceivable BESS capacity for the contracted primary reserve
consumed. All suppliers are paid a fixed price for the whole is limited to 0.25 hours, since the BESS has to furnish pri-
tendering period per kW of “standby” reserve. Thus, the reserve mary reserve for at least 15 min (see Table II). The realistic min-
availability payment can be calculated as a product of contracted imum will be higher because primary control events can follow
reserve power in kW, the primary reserve price (Euro/kW/pe- each other at short intervals so that there is not enough time to
riod), and the tendering period (1). The primary reserve price is recharge the BESS. For this reason, the recharge strategy has
based on the final accepted bid submitted to the TSO as follows: a strong influence on the optimization result. Another issue is
the use of emergency resistors. While they help to reduce the
Payment Reserve Price Period (1) needed capacity of the BESS unit by dissipating energy during
rare events when an extreme frequency excursion occurs while
The German market for primary frequency reserve has been es-
the BESS happens to be fully charged, we propose to use them
tablished in 2001 and has the longest historic records in UCTE.
as infrequently as possible. This goal enters as a side condition
Awarding of contracts for the provision of primary reserve is
into the optimization. It would be neither an environmental nor
realized in a nondiscriminatory manner on the basis of public
an economic optimum to avoid emergency resistors completely.
tenders. Fig. 4 illustrates the results of tenders in four German
TSOs starting from 2001. The current average price for primary A. UCTE Frequency Profile
reserve is about 60 Euro/kW/six month. Thus, the annual pay-
The measured UCTE frequency for several months of 2005
ment for keeping a primary reserve of MW in Germany is
has been obtained from the Swiss coordinator—ETRANS. Mea-
120 k Euro.
surements have 1 Hz of sampling with less than 1 mHz of accu-
racy. Fig. 5 shows the measured UCTE frequency in April 2005.
IV. ECONOMIC VIABILITY OF BESS APPLICATION The UCTE rules [13] specify the nominal frequency 50 Hz
FOR PRIMARY FREQUENCY CONTROL RESERVE
and noncritical frequency window of mHz. This window
To determine the economic viability of BESS application allows avoiding a calling up of primary control at near nominal
for primary frequency control, the total revenue from selling frequency.
reserve must be compared with capital, operating, The main observation from this data is that the frequency
and maintenance (O&M) costs over the BESS life cycle quality in UCTE is maintained very well. Most of the time
(70%), the frequency in UCTE stays within the noncritical
window of mHz. There are only few frequency excursions
Profit (2) outside of mHz per month. A frequency deviation of
Revenue mHz was never reached throughout the year 2005. High
(3) and low frequency deviations are symmetrical over the long
period (month).
Cost In the short term, however, there are many deviations, and
(4) they are not necessarily balanced. They can occur at random
points in time with random amplitude and random repetition
In these equations, is the BESS life cycle and the dis- rates. As the frequency deviates to values higher than 50.020
count rate (here years, %). Revenue and Cost Hz, the primary frequency controller has to absorb power; as the
are annual values. The revenues depend on the primary reserve frequency deviates to values below 49.980 Hz, it has to supply
prices and are subject to uncertainty. In this work, we use rev- power (see the left-hand side of Fig. 6).
enues of 120 Euro/kW/year. Looking back to Fig. 2 provides a The power-frequency (p-f) characteristic of the primary
first impression on the potential benefit of the BESS application. frequency controller is linear outside the noncritical frequency
from selling MW of primary reserve during 20 window (see then right-hand side of Fig. 6). It is defined in the
1262 IEEE TRANSACTIONS ON POWER SYSTEMS, VOL. 22, NO. 3, AUGUST 2007

Fig. 5. Measured frequency in UCTE, April 2005. From a frequency quality

perspective, April 2005 was an average month.
Fig. 7. Primary reserve power curve, April 2005. Power supplied to the grid is
below zero (BESS in discharge mode). Power absorbed from the grid is above
zero (BESS in charge mode).

TABLE III
RESULTS OF STATISTIC ANALYSIS OF PRIMARY RESERVE POWER CURVE

Fig. 6. Power-frequency (p-f) characteristic of the BESS primary frequency

controller in accordance to the UCTE requirements [13].

grid code [13], which states that the full primary reserve
has to be activated when the frequency deviation reaches
mHz. As an example, we choose a contracted primary
reserve Pn (nominal power of the BESS) of 2 MW, which is
0.06% of the total UCTE primary reserve. The BESS follows
the frequency profile; however, its influence on the frequency The curve named “No recharge” in Fig. 8 shows the change
is neglected. in the BESS state of charge (SoC) upon furnishing primary fre-
The p-f characteristic permits the transformation of the mea- quency control reserve in April 2005. Due to the limited BESS
sured frequency deviations into a required output/input power efficiency, the overall SoC trend leads to a continuous discharge
for every second (see Fig. 7). The duration, the maximum power, of the BESS in the long run. In this case, the minimum required
and the energy of each discharging/charging pulse have been an- BESS capacity for one month (April 2005) is 1.62 hours, a
alyzed in Table III. large and costly BESS capacity. This is of course not a prac-
tical operating strategy; the BESS must rather be recharged at
B. Simulation of BESS Supplying Primary Reserve in UCTE certain time periods in order to maintain a reasonable state of
charge during the course of operation.
1) BESS With No Additional Charging: We start the analysis 2) BESS With Additional Charging: In a second stage, we
with an unrealistic case for illustration purposes only: a BESS appended the operating rules in the following way:
with an efficiency of 70% that is charged and discharged only in • Recharge the battery up to 100% when the frequency is
function of the frequency variation with no additional charging. inside the noncritical window .
At the beginning of the month, the BESS is 100% charged, and • Use a small recharge power (a few percent of ).
then, it supplies/absorbs power depending on the system fre- • If the battery is full (100%), absorb power with auxiliary
quency variation (see Fig. 7). resistors.
With the given p-f characteristic, the operating rules are very The required minimum BESS capacity is drastically reduced
simple: with an increase in BESS recharge power. Fig. 8 shows the state
• Discharge the battery when mHz. of charge of the BESS with efficiency 70% as a function of
• Charge the battery when mHz. time and recharge power 1%, 3%, and 5% of . The higher
OUDALOV et al.: OPTIMIZING A BATTERY ENERGY STORAGE SYSTEM FOR PRIMARY FREQUENCY CONTROL 1263

enables the BESS to absorb more power if needed and reduces

the use of resistors. Additionally, there is a procedure incor-
porated into the control algorithm that allows selling relatively
small amounts of excessive energy on the intra-day market. The
main goal of the selling procedure is not to gain a profit but again
to keep the SoC parameter of the BESS between the two limits
and .
The complete set of operating rules is the following:
• Discharge the battery when mHz.
• Charge the battery when mHz.
• Discharge the battery (sell energy on the market) when
actual (maximum discharge power %
of ).
• Recharge the battery when and
. (maximum recharge power % of ).
Fig. 8. SoC of the BESS with efficiency 70%, no recharge, and recharge with • Dissipate energy in emergency resistors when
1%-5% of BESS nominal power, April 2005.
mHz and %.
• Idle the battery when and
.
Operating rules with SoC limits allow a considerable decrease
in utilizing resistors to only a few times per month compared
to typically 10 000 operations (pulses) of the BESS during that
month, hence justifying the term “emergency resistors.”

C. Optimizing BESS Capacity

The main objective is to maximize the profit for the potential
BESS owner acting at the ancillary service market. Because
the main cost driver is the battery capacity, the optimization
Fig. 9. Impact of recharge power on the BESS design and operation.
will be essentially equivalent to a minimization of the battery
capacity.
On the other hand, the use of emergency resistors is usually
the recharge power, the higher the deepest state of charge and considered negative because it means wasting useful energy.
the smaller the needed battery capacity. This is a requirement that is opposed to our main optimization
There are, however, also disadvantages with a high recharge goal and that constitutes conceptual difficulties. We let it enter
power. With the high recharge power, the BESS is fully charged as a side condition. In this way, it is possible to express any
most of the time and is not able to absorb power from the grid possible preference in that regard. The extremes may range
in the case of a frequency rise above the noncritical window. from putting no value to the resistively dissipated energy other
We use auxiliary resistors to dissipate power under these cir- than its purchase price to putting a very high penalty factor
cumstances. This costs money. Energy has to be bought on the (e.g., 100) on this price to squeeze out the resistor use almost
market that could be obtained for free during the high frequency completely. In our opinion, it is neither an environmental nor an
periods. economic optimum, to avoid emergency resistors completely.
Fig. 9 puts together two plots that demonstrate the impact of The annual values Revenue and Cost in (3) and (4) may be
the recharge power percentage on BESS design and operation. expressed as (5) and (6), respectively
The black line shows how the BESS capacity depends on the
recharge percentage. The BESS capacity is the minimum value Revenue Payment (5)
in Fig. 8 plus 0.25 hours to ensure a further maximum dis-
charge event (15 min full nominal power) at the lowest statisti- Payment is the primary reserve availability payment (1). is
cally occurring level of charge. the revenue from selling energy on intra-day market. Sold en-
The grey line in Fig. 9 shows how the total energy losses in ergy is the additional source of profit for the owner of the BESS,
the auxiliary resistors during one month depend on the recharge although its monthly amount of energy will rarely exceed 0.1
percentage. hours
3) BESS With Additional Charging and SoC Max/Min Limits:
If the target level for the state of charge is chosen slightly below Cost (6)
100%, we keep some charging reserve, and most of the events
can be avoided where energy would be dissipated in the resis- where is the BESS installation and O&M cost. is the
tors. Based on this idea, we have improved the BESS operating cost of recharge energy bought from the grid.
strategy. The goal is now not to keep the battery 100% charged All parameters and variables used in the optimization are
but to keep it in a range between two defined SoC levels. This grouped in Table IV.
1264 IEEE TRANSACTIONS ON POWER SYSTEMS, VOL. 22, NO. 3, AUGUST 2007

TABLE IV
BESS PARAMETERS AND VARIABLES FOR OPTIMIZATION

Fig. 10. Contour plot of the required battery capacity in P 3 h as a function of

the variables sell power and the upper level SoC . The lines that are schemat-
ically shown in addition correspond to resistor losses of 5%, 10%, and 20% of
the total energy uptake into the BESS. Frequency data: April 2005.

TABLE V
RESULTS OF BESS CAPACITY OPTIMIZATION
The objective function is to maximize the profit (2)

Profit Revenue - Cost

Payment

(7)

The method provides the optimal BESS capacity and can be

divided in two steps.
At the first step, the NPV of profit is calculated for each set
of combinations of the variables in the parameter range (see
Table IV). Simultaneously, for each set of combinations, we ad-
ditionally calculate the battery capacity and resistor losses.
Fig. 10 shows a contour plot of the required battery capacity
in as a function of the variables sell power and the upper
level . For example, with a sell power of % and
a level of 0.04 , the required battery capacity is
0.625 . In this example the other variables (see Table IV)
are fixed to recharge power % and
.
It can be seen from Fig. 10 that the smaller the parameters
sell power and —in other words the smaller the amount
of sold energy—the smaller the required battery capacity. This
Fig. 11. Monthly SoC variation of the 2-MW BESS with adjusted SoC
is good for the total BESS cost because the battery capacity is max/min limits. Frequency data: April 2005.
the main cost diver.
We indicate in the same plot contours of a fixed amount of
energy dissipated in the emergency resistors. It is clear that the The corresponding optimal BESS capacity is 0.62 h. Other
smaller the parameters sell power and the higher , the variables and outputs are listed in Table V. The corresponding
higher the resistor losses. BESS state of charge profile is depicted in Fig. 11.
The second step is the search for the optimal BESS capacity Fig. 12 illustrates the BESS operation algorithm for a period
taking into account the energy losses in the resistors. Having no of 32 h during April 2005. At the beginning of time interval,
mathematical formulation for penalty factor of using resistors, the BESS SoC is at Pn*h within the interval
we restricted the resistor losses to maximum 10% of the total . Next, there is a period with prevailing positive
energy uptake into the BESS. frequency deviations so that the SoC moves above the
OUDALOV et al.: OPTIMIZING A BATTERY ENERGY STORAGE SYSTEM FOR PRIMARY FREQUENCY CONTROL 1265

Fig. 13. BESS multi-strings layout and operating principle.

Fig. 12. Provision of primary reserve by the BESS in April 2005.

and a certain portion of energy is sold on the intra-day market.

This energy is delivered as constant power (“Sell power” in
Table IV) during one hour. It has to be scheduled at least 30
min in advance.
During this period, the frequency remains high, and the SoC
reaches 100%. Therefore, excessive energy is dissipated in
emergency resistors. Shortly after that period, the scheduled
power delivery starts, and it brings the SoC back into the SoC
target operating range. Then, there is a long interval with the
prevailing low system frequency. The SoC progressively goes
down and passes the limit after some hours. At this
moment, the BESS control system issues a charging command
with an appropriate value (3% of ) during periods when the
frequency stays inside a noncritical window .
Fig. 14. Average daily frequency variation during April 2005.
Toward the end of the period, the system frequency is high, so
that the SoC comes back into the operating range by primary
control action itself.
Fig. 14 represents the average daily frequency variation
D. Multi-String Operating Algorithm during April 2005. It can be observed that the frequency varies
substantially during a transition from one hour to another, due
The well-known operating rules for multi-string BESS units to a change in the electricity production schedule. This infor-
can be applied to our control algorithm in a straightforward mation can be used for a dynamic adjustment of SoC min/max
manner (see Fig. 13). The BESS monitoring and control system limits just before the change of the hour.
uses the frequency measurement and the SoC level as input As another example, the can be decreased during
signals. The BESS is composed by a large number of battery the night hours because the system frequency is expected to be
cells grouped into parallel strings. Each string is connected to rather high (see Fig. 14). The values of recharge and sell power
a common ac bus via an individual power conversion system. can also be adjusted dynamically in function of market condi-
Cells operate in two symmetric groups; one is used for the down- tions.
ward frequency regulation and another one for the upward fre-
quency regulation. The two groups shift their roles at convenient
points in time. The consistent charging direction for each cell re- VI. CONCLUSION
duces the battery aging and prolongs the battery lifetime. The supply of a sufficient frequency control reserve is impor-
tant to balance power system load and generation at any instant
E. Dynamically Adjustable SoC-Limits for a secure and high-quality power supply. BESS are able to
The BESS operating strategy might be further improved by satisfy the technical requirements for primary frequency con-
introducing dynamically adjustable SoC limits. Rules for the trol.
adjustment of SoC min/max limits can be derived by the statistic This paper provides a method for the dimensioning of a BESS
analysis of frequency measurements. Here we see trends in the to provide primary reserve based on historic frequency measure-
daily, weekly, and seasonal frequency variations. ments. A lead-acid BESS can be a profitable utility solution at
1266 IEEE TRANSACTIONS ON POWER SYSTEMS, VOL. 22, NO. 3, AUGUST 2007

current battery system costs and current market prices for the [10] T. De Vries, J. McDowall, N. Umbricht, and G. Linhofer, “A solution
provision of primary reserve capacity. for stability,” Power Eng. Int., vol. 11, no. 10, pp. 57–61, 2003.
[11] Y. Tachibana, “A new power supply system using NaS battery,” in
With rare use of emergency resistors, the optimum capacity is Proc. 17th WEC Congr., 1998.
0.62 h multiplied by the nominal power rating. For a cost-effec- [12] K. Emura, “Recent Progress in VRB battery,” in Proc. EESAT Conf,
tive sizing, it is essential to use adjustable maximum and min- 2003.
[13] UCTE Operation Handbook, 2004. [Online]. Available: http://www.
imum state of charge limits, to recharge at moderate rate while ucte.org/ohb/cur_status.asp.
the system frequency is within the noncritical window, and to [14] UCTE technical document, Geographical distribution of reserves,
sell some power to the intra-day market if the state of charge is 2006. [Online]. Available: http://www.ucte.org/publications/li-
brary/e_default_2006.asp.
on the high side. An economically optimum BESS for primary
frequency control includes emergency resistors to dissipate en- Alexandre Oudalov received the Ph.D. degree
ergy during rare events when an extreme over-frequency excur- in electrical engineering in 2003 from the Swiss
sion occurs while the BESS is fully charged. Federal Institute of Technology of Lausanne (EPFL),
Lausanne, Switzerland.
ACKNOWLEDGMENT Since 2004, he has been a Research Engineer in
the Information Technologies Department at ABB
The authors would like to thank G. Linhofer (ABB Switzer- Switzerland, Corporate Research, Daettwil-Baden,
land) and W. Sattinger (ETRANS) for their contributions. Switzerland. His research interests include power
system control, distributed generation, energy
storage, and distribution automation.
REFERENCES
[1] A. Oudalov, D. Chartouni, C. Ohler, and G. Linhofer, “Value analysis
of battery energy storage applications in power systems,” in Proc. 2nd
IEEE Power Eng. Soc. Power Systems Conf. Expo., Atlanta, GA, 2006.
6
[2] R. Saupe, “The power conditioning system for the 8:5=17 MW en- Daniel Chartouni received the Ph.D. degree in solid
ergy storage plant of BEWAG,” in Proc. 3rd Int. Conf. Power Elec- state physics from the University of Fribourg, Fri-
tronics and Variable-Speed Drives, 1988, pp. 218–220. bourg, Switzerland.
[3] J. T. Alt, M. D. Anderson, and R. G. Jungst, “Assessment of utility side He joined ABB Switzerland, Corporate Research,
cost savings from battery energy storage,” IEEE Trans. Power Syst., Daettwil-Baden, Switzerland, in 2000 and is a Project
vol. 13, no. 3, pp. 1112–1120, Aug. 1997. Leader in the Electro-technologies Department. His
[4] D. Kottick, M. Blau, and D. Edelstein, “Battery energy storage for fre- areas of interest are devices for energy storage and
quency regulation in an island power system,” IEEE Trans. Energy thermal management of power devices.
Convers., vol. 8, no. 3, pp. 455–458, Sep. 1993.
[5] J. D. Boyes, M. F. De Anda, and W. Torres, “Lessons learned from the
Puerto Rico battery energy storage system,” presented at the 6th Int.
Conf. Batteries for Utility Energy Storage, Gelsenkirchen, Germany,
1999.
[6] W. H. Von Kleinsmid, “Chino battery: An operations and maintenance
update,” in Proc. 3rd Int. Conf. Batteries for Utility Energy Storage, Christian Ohler (M’06) received the Ph.D. degree in
Kobe, Japan, Mar. 18-22, 1991, pp. 38–42. semiconductor physics from the Technical University
[7] S. K. Aditya and D. Das, “Battery energy storage for load frequency of Aachen, Aachen, Germany.
control of an interconnected power system,” Elect. Power Syst. Res., He joined ABB Switzerland, Corporate Research,
vol. 58, no. 3, pp. 179–185, 2001. Daettwil-Baden, Switzerland, in 1998 and is Group
[8] T. Sasaki, T. Kadoya, and K. Enomoto, “Study on load frequency con- Leader in the Electro-technologies Department. His
trol using redox flow batteries,” IEEE Trans. Power Syst., vol. 19, no. areas of interest are devices for energy storage and
1, pp. 660–667, Feb. 2004. switchgear.
[9] C. D. Parker, “Lead-acid battery energy storage systems for electricity
supply networks,” J. Power Sources, vol. 100, pp. 18–28, 2001.