J.

Energy Power Sources Journal of Energy

Vol. 1, No. 6, 2014, pp. 287-295 and Power Sources
Received: September 1, 2014, Published: December 30, 2014 www.ethanpublishing.com

Application of Frequency Regulation Control on the

4MW/8MWh Battery Energy Storage System (BESS) in
Jeju Island, Republic of Korea

Manuelito Del Castillo, Jr., Gun Pyo Lim, Yongbeum Yoon, Byunghoon Chang
Korea Electric Power Corporation Research Institute, Daejeon, 305-760, Republic of Korea
Corresponding author: Manuelito Del Castillo, Jr. (14990070@kepco.co.kr)

Abstract: Energy Storage Systems (ESS) have an impressive track record for providing stability and efficient power management in
power system. With recent technological advancements and the decline in costs, the use of battery for ESS is gradually becoming
popular in the power industry. In particular, Lithium-ion Battery (LiB) is being used in various power grids around the world. In
Jocheon, Jeju Island, Republic of Korea, operation and control of grid-connected Battery Energy Storage System (BESS) of
4MW/8MWh capacity has been in demonstration phase since 2013. Several simulation tests were already performed for peak shaving,
renewable energy output smoothing, and most recently, frequency regulation control. On-site tests are continuously performed for
frequency regulation. In previous tests, control algorithm was coded, modified and tested to load into controller and communication
system, and human-machine interface was developed. The BESS Energy storage was connected to grid for frequency regulation using
both simulated and actual frequency variations. The most recent results are presented in this paper showing BESS frequency
regulation with continuous operation duration for three days. It was also observed in the results that BESS frequency regulation has
faster response, less costs and less capacity of energy storage systems which cover for frequency regulation of power plants. Time
reduction and time-delay elements are also investigated.

Keywords: Battery, energy storage system, frequency regulation, grid-connected control system.

1. Introduction frequency regulation on real system, in this case in

Jocheon substation in Jeju Island. Frequency
The developments on Energy Storage System (ESS) fluctuations are caused by small load perturbations
recently focus on the compensation of maximum which continuously disturb the normal operation of
power system load, frequency regulation, and power systems. Therefore, the generation rate must be
renewable energy output smoothing changed until the frequency and tie-line power are
To assess the performance and benefits of ESS, maintained close to their acceptable limits [1]. To
Korea Electric Power Corporation (KEPCO) has an some extent BESS will be able to execute this
on-going demonstration project using Li-ion function instead of the generators. Several researches
Battery-ESS (BESS) with capacity of 4MW/8MWh. [1-3] have proposed several control schemes for
Peak shaving and renewable energy output smoothing frequency regulation function of BESS. This time we
had already been assessed in this project and results implement a control algorithm for the demonstration
are continuously evaluated. This is KEPCO’s first step of frequency regulation on actual power system with
towards the commercialization of ESS with the use of BESS.
Li-ion battery. The frequency control algorithm developed for
In this paper, we concentrate on using BESS for power system simulation programs in Jeju power grid
288 Application of Frequency Regulation Control on the 4MW/8MWh Battery Energy
Storage System (BESS) in Jeju Island, Republic of Korea

uses PSS/E with C# programming language. The Table 1 Frequency operation recorded by KPX.
frequency control used in the substation with BESS in Freq No. of freq. variations Share rate
conjunction with the installation requirements for 59.8 0 0
59.81 26 0
commercial use. Control algorithms were modified to
59.91 3730 0.28
communicate with other already installed control 59.96 635699 51.2
equipment. 60.1 644055 48.09
Since previous results of the demonstration project 60.06 5690 0.42
confirmed the use of BESS to control frequency 60.11 0 0
60.2 0 0
fluctuations for actual power system, we continue in
investigating the effectiveness of BESS frequency
regulation response in longer duration.

2. Frequency Regulation Control Theory and

Methodology
2.1 Domestic Power System Frequency Operating Status

The frequency operation reported by Korea Power

Fig. 1 AGC provided by KPX.
Exchange (KPX) is shown in Table 1. Sampling is
done for every 2 seconds in one month during
December 2013. The provisions in Electricity Act
require maintaining the power system frequency range
of 60 ± 0.2 Hz [4].

2.2 Frequency Control in a Thermal Power Plant

Power plant turbine speed is 3600 [rpm] and varies

depending on the turbine type, normally ranging from
Fig. 2 Frequency variation after power drop.
0.06 to 0.4. Deviation of 0.036 Hz (± 0.018 Hz)
corresponds to 0.06 % of the reference frequency. Fig. Fig. 2 shows the frequency variation with the
1 shows the KPX’s Automatic Generation Control generator power output when the power exchange falls
(AGC) signals when frequency changes continuously to 455 MW from 485 MW with AGC signal. This
considering 482 MW generator output. fluctuation continues until a 5 MW deviation occurs as
Points A, B, and C (uppercase letters) is set to the seen on the AGC signal output after point D.
upper limit when the power system frequency is Frequency is rising after point D despite the fall after
greater than 0.018 Hz-frequency dead band. Points a, b, point d, where you can see frequency adjustment by
and c (lowercase letters) are the points after the upper AGC.
limit frequency points A, B, and C. The power output
2.3 Configuration of BESS Frequency Regulation
is increased to reduce the frequency. If the power
Controller
system’s frequency variation is kept constant AGC
setting maintains the frequency within the tracking Fig. 3 shows the BESS frequency regulation
operation limits. It can be seen that it has a time delay, controller configuration, the power charge and
and then the generator output fluctuation is outside the discharge system (Power Conditioning System (PCS)),
dead band frequency. and connection to the battery.
 Application of Frequency Regulation Control on the 4MW/8MWh Battery Energy 289
Storage System (BESS) in Jeju Island, Republic of Korea

The frequency regulation controller device consists the battery monitoring system. The BESS has eight (8)
of control algorithm computing devices, communication 1-MWh batteries, and two of them are linked to one (1)
apparatus, and a frequency measuring unit. The MW PCS, one set of frequency regulation controller,
frequency controller receives the frequency signals as and corresponding set of communication controllers.
input and is sent to the PCS in order to determine the
2.4 Frequency Regulation Control Method for the BESS
dispatch operation. The PCS system consists of a
control panel, the drive panel, and the input panel. The The frequency regulation control method is
charging and discharging dispatch for the battery will divided into three separate control method categories
be according to the request signal received from the or states for adjusting the frequency of the BESS.
frequency controller. These states are transient control state, steady state,
The actual frequency regulation control system for the exit state.
BESS configuration is shown in Fig. 4. The whole Fig. 5 shows an overview of the frequency
control system consists of a control algorithm regulation control for BESS. The algorithm is coded
computing devices, communication apparatus, the using C# programming language. The algorithm first
frequency measuring device and a set of console for receives input data, such as frequency and battery
driving the frequency regulation controller device. charge status. Then, it calculates a frequency error to
The PCS system is composed of four sets of 1 MW determine the state of operation.
power charge and discharge system. The PCS is in During transient state, signal is sent to the PCS, and
conjunction with one set 1 MW frequency regulation then battery output target value is now set to transient
controller with a set of controls and communication state control mode. While in a non-transient state, the
configuration. The smallest unit of a battery (cell)
constitutes a 1MWh multiple cell battery connected to

Fig. 3 Frequency regulation controller configuration.

Fig. 4 Frequency regulation control system. Fig. 5 Frequency regulation control algorithm.
290 Application of Frequency Regulation Control on the 4MW/8MWh Battery Energy
Storage System (BESS) in Jeju Island, Republic of Korea

battery output target value is set to steady-state control amount of the minimum frequency will be eliminated
mode. After experiencing a transient state, the process at the generators that correspond to the values for
returns to the normal state control mode which passes frequency error. Since it is possible to control the
through the exit state control mode. Steady-state nature of fast energy storage device to store excess
control mode is applied when there is no fault and in energy when the power system based on the frequency
consideration of the battery State of Charge (SOC). It required by the system, constant K was estimated by
does not require a control system to ensure a fast calculating the requirements of the device in
control response of the available capacity of the battery proportion to the drop in the operable region before
energy storage system (to minimize the steady-state). reaching the point of maximum yield point and then to
Eq. (1) represents the control strategy of the steady-state the minimum frequency grid disturbance. There is a
control mode [5]. need to avoid sudden changes in the output when it
 0, f  30mHz  returns to the steady-state control mode, starting from
0%  SOC  40%10%P ,30mHz f  30mHz  the transient period until the system recovers. When
 n
Preq,30mHz f  
  the exit state mode changes to the normal state control
  req
P , f  30mHz 
   (1)
P  40%  SOC  80% 5%Pn ,30mHz f  30mHz  mode, it yields a dynamic or transient control mode
ess
 Preq,30mHz f  
 state of the energy storage device.
Preq, f  30mHz 
   0%  40%,0 
 80%  SOC  100 %  10 % P n ,30mHz  f  30mHz 
 0, f  30mHz    df 
  dt  0,Kf 
  (2)
P  40%  SOC  100% df  0&P  (K f),P

We set the discharge limit and this is decreasing all ess
  af d af
throughout the BESS operation lifetime. The dead dt
 df 
   0&P  (K f),K f 

zone was set to control the frequency shift in order to   dt af d d 
prevent a reduction in battery life due to frequent Eq. (2) represents the dynamic state control mode
charging and discharging operations. Very small the control strategy of the energy storage device, the
amounts of charging and discharging for the BESS as formula [4].
response to the fast-changing frequency fluctuations
may result to decreased efficiency and may negatively 3. Simulation and Demonstration Results
affect the overall facility life [6]. These small amounts Tests using simulated frequency variations were
of charging and discharging are not recognized by the conducted in previous demonstrations of the project
system. to verify the control algorithm before it is used in
In the state of charge between 40 % to 80 %, and the BESS for actual power system frequency regulation.
charge and discharge control in other areas other than In the case of large energy storage devices, a
the dead band, constant K is applied with output proportional constant is determined for the control of
change rate of 4 % of the existing power generation the output value of the BESS in order to correct the
source in accordance with the grid frequency variations. error on the power system frequency. In testing
Control is enabled for discharging during 80 % - 100 % energy storage system devices in different power
state of charge while control for charging is enabled ranges, KPX’s system auxiliary services operating
during 0 % to 40 % state of charge. criteria shall apply. Control is applied only when the
In the above mentioned charge-discharge limits and measured driving speed is out of the dead band or
controls, the output of the BESS is determined by deviated by ± 0.03 Hz from the reference value of the
applying K (domestic power system). Constant integer power system frequency [7]. Separate conditions are
K is eliminated from the generator power system: The simulated whether on steady-state or transient state.
 Application of Frequency Regulation Control on the 4MW/8MWh Battery Energy 291
Storage System (BESS) in Jeju Island, Republic of Korea

Figs. 6-7 show the results of the said simulations for Each battery charging status, 40 % or more, has to be
both steady state control mode and transient state raised to the maximum power. After the transient
control mode. The 102.21 MW/Hz Line Regulation control mode is complete, the control was found to
are subjected to a constant strain of 4 % by the Jeju decrease the battery output at a constant rate by
system simulation testing. switching the control mode to the exit state. After
In Fig. 6, system power generation capacity is 702 completing the simulation test, the frequency
MW, the system load is 699 MW, Jeju’s peak load for regulation control system is connected to the PCS
the year 2013. Battery charging status of # batteries system with all the necessary communication links as
1-4 are 5, 55, 75, and 95 % respectively. shown in Figs. 3-4.
In Fig. 7, power generation is 702 MW, and system Without connection to the BESS, we used a console
load is 699 MW during the state of the frequency controller and Rx3i of GE for frequency regulation
regulation operation of the High-Voltage Direct control performance and install the corresponding
Current (HVDC) power transmission. Battery SOC for control algorithm in the device. Proficy Machine
the four units are 5, 55, 75, and 95 %, by applying a Edition 7.0 was used to control the logic of Rx3i
peak load of Jeju Island in 2013 and were tested by development program. A control algorithm is
dropping 35 MW. developed initially in C# programming language. This
It was confirmed that it is possible to control the must be loaded into this controller device. However,
output of the battery, as shown in the simulation because the inputs that are loadable in the Rx3i device
results in Figs. 6-7. It can be discharged even when the must be in C programming language, the algorithm
battery is partially discharged. In contrast to the steady structure associated must be modified. The new
state, the transient state is determined based on the structure could be expressed in the form of a combined
frequency variation exceeding a continuous span of time. ladder diagram and functional block diagram. It is
therefore necessary to debug the manually modified
automatic control logic [8].
Once the control algorithm is loaded into the
controller similar results are obtained between the
simulation shown in Figs. 6-7 and the simulation
shown in Figs. 8-9. The process proceeds without the
communication associated with PCS. This confirms
that the controller is operating normally in accordance
with the control algorithms. As the coded algorithm is
Fig. 6 Steady state control mode using BESS.
verified and performed properly, we proceeded with
the field test connecting the BESS.
The task was to develop a parallel operation of the
console screen and the control strategy developed in
the controller. Fig. 10 shows the configuration of the
console controller for individual and collective control.
This includes controller tuning function, frequency
fluctuations simulated function, alarm and event
functions, trend function, communication network
Fig. 7 Transient state control mode using BESS. surveillance, starting sequence, automatic control and
292 Applicatio
on of Frequen
ncy Regulation
n Control on the
t 4MW/8MW Wh Battery Energy
Storage Sysstem (BESS) in Jeju Island
d, Republic of Korea

monitoring fuunctions. dispatcch to respond d after signal is sent. The delay in

A phenom menon occurreed when thee scan time is the enntire loop waas found out to be 79 m ms. The
relatively lonng, and indicaated the lineaar portion from m measurred time dellay was veriffied considerring the
the measuredd results. Thee delay occurrred for 79 m ms output target value ofo (180 ~ 100 kW.)
during the steeady state conntrol. Similarlly, there is alsso If thhe frequency variation
v rate is -0.306 [Hzz/sec] or
a communicaation delay beetween the con ntroller and thhe less, thhe system is considered to be
b in transientt state. It
PCS. Duringg charging and discharg ging dispatchh, is show wn in Fig. 111 that it took around 7 mss for the
delays are reequired to reaach the battery y output targeet algorithhm process to
o calculate freequency by meeasuring
value. Currenntly, the systemm has a very short time (mss) voltagee and currentt from the disstribution linee and to
during dispattch includingg the processiing time of thhe assign a battery ou utput target value
v in conjjunction
controller syystem itself, and the communicatio
c on with thhe charging and dischargging dispatchh of the
between conttroller and PC CS. The system m latent periood BESS. The simulateed field test shhown in Figs. 6-7 has
was measuredd as the amouunt of time it takes
t for BESSS an 8.3 ms time in nterval. It toook 10120 mss (10.12
secondds) for contro
olling the BESS output ussing the
target value
v when thhe controller is in the transieent state
of freqquency variatiion. The targeet value is determined
in accoordance with the measuredd battery outpput from
the power quality measurement
m devices which took
279 mss. Therefore that
t took a tottal delay of 100399 ms
(10.4 seconds).
s It to
ook the 400 ms m from the start of
discharrging a batteryy until it reachhed 1000 kW..
Fig. 8 Steadyy state control m
mode using con
ntroller only.
Althhough a time delay
d occurredd in the entiree loop, it
was confirmed that th he control of the
t battery outtput is in
accordaance to the baattery output target
t value. A
Also, the
exit staatus mode con ntrol operates normally in order to
eliminaate the shock in
i the transientt state when sw
witching
to the normal
n or steady state. Thee battery outpuut target
value of
o the transiennt state is a vaalue controlledd by the
proporttional gain as given in Eq. (3)
( [9]:
Fig. 9 Transiient state control mode using controller
c only.. M  Kc E (3)
where M is the baattery outputt demand; Kc is the
controlller proportio
onal constant;; and E is thhe error.
Duringg exit control mode
m operatioon Eqs. (3) andd (4) are
used:
M  Kc E  M r (4)
where Mr is the exit mode controll reduction biaas. In Eq.
(5) andd the initial vaalue of the Mr(M
( r0) are calcuulated:
M r 0  E0  R x E0 (5)
where M0 is the baattery output when transieent state
ends; Rx is decreasiing rate of exxit control moode; and
Fig. 10 Confiiguration of thee console contro
oller. E0 is the deviationn between thhe battery ouutput of
 Application of Frequency Regulation Control on the 4MW/8MWh Battery Energy 293
Storage System (BESS) in Jeju Island, Republic of Korea

transient state and steady state when transient state is 400 ms was spent to reach the output to 1000 kW, and
finished. it took 70 ms to measure the results directly from the
Fig. 11 shows the steady state frequency regulation controller. It was found that it took approximately 130
with some delay time from the energy storage device ms to generate the required battery output after
to the battery output target value that is input to the frequency variation during the transient state. 130 ms
console for operating the grid frequency. contains the required process time for the control logic
Fig. 12 shows the frequency change rate within the and 14 ms for the sweep controller.
transient criteria, between -0.0306 [Hz/sec] and -0.486 With all the necessary tests done, the whole BESS
[Hz/sec], and the frequency change lasted for about system together with the frequency regulation
120 ms. Based on the duration of the transient state controller is connected to the actual power system in
where in the corresponding frequency regulation Jocheon substation, Jeju Island. Previously frequency
control command is applied. regulation of the actual system was only done for a
Power was transferred to PCS, at 180 ms until output short time (3 seconds) as shown in Fig. 13. Our latest
target value (1000 kW) is reached and thus there will data shows frequency regulation for 3 days. This is
be a continuous rise of frequency and corresponding shown in Fig. 14. Real time communications are also
variability. Fig. 12 shows the generation delay time of done for the entire BESS frequency regulation control
60 ms, which represents the time for the BESS to system and grid with corresponding actual battery
discharge power to the distribution line as commanded charging and discharging dispatch.
by the controller. In Fig. 12, it shows that 1000 kW It is necessary to control the supply time of the
BESS power supply switch by three conditions [10].
First, it should be controlled within the dead-time of
the generator to the frequency change time. Second, it
must be able to ensure the supply time of about a few
seconds for the supply switch to change from transient
state to the normal state. Then it should be able to
execute control that does not impact the system overall
dispatch time. Considering the three conditions, the
algorithm could be able to calculate the amount of
frequency adjustment period required for the BESS to
Fig. 11 Steady state frequency regulation using BESS with operate for the frequency regulation of the power
controller. system If deployment is faster, there will be shorter
supply time and will ease the requirement for longer
feed time for deployment. Faster deployment time for
operating an energy storage device gives a stable
system with the capacity to operate for much less time
in using the energy storage system. In this paper, the
state of control mode was confirmed by the operation
control of the supply switch. Future research could
concentrate in determining deployment time for
Fig. 12 Transient state frequency regulation using BESS optimum performance of BESS.
with controller. It was therefore confirmed based from the simulation
294 Application of Frequency Regulation Control on the 4MW/8MWh Battery Energy
Storage System (BESS) in Jeju Island, Republic of Korea

test results that it is possible for a power system 4. Conclusions

frequency regulation control operation to be applied
Based on the results, it is confirmed that simulation
for an energy storage device with a response time of
results for the power system frequency regulation used
130 ms using commercially available programming
in this paper are similar to real-time operation of the
languages and controller devices.
power system frequency regulation in JejuIsland. It
was confirmed that frequency regulation can be
performed by the energy storage device, BESS with
4MW/8MWhr capacity. It was found out that the
frequency control response is 130 ms to drive the
BESS to correct the error of the frequency of the power
system based on 60 Hz reference frequency.
Communication links between the different
equipment, the control algorithm can be coded in the
field control system and the drive to modify
complement, was the most difficult part of the screen
for the operation control of the new system operation
Fig. 13 Continuous frequency regulation of the total BESS during the development processing subsequent
system for 3 seconds. development speed and reliability of data for the
energy storage device. Development of a
communication system with, was considered to be
pursued in parallel from the initial research and
development is suitable for the latest control algorithm
development.
Rate response to the frequency of the energy storage
device error will accelerate more total reduction of the
energy storage capacity required for the follow-up
frequency of the power plant operation roles. We
(a)
expect to be able to replace the role frequency tracking
operation is also less with plant capacity by studying
the factors identified in this paper, and shorten the time
delay element
All of these findings and results on implementing
the frequency regulation control of the total BESS to
the actual grid marks a significant milestone to the
commercialization of large scale BESS for the power
system in Republic of Korea.

References
(b) [1] S. Kalyani, S. Nagalakshmi, R. Marisha, Load frequency
Fig. 14 Continuous frequency regulation of the total BESS control using battery energy storage system in
system for 3 days: (a) actual 3 days; and (b) magnified 2 interconnected power system, in: Third International
hour period. Conference on Computing, Communication and
 Application of Frequency Regulation Control on the 4MW/8MWh Battery Energy 295
Storage System (BESS) in Jeju Island, Republic of Korea
Networking Technologies (ICCCNT’12), India, July, 2012. Engineers 62 (4) (2013) 444-450.
[2] P. Mercier, R. Cherkaoui, A. Oudalov, Optimizing a [6] P.W. Murrill, Fundamentals of Process Control Theory,
battery energy storage system for frequency control Instrument Society of America, 2009, p.100.
application in an Isolated power system, IEEE [7] G. Mcmillan, Good Tuning: A pocket guide, 2nd ed., The
Transactions on Power System 24 (3) 1469-1477. Instrumentation, Systems and Automation Society, 2005,
[3] C.F. Lu, C.C. Liu, C.J. Wu, Effect of battery energy pp. 88-91.
storage system on load frequency control considering [8] G.P. Lim, Demonstration to operate and control
governor deadband and generation rate constant, IEEE frequency regulation of power system by 4MW energy
Transactions on Energy Conversion 10 (3) 555-561. storage system, The Transactions of the Korean Institute
[4] Power System Operation Records, KPX, December 2013, of Electrical Engineers 63 (3) 169-177.
p. 11. [9] C.L. Smith, Practical Process Control: Tuning and
[5] J.B. Han, U. Garam, K.S. Kook, B. Chang, A study on the Troubleshooting, John Willet & Sons, Inc., 2009, pp.
criteria for setting in the dynamic control mode of battery 151-153.
energy storage system in power systems, The [10] K. Tweed, Faster Frequency Regulation Triples in PJM,
Transactions of the Korean Institute of Electrical Greentech Media, November 2013.