982

IEEE TRANSACTIONS ON SMART GRID, VOL. 5, NO. 2, MARCH 2014

Sizing Strategy of Distributed Battery Storage System

With High Penetration of Photovoltaic for Voltage
Regulation and Peak Load Shaving
Ye Yang, Student Member, IEEE, Hui Li, Senior Member, IEEE, Andreas Aichhorn, Student Member, IEEE,
Jianping Zheng, Senior Member, IEEE, and Michael Greenleaf, Student Member, IEEE

AbstractThis paper proposes an effective sizing strategy for

distributed battery energy storage system (BESS) in the distribution networks under high photovoltaic (PV) penetration level. The
main objective of the proposed method is to optimize the size of
the distributed BESS and derive the cost-benefit analysis when the
distributed BESS is applied for voltage regulation and peak load
shaving. In particular, a system model that includes a physical battery model and a voltage regulation and peak load shaving oriented
energy management system (EMS) is developed to apply the proposed strategy. The cost-benefit analysis presented in this paper
considers factors of BESS influence on the work stress of voltage
regulation devices, load shifting and peaking power generation, as
well as individual BESS cost with its lifetime estimation. Based on
the cost-benefit analysis, the cost-benefit size can be determined for
the distributed BESS.
Index TermsBattery storage, cost-benefit analysis, distributed
PV system, overvoltage.

I. INTRODUCTION

N distribution network, grid-tied photovoltaic (PV) installations have grown dramatically [1]. Under high PV
penetration levels, large reverse power may lead to voltage
rise, which will cause potential issues to the distribution system
[2][6]. Traditional voltage control devices, such as on-load tap
changer (OLTC) transformer and step voltage regulator (SVR),
are usually used to regulate the voltage within the normal
voltage range. However, this voltage regulation technology will
suffer from increased work stress resulting in reduced system
lifetime under high penetration PV applications. Therefore,
improved voltage regulation methods have been developed
to solve the overvoltage issue [7][11]. Limiting real power
has been verified to be more effective than reactive power to
regulate the voltage in the distribution line [12]. Accordingly,

Manuscript received February 06, 2013; revised May 24, 2013; accepted
September 03, 2013. This work was supported by National Science Foundation
under Grant ECCS-1001415. Date of publication September 25, 2013; date of
current version February 14, 2014. Paper no. TSG-00089-2013.
Y. Yang, J. Zheng, and M. Greenleaf are with the Center for Advanced Power
Systems, Florida State University, Tallahassee, FL 32310 USA.
H. Li is with the Center for Advanced Power Systems, Florida State University, Tallahassee, FL 32310 USA, and also with the Department of Electrical and
Computer Engineering, Florida State University, Tallahassee, FL 32310 USA
(e-mail: hli@caps.fsu.edu).
A. Aichhorn is with the R&D Department, Company Trench, Austria (e-mail:
andreasaichhorn@gmail.com).
Color versions of one or more of the figures in this paper are available online
at http://ieeexplore.ieee.org.
Digital Object Identifier 10.1109/TSG.2013.2282504

a possible solution for the overvoltage issue is to integrate

battery energy storage system (BESS) into PV systems in
order to achieve a flexible real power control. Recent research
works have demonstrated that the BESS could help OLTC and
SVR to prevent the overvoltage caused by high penetration
PV in distribution systems [7], [8], and meanwhile reduce or
eliminate voltage violation without OLTC or SVR [9][11].
However, the optimized BESS size has not been considered
in these researches which usually lead to an oversized BESS
usage resulting in extra cost. This has hindered the BESS commercial applications in the distribution systems. As a result, a
cost-benefit battery size research is important and beneficial.
Aiming to obtain an economical and serviceable BESS, a battery sizing strategy has been developed focusing on shaving the
peak demand in a residential power distribution feeder with PV
system [13]. The cost and size of hybrid PV and battery system
have been also analyzed for demand side application by taking
the advantages of the on-peak and off-peak electricity charge
difference [14]. However, these studies have not considered factors related to BESS sizing on the supply side of the distribution
system, such as reduced workload on OLTC and SVR, as well as
reduced peaking power generation cost, which has degraded the
BESS size planning and evaluating. Furthermore, the lifetime
of BESS should also be considered and evaluated in the sizing
procedure. Previous studies [13], [14] could not provide precise
cost estimations since they are based on general battery lifetime estimation. However, BESS lifetime varies greatly under
different usage, which depends on the PV-bus feeders location,
PV penetration level, local weather and battery type. As a result,
these sizing strategies and cost analysis need to be improved due
to the lack of the detailed battery lifetime estimation and other
operational information of the power system.
In this paper, a novel method is presented to solve the above
issues. Firstly, a BESS was integrated into each PV bus of the
General Electric (GE) distribution power system model. The
real annual load information, PV power profile, and temperature data were adopted. The proposed method is applied on the
modified GE model. Secondly, a physical model of a lithium-ion
phosphate (LiFPO4) battery was developed with aging effect
so the lifetime information under different operation conditions
can be derived. In addition, several factors that impact the BESS
sizing including OLTC/SVR operation positions, load shifting
and peaking power generation were recorded during the whole
BESS lifetime. The BESS usage, lifetime and system performance in terms of battery size were then analyzed on every bus

1949-3053 2013 IEEE. Translations and content mining are permitted for academic research only. Personal use is also permitted, but republication/
redistribution requies IEEE permission. See http://www.ieee.org/publications_standards/publications/rights/index.html for more information.

YANG et al.: SIZING STRATEGY OF DISTRIBUTED BATTERY STORAGE SYSTEM

983

Fig. 1. The proposed battery cost-benefit analysis system diagram.

as well as on the total feeder of the developed distribution power

system under different PV penetration levels. Finally, the cost of
each bus was evaluated based on the above information to derive a cost-benefit BESS size.
II. BATTERY COST-BENEFIT ANALYSIS SYSTEM DESCRIPTION
Fig. 1 shows the proposed battery cost-benefit analysis system that is applied to investigate battery lifetime,
OLTC&SVR work stress, peaking power generation and peak
load shifting in a distribution power system with BESS and
PV units. It consists of two subsystems. Subsystem 1 includes
a distribution power system model, a physical battery model
and a voltage regulation and peak load shaving oriented energy management system (EMS); subsystem 2 contains a cost
calculation block and a cost-benefit BESS sizing block.
In subsystem 1, the battery cost-benefit analysis process
starts with a three-step cycle for a given battery size and PV
penetration level. Firstly, the distribution power system model
reads current input data including annual load power profile, PV
power profile and temperature profile, records the SVR/OLTC
operations, then delivers the OLTC&SVR work stress, peaking
power generation and peak load shifting information to subsystem 2 and generates dynamic buses voltages for EMS. EMS
is the control center for voltage regulation and battery protection. Secondly, EMS compares the buses voltages with the
bus voltage references which are the upper and lower voltage
limits. EMS is able to control BESS to regulate the bus voltage
within normal range by sending charge/discharge requests to
the battery. OLTC and SVR will take actions if there is still
overvoltage or undervoltage on the buses after BESS operation.
Depth of discharge (DOD) and maximum charge/discharge
power will be evaluated to protect the battery. The DOD reveals
the batterys operation range in state of charge (SOC) [15].
In this paper, the operation range of SOC is set from 15%
to 85%. The output of EMS block is the charge or discharge
command to each physical battery that is modeled in this paper.

In addition, the charge/discharge power is dynamically adjusted

to protect battery from overcharging/overdischarging which
could degrade its lifetime. Finally, the physical battery model
updates the SOC and state of health (SOH) after obtaining the
new charge/discharge power command. The above three steps
will be repeated and new input data will be read again until the
batterys SOH reaches to the end-of-life condition of battery,
which means the battery is scraped and the desired battery
lifetime evaluation is finished.
After obtaining each battery units lifetime, OLTC&SVR
work stress, peaking power generation and peak load shifting
information from subsystem 1, the economical annual cost
calculation is performed considering BESS unit annual cost,
the benefit from SVR/OLTC work stress relieve, load shifting
and shaved peaking power generation in the subsystem 2.
Therefore, the cost of each bus can be calculated under every
possible battery size with different PV penetration level. As
a result, a desired cost-benefit BESS size can be derived.
The detailed distribution power system model, battery model
and EMS development in subsystem 1 are firstly discussed
in Section III. The cost-benefit analysis and BESS sizing is
presented in Section IV. The simulation results are provided in
Section V.
III. SYSTEM MODELING AND ENERGY MANAGEMENT
STRATEGY
A. Distribution System Description
A distribution power system model developed by the GE [3]
is selected and modified in this paper to investigate voltage
rise, BESS usage and system performance. Fig. 2 shows that
the distributed PV units, BESS and loads are integrated into
the GE model. The model includes some fundamental distribution system components: OLTC as the substation transformer,
SVR and switched capacitors. This study is focused on Feeder
2. This feeders length is about 6 miles and the total peak load

984

IEEE TRANSACTIONS ON SMART GRID, VOL. 5, NO. 2, MARCH 2014

Fig. 2. The modified GE distribution power system model.

TABLE I
FEEDER INFORMATION IN THE DISTRIBUTED POWER SYSTEM

is 11MVA. Seven loads represent a mixture of both residential

loads and commercial peak loads ranging from 0.3MW to 5MW
with 0.92 power factor. Hybrid PV and BESS units are connected on 201 to 207 buses. The primary feeder base voltage is
12.5 kV. The secondary feeder base voltage is 240 V for residential loads. The service capacity is rated at 1.5 per unit (p.u.) related to the served load. The impedance of transformers is 2.5%
and X/R ratio is 1.5. An average length of 200-ft was selected
for the distance from the transformer to the load. The secondary
feeder impedance is calculated based on the conductor with 200
A of thermal capacity. The detailed feeder information is listed
in Table I.
B. The Energy Management System (EMS)
The EMS is the control hub in the distribution power system.
The voltage regulation and peak load shaving oriented EMS
controls the power flow among the battery, PV, load and grid.

Although the BESS can achieve more functions, voltage regulation and peak load shaving are two main functions considered
in this study. The EMS sends charge command to BESS when
overvoltage caused by reverse PV power happens, and conveys
discharge command to BESS during peak load period. The bus
205 is selected as an example to describe the power control flow
and shown in Fig. 3. The same strategy can be applied to other
buses. The initial voltage of bus 205 is obtained based on the
load and PV power data. The bus voltage is then compared with
the voltage upper and lower limits. If the bus voltage is controlled within the normal range, EMS collects next load and PV
power data. Otherwise, one of the following two conditional
voltage control strategies is activated:
(a) Overvoltage regulation flowchart: On the one hand,
when overvoltage occurs on bus 205 and PV is generating power, battery will be charged with the excess PV
power until the bus voltage drop to normal range. On
the other hand, when overvoltage occurs but PV is not
generating power or the battery charge flag shows that
battery reaches to its power limit, the closest voltage
control device will step down the tap position to reduce
the voltage.
(b) Undervoltage regulation and peak load shave flowchart:
On the one hand, during peak load time or when voltage
sag happens, the battery will discharge to support the bus
voltage. On the other hand, when the battery discharge
flag shows that battery reaches to its limit and voltage sag
still exists, the closest voltage control device will step up
the tap position to boost the voltage.
C. Physical Battery Model With Aging Effect Estimation
Physical battery model is critical to investigate the battery
lifetime including SOC and SOH. In this paper, a physical
battery model is developed based on a commercial LiFePO4
battery from A123 systems (APR18650) due to its good lifetime performance and high power and energy density [16].
The developed battery equivalent circuit model is shown in

YANG et al.: SIZING STRATEGY OF DISTRIBUTED BATTERY STORAGE SYSTEM

985

Fig. 3. The control flowchart of the proposed EMS: (a) overvoltage regulation flowchart; and (b) undervoltage regulation and peak load shave flowchart.

Fig. 5. Cyclation based irreversible capacity loss.

Fig. 4. Battery equivalent circuit model.

Fig. 4.The detailed parameters may change with temperature

and have been described in [16][18].
The developed battery circuit model is integrated into the distribution power system model of Fig. 2. The SOC and SOH of
the battery can be calculated and updated. The SOC is the feedback to the EMS which indicates the energy can be converted
for the next status. The SOH specifies the leftover lifetime of
the battery. The manufacturer defines the battery life ends when
the capacity reaches 80% or less of the initial one in a fully
charged state [16]. In order to obtain the battery lifetime, two
major aging effects are considered to estimate the irreversible
capacity loss of the battery. One is the cyclation based state of
health
and the other is calendrical aging
.
1)
: The cyclation based capacity loss is derived from
counting the amount of transferred coulombs. The battery cell
(APR18650) is capable of 2,000 cycles at 100% DOD [16]. Accordingly, the
can be reflected from the battery capacity
change in the available cycles. Fig. 5 shows the battery capacity
in terms of the available battery cycles to illustrate the estimated
available volume of coulombs before the battery is scraped. The
linear function for capacity reduction rate is selected in this

paper to calculate the volume of coulombs during battery discharge in order to simplify the analysis. Once the discharged
coulombs are equal to the total area highlighted in Fig. 5, the
battery is scraped. The challenge case is the battery is always
working at 100% DOD. When the battery life ends, the battery
cycles will reach to the limit value 2000 cycles. In the real application, 100% DOD is not practical and the capacity reduction rate may be non-linear, therefore, the battery cycles may be
more than 2000 cycles when the battery capacity drops to 80%.
In addition, the variation of environment temperature (T)
also has an effect on
calculation. The terminal voltage
of the battery changes with [18]. In this case, the current and
the amount of used coulombs under the same amount of power
change with . Therefore, the temperature influence is also
considered in the evaluation of
.
can be estimated
by the division of the used coulombs with the total available
coulombs (1-b).
2)
: The
is the effect of capacity loss occurring
at non-operating state of the battery. The three main factors for
the calendrical aging effect are T, current SOC, and the non-operating time of the battery. The estimation of this aging effect
is shown in Fig. 6 which is based on the measurement results
from [19]. In these functions, the capacity loss/day and can be

986

IEEE TRANSACTIONS ON SMART GRID, VOL. 5, NO. 2, MARCH 2014

where the BESS on bus i is defined as

. The power conversion system (PCS) cost is proportional to the batterys power
rating
.
is the unit cost of the PCS. The cost of
the storage is proportional to each batterys capacity
.
is the unit cost of the battery. is the system efficiency
[21]. The installation cost is proportional to the sum of PCS and
storage cost [22], and it can be represented by introducing a installation cost factor
.
. is the levelized factors related to the
s lifetime.
For generating the same amount of peaking power, gas
combustion turbine plant cost is reduced due to the unleashed
power from
during peak load time. The annual benefit
depending on the peaking power generation from
can
be derived by:
(4)
Fig. 6. Experimental measurements of the capacity loss per day [19].

determined by the given SOC and T. Finally, the

is estimated by the product of capacity loss/day with non-operating
time (1-c).
The above two factors are combined to calculate the battery
aging effect and obtain the battery lifetime. Therefore, the state
of health of the battery can be expressed as:
(1)

IV. COST-BENEFIT ANALYSIS AND BESS SIZING

Fig. 1 shows that the BESS lifetime, OLTC&SVR work
stress, peaking power generation and peak load shifting information can be obtained from subsystem 1. Therefore, the cost
of each BESS and the whole feeder can be calculated based
on the BESS lifetime. The cost evaluation will determine a
cost-benefit BESS size. In order to design a battery size more
accurately, the cost calculation will consider multiple factors
including the battery annual cost, annual benefit from reduced
OLTC/SVR work stress, annual benefit from shaved peaking
power generation and annual benefit from load shifting.
In view of the BESS lifetime varies a lot under different scenarios, the lifetime levelized annual cost analysis method [20]
is adopted in this paper. In general, the levelized cost over its
lifetime
is given by:
(2)
where is the current cost; is its lifetime (years) and is the
discount rate [20]For simplification, the levelized factor over n
years is denoted as
.
Therefore, the levelized annual cost of BESS on bus can be
expressed as:

(3)

is the annual average peaking power generation

where
from
and
is the unit levelized annual cost of gas
combustion turbine [23].
The annual benefit for bus from load shifting can be calculated by:

(5)
where
is the annual stored energy on
during overvoltage regulation and the energy is used to shave peak load
during on-peak time.
and
are the electricity rate at peak load and off-peak load, respectively.
The annual benefit as of the annual saved OLTC&SVR operation is reflected in the reduced operation and maintenance
(O&M) cost. Since all 7 BESS units contribute to the work load
mitigation of OLTC&SVR, this benefit from
can be calculated as:

(6)
where

is the capital cost for both OLCT and SVR.

is the O&M cost factor [24]. According to [25], the
OLTC transformer requires one major maintenance after every
150,000 operation times
.
is
the annual saved operation times when BESS is applied.
is the annual operation times of BESSi and it is used to assign
the benefit to each.
Based on (3) to (6), the annual cost of installing
to
regulate voltage and shave peak load can be obtained by:

(7)
The
size can minimize the annual cost is defined as the
cost-benefit size. Therefore, the cost-benefit BESS size on bus
can be obtained by minimizing
. Equation (7) can
be applied to obtain the annual cost and the cost-benefit size for
every
in the distribution network.

YANG et al.: SIZING STRATEGY OF DISTRIBUTED BATTERY STORAGE SYSTEM

987

V. CASE STUDY AND SIMULATION RESULTS

The proposed method of cost-effect BESS size evaluation has
been applied on the distribution system model and simulated
using Matlab simulation. Four cases are selected to evaluate the
proposed method. Case A compares the system performance
with and without BESS in one day. Case B shows the BESS lifetime and usage. Case C highlights the OLTC/SVR work stress
and peaking power generation with BESS. Case D presents the
BESS cost analysis results under different scenarios.
The simulation results were derived based on the strategy of
Fig. 1 where the local temperature profile is from [26]. The load
data is obtained from [27] and the peak load is rated to the 7
buses peak load on feeder 2 according to [3]. The PV power
profile is provided by [28]. The local PV penetration level is
defined as:
(8)
where
is annual PV peak power, and
is the annual
local peak load. Then the data from [28] is linearly rated to the
desired value on each bus.
The
size in the simulation study can be expressed as:

Fig. 7. 24 hour results without BESS.

(9)
where
capacity of BESSi.
is defined as the number of
desired operation hours of BESSi when it is providing local peak
load
.
A. Case Study A: System Performance With and Without BESS
in One Day
In order to investigate the effect of BESS integration with
proposed EMS on system operation, the system performance
with BESS and without BESS in one particular day under
are firstly shown in Figs. 7 and 8. The BESS size applied for
this case study is 2-hour peak load and the PV, Load and BESS
power profile is based on bus 205.
The same power profiles in 24 hours of PV and load are
shown in (a) of both Figs. 7 and 8. The
action for voltage
regulation is illustrated in Fig. 8(a) with the stored energy in
BESS from 8:00 am to 12:00 am, which contributes to eliminate
the bus overvoltage caused by large PV output power. During
the peak load time, BESS provides energy to load in order to
shave the peak load. Figs. 7 and 8(b) show the primary and
service voltage profile. The voltage limits of primary voltage
and the service voltage are adopted as 0.97 p.u.1.05 p.u. and
0.94 p.u.1.04 p.u., respectively [3], [29]. In Fig. 7(b) the service voltage is higher than primary voltage due to the reverse
power flow. With the BESS operation, these two voltages are
very close as shown in Fig. 8(b). The OLTC&SVR tap positions in one day are shown in (c). Fig. 7(c) shows the tap positions has to be changed frequently in order to regulate the bus
voltage within normal range if BESS is not applied. But the tap
position can be kept to be constant with the help from BESS in
Fig. 8(c). Therefore the work stress of OLTC/SVR can be relieved with BESS by reducing the tap change operations.

Fig. 8. 24 hour results with BESS.

B. Case Study B: BESS Lifetime and Peak Load Shifting

In this paper, bus 205 is selected as an example to investigate the BESS lifetime and usage. The similar analysis can be
applied to other buses. Fig. 9 shows
lifetime under
different BESS sizes and PV penetration levels for the applications of voltage regulation and peak load shaving. The BESS
lifetime can be estimated based on (1). The PV penetration level
and BESS size are defined by (8) and (9) respectively. It can be
observed that under low PV penetration,
lifetime increases with the size but the increasing rate becomes slow after
2-hour peak load BESS size. On the other hand, under high PV
penetration the battery will be used more frequently, therefore,
the lifetime increases with the size growth nearly linearly. Moreover, for a chosen size of
, its lifetime decreases significantly with the growing of PV penetration .

988

Fig. 9. BESS lifetime on bus 205.

IEEE TRANSACTIONS ON SMART GRID, VOL. 5, NO. 2, MARCH 2014

Fig. 11. Operation time with and without BESS.

Fig. 12. Peaking power generation on bus 205.

Fig. 10. Annual stored energy on bus 205.

As described in (5), the benefit from load shifting depends

on the stored energy in
. Fig. 10 presents the annual
stored energy by
under different PV penetration level
and BESS size. Under higher PV penetration level, BESS is required to absorb more energy to solve overvoltage issue, therefore, the annual stored energy increases rapidly if the BESS
size is expanded. However with lower penetration level, there
may be only less reverse PV power resulting in bus overvoltage.
Therefore, there is less energy needing to be stored in BESS for
voltage regulation. As a result, a certain BESS size is enough
to meet the requirement of stored energy. Expanded BESS size
will not contribute to the saved energy.
The overall trends of
lifetime and annual stored energy of other buses are similar to the results of
.
C. Case Study C: OLTC/SVR Work Stress and Peaking Power
Generation
As shown in Fig. 7, the BESS is helpful to reduce the
work stress on OLTC/SVR and achieve the peaking power
generation. The integrated BESS on all 7 buses cooperatively
regulate the buses voltages and reduce the OLTC/SVR operation times. Fig. 11 shows the annual operation times with
and without BESS action under different PV penetration level
and BESS size. The top line shows the annual operation times

of OLTC and SVR without BESS action. The operation times

increase significantly with the growth of PV penetration level.
The number under
is almost doubled the one under
. The lines below show the annual operation times with
different selected BESS sizes. The difference between these
lines and the top one represents the saved operation times.
Under higher and larger BESS size, the difference is more
apparent. This means the saved operation time is more.
This result is used to evaluate the benefit from mitigating the
work stress of OLTC/SVR in (6). It also can be seen from Fig. 11
that the operation times under small BESS size increase quickly
with . But the number is still much smaller than the one without
BESS action. In addition, it is obvious that large size BESS
helps to reduce operation times rapidly under higher . However, a certain BESS size even small size is enough to achieve
the desired operation times under lower . Consequently, the
optimized BESS size should be decided based on and desired
operation times.
The peaking power generation due to
integration
dominates the benefit derived in (4). Fig. 12 shows the annual
average peaking power generated by different
sizes
under different . Since the power supplied by BESS during
peak load time is determined by the energy stored in
,
the trend of the shaved peaking power is similar as the annual
saved energy in Fig. 10. It is clearly shown that the peaking

YANG et al.: SIZING STRATEGY OF DISTRIBUTED BATTERY STORAGE SYSTEM

Fig. 13. Annual cost-benefit analysis under

: (a) quantitative annual cost on all buses; and (b) normalized annual cost on all buses.

TABLE II
KEY PARAMETERS RELATED TO COST EVALUATION IN (3)(7)

power generation increases with the PV penetration level and

BESS size under higher . But the peaking power generation
will be constant under lower when the BESS size reaches to
a certain value. Accordingly, BESS size can be selected based
on the and peaking power generation requirement.
D. Case Study D:

989

Cost-Benefit Analysis Results

In order to illustrate the

cost evaluation, the BESS annual costs on all 7 buses of feeder 2 are analyzed and compared
according to BESS sizes. The
size will change from
half-hour peak load power provision to 5-hour peak load power
support. The annual costs are also explored under different
as shown in Figs. 1315. Table II shows some key parameters
applied in this analysis to derive the results of Fig. 1316 from
(7). These parameters can be changed based on the users applications.
Fig. 13(a) shows the quantitative annual costs of BESSi under
. Due to different peak load on these buses, the BESS
actual capacity is different especially for bus 207 which has the
largest peak load [3]. It can be observed that the annual costs
of all 7 BESS are positive with all the possible BESS sizes.
Therefore, there is no economic profit of every BESS. But, the

lowest annual cost can be determined from this result. In order

to clearly reveal the relationship between size of
and its
annual cost, the quantitative annual cost results are normalized
in the range from
to 1 based on the peak-peak annual cost of
. Fig. 13(b) shows the normalized cost results. It is much
clearer to illustrate the variation trend of
annual cost
with growth of size. Then, the lowest cost can be found and the
cost-benefit
size can be obtained. In the demonstrated
system, the 3-hour peak load BESS size can bring the minimum
cost for
; 2.5-hour peak load for
; 2-hour peak
load for
; 1.5-hour peak load for
. The cost
keeps increasing with the BESS size for others. The normalized
results facilitate to find the cost-effect BESS size of all the buses,
but the real value of the annual cost should be obtained from
Fig. 13(a).
The
quantitative annual costs under
are presented in Fig. 14(a). Similarly, the normalized annual cost results are shown in Fig. 14(b). For
, the cost-benefit size
is around 2-hour peak load. 1.5-hour peak load is the cost-benefit BESS size for
and
. The cost keeps increasing with the BESS size for others. It is can be seen from
Figs. 13 and 14 that the cost-benefit size of the same BESS reduces with the PV penetration level.
Fig. 15 shows the annual cost of
changes with BESS
size under lower PV penetration level
. The
quantitative annual costs are always increasing with the BESS
size. Therefore, it is difficult to find the cost-benefit size for
BESS when the PV penetration level is 50% or lower.
The proposed method can be applied to other distribution systems that include PV and BESS units. If there is no OLTC/SVR,
the benefit in (6) can be ignored. In addition, the system parameters of Table I and some key parameters related to cost evaluation of Table II need to be changed according to different applications. If the applied batterys life cycle and
information
are available, the battery aging effect could also be estimated
based on (1). EMS can be modified flexibly to provide other
functions of BESS besides voltage regulation in the applied distribution system.

990

IEEE TRANSACTIONS ON SMART GRID, VOL. 5, NO. 2, MARCH 2014

Fig. 14. Annual cost-benefit analysis under

: (a) quantitative annual cost on all buses; and (b) normalized annual cost on all buses .

REFERENCES

Fig. 15. Quantitative annual cost on all buses under

VI. CONCLUSIONS
This paper has proposed a new method to discover a cost-benefit BESS size when distributed BESS is integrated into the
distribution system under high PV penetration. The proposed
method is able to obtain the BESS cost and benefits of each bus
under different PV penetration level for any selected BESS size.
Therefore the tradeoff between economic profits and operational
benefits can be evaluated quantitatively. This method has been
demonstrated on a modified GE distribution system model in
this paper. But it can be applied for other distribution systems
with distributed BESS and PV units. When the BESS is used to
achieve other functions including spinning reserve, frequency
regulation, etc. [33], the same method can be applied to determine the cost-benefit battery size with necessary modifications.
Although the economic profits cannot be achieved in this paper
based on current battery price, it is possible to gain economic
revenue when BESS is applied to achieve multiple functions or
the price of BESS is reduced in the future.

[1] C. Whitaker, J. Newmiller, M. Ropp, and B. Norris, Renewable Systems Interconnection Study: Distributed Photovoltaic Systems Design
and Technology Requirements. Albuquerque, NM: Sandia National
Laboratories, 2008.
[2] R. Tonkoski, D. Turcotte, and T. H. M. EL-Fouly, Impact of high
PV penetration on voltage profiles in residential neighborhoods, IEEE
Trans. Sustain. Energy, vol. 3, no. 3, pp. 518527, Jul. 2012.
[3] E. Liu and J. Bebic, Distribution System Voltage Performance Analysis for High-Penetration Photovoltaic. Niskayuna, NY, USA: GE
Global Research, 2008.
[4] F. Katiraei and J. R. Agero, Solar PV integration challenges, IEEE
P&E Mag., vol. 9, no. 3, pp. 6271, May-Jun. 2011.
[5] Y. Ueda, K. Kurokawa, T. Tanabe, K. Kitamura, and H. Sugihara,
Analysis results of output power loss due to the grid voltage rise in
grid-connected photovoltaic power generation systems, IEEE Trans.
Ind. Electron., vol. 55, no. 7, pp. 27442751, Jul. 2008.
[6] J. Matsuki, S. Kawasaki, S. Hosokawa, and N. Kobayashi, Optimal
grid voltage control in distribution feeder connected with PV systems,
in Proc. Photovoltaic Energy Conversion, Conf. Record of the 2006
IEEE 4th World Conf., May 2006, vol. 2, pp. 23272330.
[7] H. Ravindra, M. O. Faruque, K. Schoder, M. Steurer, P. Mclaren, and R.
Meeker, Dynamic interactions between distribution network voltage
regulators for large and distributed PV plants, in Proc. Transm. Distribution Conf. Exposition (T&D), May 2012, pp. 18.
[8] X. Liu, A. Aichhorn, L. Liu , and H. Li, Coordinated control of
distributed energy storage system with tap changer transformers for
voltage rise mitigation under high photovoltaic penetration, IEEE
Trans. Smart Grid, vol. 3, no. 2, pp. 897906, Jun. 2012.
[9] M. Zillmann, R. Yan, and T. K. Saha, Regulation of distribution network voltage using dispersed battery storage systems: A case study of a
rural network, in Proc. 2011 Power Energy Society General Meeting,
Jul. 2011, pp. 18.
[10] J. Cappelle, J. Vanalme, S. Vispoel, T. V. Maerhem, B. Verhelst, C.
Debruyne, and J. Desmet, Introducing small storage capacity at residential PV installations to prevent overvoltage, in Proc. Smart Grid
Commun., Oct. 2011, pp. 534539.
[11] K. H. Chua, Y. S. Lim, P. Taylor, S. Morris, and J. Wong, Energy
storage system for mitigating voltage unbalance on low-voltage networks with photovoltaic systems, IEEE Trans. Power Delivery, vol.
27, no. 4, pp. 17831790, Oct. 2012.
[12] R. Tonkoski and L. A. C. Lopes, Voltage regulation and radial distribution feeders with high penetration of photovoltaic, in Proc. Energy
2030 Conf., Nov. 2008, pp. 17.
[13] C. Venu, Y. Riffonneau, S. Bacha, and Y. Baghzouz, Battery storage
system sizing in distribution feeders with distributed photovoltaic systems, in Proc. 2009 IEEE PwerTech Bucharest, Jul. 2009.
[14] W. F. Su, C. E. Lin, and S. J. Huang, Economic analysis for demandside hybrid photovoltaic and battery energy storage system, in Proc.
Ind. Appl. Conf. 34th IAS Annual Meeting, 1999, vol. 3, pp. 20512057.

YANG et al.: SIZING STRATEGY OF DISTRIBUTED BATTERY STORAGE SYSTEM

991

[15] T. Guena and P. Leblanc, How depth of discharge affects the cycle
life of lithium-metal-polymer batteries, in Proc. Telecommun. Energy
Conf., Sep. 2006, pp. 1,81,8.
[16] [Online]. Available: http://www.a123systems.com
[17] A. Aichhorn, M. Greeleaf, and H. L. Zheng, A cost effective battery
sizing strategy based on a detailed battery lifetime model and an economic energy management strategy, in Proc. 2012 Power Energy Society General Meeting, Jul. 2012.
[18] M. Greenleaf, Physical Based Modeling and Simulation of LiFepo4
Secondary Batteries. Tallahassee, FL, USA: Dept.ECE., Florida
State Univ., 2010.
[19] A. Farman, Erstellen Eines Messtechnisch Gesttzten Modells Zur
Berechnung der Kalendarischen Alterung von LiFePO4-Batterien. Germany: Hochschule Mnchen, 2010.
[20] H. L. Willis and W. G. Scott, Distributed Power Generation Planning
and Evaluation. New York, NY, USA: Marcel Dekker, Inc., 2000, pp.
141144.
[21] Schoenung, M. Susan, and W. V. Hassenzahl, Long-vs. Short-Term Energy Storage Technologies Analysis. A Life-Cycle Cost Study. A Study
for the doe Energy Storage Systems Program. Albuquerque, NM,
USA: Sandia National Laboratories, 2003.
[22] D. Raslter, A. Akhil, D. Gauntlett, and E. Cutter, Energy Storage
System Costs 2011 Update Executive Summary, EPRI, 2012 [Online]. Available: http://www.eosenergystorage.com/documents/EPRIEnergy-Storage-Webcast-to-Suppliers.pdf
[23] Cost of New Entry Estimates for Combustion-Turbine and CombinedCycle Plants in PJM. HILL: CH2M, 2011.
[24] Gomatom, Phanikrishna, and W. Jewell, Feasibility evaluation of distributed energy generation and storage for cost and reliability using the
worth factor criterion, in Proc. 2012 Frontiers Power Conf., 2002.
[25] [Online]. Available: http://www05.abb.com/global/scot/scot252.
nsf/veritydisplay/f741130363e7c6de85256d6f007aade1/$file/
030129_On-load.pdf
[26] [Online]. Available: http://wunderground.com
[27] [Online]. Available: https://www.sce.com/wps/portal/home/regulatory/
load-profiles/!ut/p/b0/04_Sj9CPykssy0xPLMnMz0vMAfGjzOINLdwdPTyDDTwtfAKNDTydnDz9zdxMjA28jfQLsh0VAY010s4!/
[28] [Online]. Available: http://egauge699.d.egauge.net/index.html
[29] Willis and H. Lee, Power Distribution Planning Reference Book.
Boca Raton, FL, USA: CRC Press, 2004.
[30] [Online]. Available: http://www.nrel.gov/analysis/tech_lcoe.html
[31] [Online]. Available: http://evcargo.com/2012/05/23/121/
[32] [Online]. Available: http://www.talgov.com/you/you-customerhelpful-rates-res-elec.aspx
[33] D. Ton, C. Hanley, G. Peek, and J. D. Boyes, Solar Energy Grid Integration Systems-Energy Storage (SEGIS-ES), Sandia National Laboratories, 2008, Report# SAND20084247.

Hui Li (S97M00SM01) received the B.S.

and M.S. degrees in electrical engineering from
Huazhong University of Science and Technology,
Huazhong, China, in 1992 and 1995, respectively.
She received the Ph.D. degree in Electrical Engineering from the University of Tennessee, Knoxville,
TN, USA, in 2000.
She is currently a Professor in the Electrical and
Computer Engineering Department at the Florida
A&M UniversityFlorida State University College
of Engineering, Tallahassee, FL, USA. Her research
interests include PV converters, energy storage applications and smart grid.

Ye Yang (S12) received the B.S. degree in information engineering from Tianjin University, Tianjin,
China, in 2008, and the M.S. degree in electrical engineering from Texas Tech University, Lubbock, TX,
USA, in 2010. He is currently working toward the
Ph.D. degree in the Center for Advanced Power Systems, Florida State University, Tallahassee, FL, USA.
His research interests include integration of renewable energy sources and large scale battery energy
storage system optimization.

Andreas Aichhorn (S11) received the B.Sc. and

M.Sc. degrees from the Upper Austria University of
Applied Sciences/Campus Wels, Austria, in automation engineering, in 2009 and 2011, respectively.
He worked on the M.Sc. thesis in the Center for
Advanced Power Systems, Florida State University,
Tallahassee, FL, USA, as a visiting research scholar.
He is currently working in the R&D department of
the Company Trench Austria as a software developer.

Jim P. Zheng (M87SM13) received the Ph.D. degree in electrical engineering from the State University of New York at Buffalo, NY, USA, in 1990.
He has worked at US Army Research Laboratory,
Fort Monmouth, NJ, USA. He is currently Sprint Eminent Scholar Chair Professor at the Department of
Electrical and Computer Engineering at FAMU-FSU
College of Engineering.
Dr. Zheng is a member of JECS and MRS.

Michael Greenleaf (S06) received the B.S. and

M.S. degree in electrical engineering at Florida State
University, Tallahassee, FL, USA, in 2008 and 2010,
respectively, where he is currently pursuing the
Ph.D. degree in electrical engineering under Dr. J.
Zheng.
His research interests are in energy storage devices
and his current research topic is Modeling and Simulation of Energy Storage Devices.