Received September 14, 2020, accepted October 7, 2020, date of publication October 14, 2020, date of current version

October 26, 2020.

Digital Object Identifier 10.1109/ACCESS.2020.3031197

Optimal Energy Storage Sizing With Battery

Augmentation for Renewable-Plus-Storage
Power Plants
HUNYOUNG SHIN 1, (Member, IEEE), AND JIN HUR2 , (Senior Member, IEEE)
1 Department of Electrical Engineering, Sangmyung University, Seoul 03016, South Korea
2 Department of Climate and Energy Systems Engineering, College of Engineering, Ewha Womans University, Seoul 03760, South Korea

Corresponding author: Hunyoung Shin ([email protected])

This work was supported by the National Research Foundation of Korea (NRF) Grant funded by the Korean Government (MSIT) under
Grant 2019R1G1A110027212.

ABSTRACT The renewable-plus-storage power plant is becoming economically viable for power producers
given the maturing technology and continued cost reduction. However, as batteries and power conversion
systems remain costly, the power plant profitability depends on the capacity determination of the battery
energy storage system (BESS). This study explored an approach for optimal capacity determination of a
BESS combined with renewable energy considering the complex degradation of lithium-ion batteries. The
proposed sizing algorithm iteratively evaluates the effect of BESS operation on battery degradation and
estimates the cash flows of the power plant. In addition, we studied battery augmentation that adds the
storage capacity in the base system to sustain the BESS capacity throughout the project planning horizon.
Using data from South Korea, we showed that both the optimal storage capacity and project profitability
are higher when the BESS is combined with solar generation than when combined with wind generation.
Moreover, simulation results demonstrated that the proposed battery augmentation scheme improves the
project profitability by deferring the upfront cost of batteries and increasing the total revenue. The proposed
approach can provide a comprehensive framework for the parties involved in a BESS project to accurately
calculate the BESS sizes and maximize the project profitability.

INDEX TERMS Energy storage system, energy storage sizing, renewable energy, lithium-ion battery
degradation, battery augmentation.

I. INTRODUCTION requiring its three biggest utilities to procure 1325 MW of

Battery energy storage systems (BESSs) enable fast charg- energy storage for consistent power supply over its power
ing and discharging to effectively enhance the flexibility of system that has high RES penetration.
power grids, especially those integrating several renewable As the costs of RES generation and energy storage
energy sources (RESs). In practice, BESSs provide different are decreasing rapidly and the underlying technology is
services to power grids in many countries. For instance, in the maturing, RES-plus-storage power plants are becoming eco-
PJM energy market, energy storage is mainly used for the nomically viable to independent power producers and end
provision of the frequency regulation service [1]. Likewise, consumers. In particular, solar-plus-storage power plants will
the Korea Electric Power Corporation (KEPCO) installed be competitive with as much as 82% of the projected new
500 MW of energy storage systems (ESSs) for frequency gas peakers in the US, corresponding to 13.1 GW, within
regulation in South Korea. Energy storage can be used to pro- a decade [2]. In markets under government incentives for
vide power systems with solid capacity when combined with energy storage, solar-plus-storage power plants can pro-
RES generation. For instance, California adopted a mandate vide a higher benefit–cost ratio than standalone solar power
plants [3]. Recently, the number of RES-plus-storage projects
is notably growing, with capacity penetration to interconnec-
The associate editor coordinating the review of this manuscript and tion queues reaching 36 GW in 2018, approximately doubling
approving it for publication was Chi-Seng Lam . the 19 GW in 2017 [4].

This work is licensed under a Creative Commons Attribution 4.0 License. For more information, see https://creativecommons.org/licenses/by/4.0/
187730 VOLUME 8, 2020
H. Shin, J. Hur: Optimal Energy Storage Sizing With Battery Augmentation for Renewable-Plus-Storage Power Plants

Despite the substantial decline in prices, BESSs remain ESS size for stabilizing the power and frequency variance due
costly compared to conventional power generation. Hence, to fluctuating RES outputs has been studied in [26], [27].
the profitability of a generation project using a BESS is Lithium-ion battery capacity degradation has been
considerably affected by the determination of its capacity. analyzed using different methods such as chemical
Lithium-ion batteries are subject to capacity degradation, theory [28]–[30], experimental approach [31], and semi-
which is mostly determined by their components and operat- experimental approach [32]–[36]. The chemical-theoretic
ing conditions [5], [6]. As a BESS-based power plant project approach allows to logically explain the causes of the degra-
generally has a lifetime exceeding a decade, lithium-ion bat- dation. However, the obtained battery degradation model
teries typically undergo severe capacity fade throughout the often disagrees with practical battery operation. On the other
project planning horizon. Therefore, BESS sizing neglecting hand, an experimental approach requires a large amount of
battery degradation results in overestimated revenues and data to ensure accuracy, and an empirical model for one
may severely reduce profitability. A precise BESS sizing application obtained from limited data may not be suit-
method must consider that battery degradation is influenced able for other applications. The semi-experimental approach
by the BESS operation and specifications. combines both theoretical analysis with empirical observa-
Several methods are available for BESS sizing. Oversizing tions. A mathematical model of battery capacity degradation
is the conventional method to handle battery degradation by obtained from this approach is robust for a variety of energy
installing higher battery capacity than the required one to storage applications while agreeing with experimental data.
deliver the intended amount of energy at the beginning of In the aforementioned BESS sizing studies, battery capac-
life. Another method is battery augmentation, in which new ity degradation is disregarded or assumed to be independent
batteries are added to the BESS over time. Battery augmen- of the BESS operation and size, despite being primarily deter-
tation defers initial investments and can exploit future cost mined by battery usage according to the application. Vari-
reductions in batteries. In addition, it allows maintaining the ous studies [37]–[41] on battery control strategies have been
battery capacity or performance at a specific level by the end conducted considering battery degradation. However, BESS
of the project. Nevertheless, technical challenges should be sizing imposes additional challenges besides those found in
addressed when implementing battery augmentation, such as the battery control problems. In fact, determining the optimal
interconnecting old and new battery racks without affecting BESS size depends on the prediction of the battery degrada-
the corresponding power electronics. tion according to the BESS size itself. The battery degrada-
Various studies have been conducted on ESS sizing for tion is affected by BESS operating conditions including depth
different applications. Mercier et al. [7] explored the optimal of discharge (DoD), state of charge (SoC), and C-rate, which
sizing of a BESS for frequency regulation of an isolated depend on BESS sizes. Moreover, as battery degradation is
power system using a dynamic simulator of load frequency affected by various stress factors, the prediction of its trends
control. Knap et al. [8] calculated the ESS capacity that over the BESS project planning horizon, which usually spans
provides inertial response and primary frequency reserve for more than a decade, relies on numerical experiments. All
achieving the target inertial response and power and fre- these factors influencing degradation cause high nonlinearity
quency characteristics in a power system. Zhang et al. [9] for BESS sizing.
considered BESS owners participating in the primary con- In addition, to the best of our knowledge, no research
trol reserve market and devised two-level profit-maximizing on a systematic scheme for battery augmentation consid-
BESS planning. Many studies have addressed the optimal ering BESS sizing is available. Finally, most studies on
siting and sizing of BESSs for transmission networks [10], RES-plus-storage power plants consider the BESS applica-
distribution networks [11], or both [12]. Co-optimizing trans- tion to energy arbitrage in wholesale electricity markets.
mission and BESS planning is achieved by a mixed-integer However, RES generation is typically traded via long-
linear model in [13] and a stochastic multistage model in [14]. term bilateral contracts, usually referred to as power pur-
In microgrids, the ESS is an essential power asset for reliable chase agreements (PPAs), rather than in wholesale electricity
operation and effective utilization of RESs [15]. In [16], markets.
[17], energy storage sizing for microgrids is studied consider- In this study, we aimed to address the limitations and
ing both grid-connected and islanded modes. Various works challenges of existing BESS sizing methods. To this end,
related to BESS sizing for RES power plants consider that we propose an optimal BESS sizing algorithm for RES-
a BESS owner participates in the electricity markets while plus-storage power plants considering the effects of BESS
pursuing profit maximization [9], [18]–[23]. In this context, operation and size on battery capacity degradation. Such
energy storage can be used to compensate for the difference effects and the accurate battery degradation estimation allow
between the predicted and actual generation, relieve the gen- to realistically determine the power plant profitability, which
eration output variability, and arbitrage by scheduling charge is directly affected by the battery capacity retention. Addi-
and discharge profiles according to the expected electricity tionally, we integrate a battery augmentation scheme (BAS)
prices and expected RES generation. Other works have been into the proposed BESS sizing algorithm to further increase
conducted on the ESS capacity that enhances the performance the economic benefits. The main contributions of this study
of RES power plants [20], [24], [25]. In addition, the proper can be summarized as follows:

VOLUME 8, 2020 187731

 H. Shin, J. Hur: Optimal Energy Storage Sizing With Battery Augmentation for Renewable-Plus-Storage Power Plants

• An optimal BESS sizing algorithm for solar-plus- storage given the unique characteristics of BESSs. According
storage and wind-plus-storage power plants is proposed. to an analysis of 38 RES-plus-storage PPAs in the US [3],
Unlike existing methods, the proposed algorithm can the various methods to compensate storage within the PPAs
accurately calculate the battery degradation rate con- can be categorized into three options: time-invariant energy
sidering the BESS operation and size to then evaluate prices, time-variant energy prices, and capacity payment.
the cash flows of the BESS project. Therefore, BESS 1) Time-invariant energy prices: This compensation
investors can determine the BESS capacity that maxi- method simply bundles the storage cost into the overall
mizes profitability. energy price. The ESS operation is specified in the PPA.
• A BAS that can be integrated into the BESS sizing 2) Time-variant energy prices: In this method, energy
algorithm is introduced. Simulation results show that prices during peak generation periods are assumed to
battery augmentation can improve the profitability of soar up to multiple times their normal level. For exam-
the RES-plus-storage project. The BAS defers part of ple, several PPAs in Nevada, such as that in Arrow
the initial battery investments, thus providing economic Canyon, allow peak energy pricing from 16:00 to
benefits. Furthermore, the BAS allows to achieve a more 21:00 each day throughout June–August. In South
balanced cash inflow throughout the project planning Korea, the energy discharged from the ESS during peak
horizon compared to the lack of battery augmentation. generation periods on each day is eligible for additional
• Regarding market conditions, a long-term contract (usu- REC (renewable energy certificate) multipliers.
ally a PPA) between the project owner and off-taker is 3) Capacity payment: Some projects have capacity and
considered as a common form of renewable energy trad- energy PPAs for the energy storage to be compensated
ing in many regions including the US, Europe, China, through a fixed capacity payment. A new contract
India, and South Korea. Among the various types of option in eight recent projects in Hawaii allows treating
contracts, we adopt the time-variant energy price in RES-plus-storage as ‘‘renewable dispatchable gener-
South Korea as a representative example. The proposed ation.’’ For compensation, the utility makes a fixed
algorithm can be tailored to other market conditions with payment per month for the ‘‘dispatchable capacity" and
simple modifications. a variable payment to cover operation and maintenance
costs.
The remainder of this paper is structured as follows.
We focus on the second option as case study for opti-
Section II describes various energy compensation rules for
mal BESS sizing. In particular, we consider the time-variant
the RES-plus-storage power plants and details the South
energy pricing rules for RES-plus-storage power plants in
Korean market rules. Section III presents the mathematical
South Korea and derive the corresponding BESS operation
model of lithium-ion battery degradation based on the work
algorithm for accurate sizing. Nevertheless, the proposed
by Xu et al. [36]. In Section IV, we propose the optimal
sizing algorithm can integrate other compensation options
BESS sizing algorithm and BAS for RES-plus-storage power
without major changes, as further detailed in Section IV-C.
plants. Numerical results based on real data are reported in
Section V, and we finally draw conclusions in Section VI. B. ENERGY COMPENSATION RULES IN SOUTH KOREA
A renewable portfolio standard is mandatory for power gen-
II. ENERGY COMPENSATION RULES FOR eration companies to supply a proportion of the total power
RES-PLUS-STORAGE POWER PLANTS generation from RESs. A REC is a tradable commodity that
A. POWER PURCHASE AGREEMENTS proves the RES energy generated by the power producer. One
A PPA is a long-term contract that governs the sale and REC is endowed per megawatt hour of generation. Although
purchase of electrical energy between a power producer and a both the generated energy and RECs can be sold in the
utility off-taker, often for a fixed price. PPAs are widespread corresponding spot markets, the price volatility is a serious
in energy markets and particularly important in renewable risk to investors, especially the small ones. To address their
energy sectors. In practice, renewable energy is commonly concerns, the South Korean government implemented ‘‘fixed-
sold through PPAs for various reasons, such as the ease of price contracts" in 2017 to allow power producers to jointly
project financing and hedging given the risk of fluctuat- trade the generated energy and RECs as one product with
ing energy prices and varying RES generation. Combining KEPCO, which is the only electricity utility in South Korea,
energy storage to RES is a viable option to mitigate such a at a fixed price over a long term.
risk encountered when participating in the wholesale elec- As RES power plants have different economical values
tricity markets. However, the regulatory and market design depending on their type and capacity, different REC multipli-
in many countries are not yet adequate for incorporating ers are imposed per REC to both promote the balanced devel-
energy storage as market participating generators. Therefore, opment of various RESs and foster the competitivity between
a RES-plus-storage power plant is usually compensated by different RESs [42]. Table 1 lists various REC multipliers for
PPAs rather than the wholesale markets. solar and wind power plants according to the type of facility.
The PPAs for RES-plus-storage projects should be estab- Although the government endows high REC multipliers to
lished differently from those for projects without energy RES-plus-storage power plants, not all the energy discharged

187732 VOLUME 8, 2020

H. Shin, J. Hur: Optimal Energy Storage Sizing With Battery Augmentation for Renewable-Plus-Storage Power Plants

TABLE 1. REC multipliers for RES power plants based on the types of RES Cref = 1C, and Tref = 25◦ C. As a result, the model of battery
and facility.
capacity degradation combines cycling aging and calendar
aging as follows [36]:
N
X
fd = fDoD (DoDi )fSoC (SoCi )fC (Ci )fT (Ti )ni
i=1
+kt HfSoC (SoCavg )fT (Tavg ), (2)
where i is the cycle index and N is the number of cycles.
The cycle count is obtained by using the rainflow-counting
algorithm [43], in which H is the total operation time in
TABLE 2. Non-peak generation periods for RES-plus-storage power seconds, and SoCavg and Tavg are the average SoC level
( N
PN
i=1 SoCi /N ) and temperature ( i=1 Ti /N ) across all
plants.
P
cycles, respectively.
Considering the battery residual capacity and solid elec-
trolyte interphase [44] for capacity degradation, the final
model for the battery state of health (SoH1 ) is given by [36]
SoH = pSEI · e−rSEI fd + (1 − pSEI ) · e−fd . (3)
The model parameters for the aging and SoH models are cal-
culated using experimental data from lithium-ion manganese-
oxide batteries [36]. Although we adopt eq. (3) to model
battery degradation, alternative definitions include the model
from an ESS is eligible for the increased multipliers. For
by Stroe et al. [32] based on experimental data from lithium
instance, energy produced by a RES during peak generation
iron phosphate batteries.
and discharged from the ESS during the non-peak generation
periods, as listed in Table 2, is subject to the REC multiplier
IV. PROPOSED BESS SIZING ALGORITHM AND BAS
of 4.0.
We propose an optimal BESS sizing strategy that maximizes
the economic value of BESS installation into an existing RES
III. LITHIUM-ION BATTERY DEGRADATION
power plant. Subsequently, we introduce a complementary
Lithium-ion batteries are subject to capacity degradation as BAS for further increasing the profitability. We begin by ana-
they are used. Therefore, it is necessary to consider battery lyzing the BESS charging and discharging schedules aiming
degradation in BESS sizing. The capacity degradation rate to maximize revenue.
depends on many factors. Thus, it is difficult to obtain an
accurate degradation model. In this paper, we employ the bat- A. BESS OPERATION
tery capacity degradation model proposed by Xu et al. [36],
As we consider the South Korean time-variant energy price,
who used chemical theory to model the battery capacity reten-
a comparison of the prices in different periods is essential to
tion with respect to multiple stress factors and determined the
develop the BESS sizing algorithm. From Table 1, the elec-
model parameters according to experimental data. In addi-
tricity discharged during non-peak generation periods is com-
tion, they calculated the total aging of lithium-ion batteries
pensated with SMP + 4.0 · REC per megawatt hour, where
as the sum of cycling aging and calendar aging. The stress
SMP represents the system marginal prices.2 As the market
factors determining the capacity retention are the levels of
value per REC is comparable to the SMP in South Korea,
DoD and SoC, C-rate, cycle count, temperature, and total
it is economically optimal to store RES generation during
operation time.
peak-generation periods until reaching the full BESS capacity
The aging models for the stress factors are defined as
to then discharge the energy during non-peak generation
follows [36]:
periods. Algorithm 1 details this optimal BESS operation.
DoD : fDoD (DoD) = (kDoD1 DoDkDoD2 + kDoD3 )−1 , (1a) Given the BESS energy capacity or equivalently battery
capacity (EBAT ), the algorithm first checks whether time t
SoC : fSoC (SoC) = ekSoC (SoC−SoCref ) , (1b)
is within a non-peak generation period. If not, the BESS
C-rate : fC (C) = ekC (C−Cref ) , (1c) charges energy (ECh (t)) until reaching its maximum energy
Tref
Temp. : fT (T ) = ekT (T −Tref )· T , (1d) level (EBAT,max ), which is obtained as R · EBAT , where R is the
maximum DoD range (e.g., 90%). ECh (t) is determined by the
where kDoD1 , kDoD2 , and kDoD3 are the DoD aging model 1 SoH defines the condition of a battery as the remaining capacity divided
coefficients, and kSoC , kC , and kT are the SoC, C-rate, by the nominal capacity.
and temperature aging model coefficients, respectively. The 2 In South Korea, the SMP is used as the wholesale market price calculated
reference values per stress factor are SoCref = 50%, without considering transmission constraints.

VOLUME 8, 2020 187733

 H. Shin, J. Hur: Optimal Energy Storage Sizing With Battery Augmentation for Renewable-Plus-Storage Power Plants

Algorithm 1 BESS Operation capacity (in kilowatts hour) that maximizes the NPV over the
Input: BESS energy capacity, EBAT project planning horizon. The power and energy capacities of
Data: RES generation, ERES (t), ∀t the BESS are determined by the size of the PCS and batteries,
Initialization (t = 1, SoC(1) = SoCint ); respectively. The NPV of the BESS installation in the RES
while t < T + 1 do system over Y years of the project with PCS capacity (PPCS )
/ non-peak period then
if t ∈ and battery capacity (EBAT ) is calculated as
if EBAT,rem (t) = EBAT,max − EBAT (t) > 0 then Y
X CFy (PPCS , EBAT,y )
ECh (t) =ρRES-BAT · NPVY (PPCS , EBAT ) =
min{ERES (t), PPCS , EBAT,rem (t)}; (1 + r)y
y=1
else −Cint (PPCS , EBAT ), (4)
ECh (t) = 0;
end where CFy (PPCS , EBAT,y ) is the net cash flow during year
EBAT (t + 1) = EBAT (t) + ECh (t); y, r is the discount rate, and Cint (PPCS , EBAT ) is the initial
else if t ∈ non-peak period then investment of the BESS comprising the PCS and battery
if t ∈ beginning of non-peak period then costs:
t ∗ = t; Cint (PPCS , EBAT ) = CPCS · PPCS + CBAT · EBAT , (5)
∗
∗ = EBAT (t )−EBAT,min ;
EDch REC duration
end with CPCS and CBAT being the unit prices for the PCS and
EDch (t) = min{EDch∗ ,P battery, respectively. The usable battery capacity in year
PCS };
EBAT (t + 1) = EBAT (t) − EDch (t); y (EBAT,y ) is determined by the battery degradation over
t := t + 1; previous years and battery augmentation, if applicable. The
end net cash flow, CFy (PPCS , EBAT,y ), is the difference between
the cash in or revenue, Revy (PPCS , EBAT,y ), and the cash
out or cost, Cy (PPCS , EBAT,y ), during year y:
CFy (PPCS , EBAT,y) = Revy (PPCS , EBAT,y) − Cy (PPCS , EBAT,y ).
RES generation (ERES (t)) but limited by the nominal capacity
of the power conversion system (PCS) (PPCS )3 and remaining The revenue during year y comprises the income from both
energy capacity in the BESS (EBAT,rem (t)), which is calcu- discharging power from the BESS and supplying RES gener-
lated as EBAT,max − EBAT (t) where EBAT (t) is the energy ation directly to the grid:
stored in the battery at time t. If time t is within a non-peak
Revy (PPCS , EBAT,y )
generation period, the BESS discharges energy (EDch (t)) until
T 
the minimum energy level (EBAT,min ), with the discharging X
rate being determined as follows. First, EDch ∗ , a candidate = λBESS · ρBAT-AC · EDch,y (t)
t=1
discharge rate, is determined as the available energy in the  
BESS at the beginning of the non-peak generation period, ECh,y (t)
+λRES ·ρRES-AC ERES,y (t)− , (6)
(t ∗ ), divided by the duration of the non-peak generation peri- ρRES-BAT
ods: EDch∗ = (E ∗ ∗
BAT (t )−EBAT,min )/(REC duration). If EDch is where t indicates the hour in a year, and thus T = 8760
greater than PPCS , the discharging energy for time t, EDch (t), h, λBESS and λRES are the prices for electricity from the
∗ .
is PPCS . Otherwise, EDch (t) is EDch BESS and RES, respectively, and ρBAT-AC and ρRES-BAT are
Algorithm 1 is designed to optimally work for the the efficiency factors from the BESS to the grid and from
South Korean PPA. Due to the energy compensation rule, the RES power plant to the BESS, respectively. Using the
the stochasticity of renewable generation can be omitted electricity pricing rule in Section II-B and the BESS operation
in the algorithm. However, under other compensation rules, algorithm in Section IV-A, the energy discharged from the
the stochasticity of renewable generation may play an impor- BESS is compensated at λBESS = SMP + 4.0 × (REC Price)
tant role to obtain optimal BESS operation and sizing per kilowatt hour. In addition, the remaining RES power after
algorithms [17]–[19], [45], [46]. charging the BESS is priced at λRES = SMP+REC Price per
kilowatt hour.
B. CALCULATION OF NET PRESENT VALUE The costs in year y include the annual operations and main-
The net present value (NPV) represents the current cash tenance cost of the BESS and the income tax from selling
flows generated by a project over a specific period and is the electricity. In addition, the opportunity cost, which is the
widely used to estimate whether a project will result in a potential gain from the standalone RES power plant, must
net profit or loss. BESS sizing aims to find the optimal be included in the costs, because we analyze the NPV of the
combination of power capacity (in kilowatts) and energy BESS installation project. Therefore, we obtain

3 We consider hourly generation, the PCS capacity in kilowatts can be Cy (PPCS , EBAT,y ) = CO&M,y (PPCS , EBAT )
directly used as the limit for charging and discharging energy over time t. +Ctax,y (PPCS , EBAT,y ) + COpp,y . (7)

187734 VOLUME 8, 2020

H. Shin, J. Hur: Optimal Energy Storage Sizing With Battery Augmentation for Renewable-Plus-Storage Power Plants

We assume the annual operations and maintenance cost to algorithm flow. First, the BESS operation algorithm in the
be 1% of the investment cost of BESS and the income tax annual loop must be adjusted to the conditions of the market
to be 10%: CO&M,y (PPCS , EBAT ) = 0.01 × Cint (PPCS , EBAT ) under consideration. Second, the revenue in eq. (6) should be
and Ctax,y (PPCS , EBAT,y ) = 0.1×Revy (PPCS , EBAT,y ), respec- redefined using the energy prices for that market.
tively. The opportunity cost is calculated as The sizing algorithm applies brute-force search to explore
T the solutions from all the possible ranges of battery and PCS
capacities. Although this type of search is computationally
X
COpp,y = λRES · ρRES-AC ERES,y (t).
t=1 inefficient, the computation time can be substantially short-
ened by setting the search ranges appropriately. Furthermore,
The variables associated with the BESS, ECh,y (t) and
as our goal is to obtain the optimal BESS planning over a long
EDch,y (t), and RES generation, ERES,y (t), are affected by the
period, it is more important to guarantee the solution accuracy
performance of the BESS and the RES power plant, whose
than to improve the computational efficiency to obtain the
performance degrades over the project operation time. Thus,
solution.
these variables differ according to year y even under the same
BESS operation.
D. BESS SIZING WITH BAS
Some BESS projects should sustain a minimum energy
C. BESS SIZING
capacity over the planning horizon. As an energy project
The overall process for optimal BESS sizing is described
typically operates for more than a decade, batteries expe-
in Figure 1. First, the PCS and battery capacities are deter-
rience severe capacity degradation, and ensuring capacity
mined, with the maximum and minimum values being based
retention to the minimum requirement throughout the project
on the type of RES and its generation capacity. The algorithm
planning horizon is difficult. Although oversizing can be a
is composed of three loops. The first two loops search the
feasible solution, to ensure sufficient energy capacity over
optimal BESS power and energy capacities, and the third
such long periods may require installing excessive battery
loop calculates the battery SoH and annual revenue and
capacity upfront. In this case, the project profitability can
cost. The PCS size, which represents power capacity of the
substantially decrease.
BESS, establishes the outer (first) loop. The battery capac-
Alternatively, battery augmentation allows to sustain the
ity in the second loop is determined considering the PCS
total energy capacity of a BESS throughout its lifetime by
capacity and the energy-to-power (E/P) ratio in hours4 as
adding battery capacity to the base system, rotating bat-
PCS capacity × E/P ratio. The algorithm uses the E/P ratio
teries in the system, or both. Given its economic benefit,
instead of kilowatt hours to examine the capacity because
battery augmentation is being increasingly adopted by BESS
1) a pair of power capacity and E/P ratio is frequently used to
installers and project developers. Nevertheless, obtaining the
describe the BESS size and 2) it allows searching the energy
optimal BAS is difficult due to the complex factors that must
capacity in more detail regarding the capacity from the first
be considered. For instance, the effect of additional batteries
loop. Given the PCS and battery capacities, the battery SoH,
on the performance of the whole BESS may be difficult to
revenues, and costs are evaluated annually in the third loop.
determine. In addition, battery cost prediction is also required
The BESS operation algorithm allows to determine the stress
for optimal battery augmentation.
factors for battery degradation, such as SoC, C-rate, time, and
We aim to propose a BAS to be easily combined with the
cycle count, and the SoH in year y (SoHy ) can be obtained
BESS sizing algorithm proposed in Section IV-C rather than
using eq. (3). The BESS operation also allows to evaluate
obtain an optimal BAS.
of the revenue and cost in year y from the charging and
discharging schedules and using eq. (6) and (7), respectively.
Algorithm 2 Battery Augmentation Algorithm
After each iteration of the annual loop, the capacities of
the RES-plus-storage power plant are updated based on the Input: EBAT,y
calculated SoH and the degradation rate of the RES systems. Compute SoHy+1 to predict EBAT,y+1= EBAT ·SoHy+1;
Req
This process is repeated until the last year of the project if EBAT,y+1 < EBAT then
SoHyLastAug −SoHy+1
planning horizon. To include battery augmentation, the BAS r̂deg = ;
y+1−yLastAug
in Section IV-D is applied after calculating the annual SoH. Aug Req
EBAT,y = min{EBAT r̂deg (Y −y)−EBAT,y + EBAT ,
After the annual loop, the algorithm evaluates the NPV
for the PCS and battery capacities by combining the revenue EBAT − EBAT,y};
Aug
and cost from eq. (4). When all the iterations are completed, EBAT,y := EBAT,y + EBAT,y ;
the algorithm selects the PCS and battery capacities providing end
the highest NPV as the optimal BESS capacity.
To consider other markets, two processes in the sizing
Algorithm 2 describes the proposed BAS that can be inte-
algorithm should be modified while maintaining the overall
grated into the annual loop of the BESS sizing algorithm
4 The E/P ratio describes the ratio of the energy capacity to the power shown in Figure 1. The timing and capacity of battery addi-
capacity of the BESS [47], [48]. tions are determined as follows:

VOLUME 8, 2020 187735

 H. Shin, J. Hur: Optimal Energy Storage Sizing With Battery Augmentation for Renewable-Plus-Storage Power Plants

FIGURE 1. Proposed BESS sizing algorithm. Battery augmentation (dashed box) is optional.

1) For year y, calculate SoHy+1 using eq. (3) to predict energy capacity to be added to satisfy the minimum
the battery capacity retention at the end of that year, capacity required for the remaining project time based
EBAT,y+1 . Check whether EBAT,y+1 satisfies the min- on the estimated degradation rate, r̂deg . This is illus-
imum energy requirement, as illustrated in curve (1) trated in curves (3) and (4) in Figure 2. The latter
in Figure 2. expression implies that the capacity of battery additions
Req Req
2) If EBAT,y+1 < EBAT , where EBAT is the minimum is restricted by the initial BESS capacity. Without this
energy capacity that must be guaranteed during the constraint, the augmentation algorithm may require
project planning horizon, apply battery augmentation. installing large additional batteries. This would require
Estimate the battery degradation rate (r̂deg ) as the rate additional balance of the system and storage containers
of SoH variation between the year of the last battery to accommodate the batteries, consequently incurring
augmentation and the current year, as illustrated in high costs.
curve (2) in Figure 2. The rate of battery capacity degradation declines over time
Aug
3) Calculate the augmented capacity (EBAT,y ) as the min- as the BESS operates for a fixed maximum DoD, which
Req
imum between EBAT r̂deg · (Y − y) − EBAT,y + EBAT can be estimated from eq. (3) (see, for example, Figure 5).
and EBAT − EBAT,y . The former expression indicates Thus, r̂deg may overestimate the battery degradation during

187736 VOLUME 8, 2020

H. Shin, J. Hur: Optimal Energy Storage Sizing With Battery Augmentation for Renewable-Plus-Storage Power Plants

TABLE 3. Parameter values used for simulations.

FIGURE 3. Costs for BESS excluding PCS from 2020 to 2034.

FIGURE 2. Illustration of parameters in proposed BAS. the Korean won (KRW) is converted into US dollars (USD)
at an exchange rate of 1100 KRW/USD. For the BESS costs,
Aug
the remaining project time. Consequently, EBAT,y may also be the historical costs and projections in [49] are employed as
greater than the required battery capacity for augmentation. reference. Figure 3 shows the cost projections for a BESS
Hence, the BAS convergence is guaranteed. excluding the PCS from 2020 to 2034. Only the cost pro-
The costs of battery augmentation include the battery cost jection before 2031 is available in [49]. Thus, we use curve
and related operations and maintenance cost. Additional costs fitting to predict the costs after that year.
such as those from labor are also included and assumed to be Table 3 lists the parameter values used for the simula-
10% of the battery cost. These costs should be reflected in the tions. The variations of battery temperature are neglected
annual cost in eq. (7) when applying the BAS. because they are complex to evaluate in numerical experi-
ments. Instead, we assume a constant temperature of 25◦ C
V. SIMULATION RESULTS obtained from air-conditioning [32], [36].
A. SIMULATION DATA The simulations were conducted using MATLAB 2019b
We evaluate the proposed algorithms using field data obtained on a computer with the Intel Core i7-4790 CPU and 16 GB
from hourly measurements in solar and wind power plants RAM.
of South Korea during 2017, whose capacities are 1 and
33 MW, respectively. For a fair comparison of the revenues, B. ESS OPERATION AND BATTERY DEGRADATION
we normalize the capacities to 10 MW. The energy price Figure 4 shows the four 24-hour profiles of solar and wind
for the RES-plus-storage power plant corresponds to the generation and the battery SoC levels when applying the
average SMP and RES prices in 2019, which are 89.49 and BESS operation algorithm. The power and energy capacities
62.95 KRW/kWh, respectively. Hence, the PPA price for of the BESS are set to 6 MW and 27 MWh, respectively, and
RES generation is λRES = 152.44 KRW/kWh, and that the maximum DoD range is set to 0.9. The combination of
for BESS discharging during RES generation is λESS = solar power patterns in the diurnal cycle and uniform non-
(89.49 + multiplier · 62.95) KRW/kWh. When the REC peak solar generation periods leads to similar daily SoC pat-
multiplier is 4.0, λESS becomes 341.29 KRW/kWh, which terns with large changes. In contrast, the wind power outputs
is approximately 2.4 times λRES . The PPA period for the are more irregular, but their daily variation is smaller than the
RES-plus-storage power plants is set to 15 years. Hereinafter, solar power outputs. In addition, non-peak wind generation

VOLUME 8, 2020 187737

 H. Shin, J. Hur: Optimal Energy Storage Sizing With Battery Augmentation for Renewable-Plus-Storage Power Plants

FIGURE 4. RES generation and SoC profiles by applying BESS operation

algorithm. Each curve corresponds to a season .

FIGURE 6. NPV, final SoH, total revenue, and total cost per combination
of PCS capacity and E/P ratio for 10 MW of solar generation.

discounted cash inflows at the rate listed in Table 3. The total

cost includes the annual cost in eq. (7) and the initial facility
investment cost in eq. (5), and it is discounted by the discount
rate. In Figures 6(c) and (d), larger PCS and higher E/P ratio
result in higher revenues and costs. In Figure 6(b), the final
SoH level is generally low, being below 63%, over the entire
search space. Especially when the PCS capacity is low and the
E/P ratio is low, the usable battery capacity in the final year
FIGURE 5. Battery capacity retention of RES-plus-storage power plant at can be less than 55% of the nominal value. It remains unclear
different maximum DoD ranges.
whether the final SoH increases with the PCS capacity at a
fixed E/P ratio. Using a larger PCS allows the BESS to oper-
periods vary according to the season. Therefore, the daily SoC
ate at higher C-rate, and larger SoC variations occur within
profile for wind generation exhibits a variety of patterns with
a cycle, accelerating the battery capacity degradation. On the
relatively smaller SoC changes than solar generation.
other hand, at fixed E/P ratio, a larger PCS results in larger
Figure 5 shows the battery capacity retention curves for
battery capacity, mitigating the battery capacity degradation,
the RES-plus-storage power plants obtained by applying the
as shown in Figure 6(b).
BESS operation algorithm for power and energy capaci-
Figure 6(a) shows the simulation results for the NPV under
ties of 6 MW and 27 MWh, respectively. For both RESs,
various PCS capacities and E/P ratios. The optimal BESS
a higher maximum DoD range causes more severe battery
capacity is that corresponding to the PCS size and E/P ratio
capacity degradation. When comparing solar and wind gen-
providing the highest NPV.
eration, the BESS combined with solar generation shows
Table 4 lists the optimal BESS capacities and correspond-
higher degradation than that combined with wind generation,
ing NPVs and final SoHs for a BESS installed in a RES power
because the daily SoC variation is usually larger for solar
plant with and without the BAS. The proposed BESS sizing
generation, as shown in Figure 4. The final capacity retention
algorithm retrieves a 4600 kW PCS and 5.9-hour battery stor-
in Figure 5 may be very small compared to that of multiple
age, totaling 27,140 kWh. For solar generation, energy stor-
studies.5 To guarantee suitable capacity retention, the BESS
age with long duration is preferable because the BESS added
owner may add battery racks to the base system.
to the solar power plant experiences harsh battery degrada-
tion from the large SoC variations shown in Figure 4. The
C. CASE STUDIES
estimated NPV is 910, 500 USD, indicating that the BESS
1) BESS SIZING AND ECONOMIC ANALYSIS FOR installation in the solar system is profitable considering the
SOLAR GENERATION South Korean market. However, the final SoH is low, being
Figure 6 shows the NPVs, final SoHs, total revenues, and only 56.7%, after BESS operation for 15 years. To maintain
total costs of various PCS capacities and E/P ratios for solar the SoH at a moderate level, for instance 80%, throughout the
generation without the BAS. The total revenue is the sum of project operation, battery augmentation should be adopted.
5 In many studies, the end of life for lithium-ion batteries are set to 80% of We also compare the BESS sizing method with two
their nominal capacity, although smaller levels are sometimes adopted, such benchmarks to verify its effectiveness. The first benchmark
as 60 and 70%, practically. performs BESS sizing using the proposed algorithm but

187738 VOLUME 8, 2020

H. Shin, J. Hur: Optimal Energy Storage Sizing With Battery Augmentation for Renewable-Plus-Storage Power Plants

TABLE 4. BESS capacity providing the highest NPV and its final SoH for four types of installations.

The third row of Table 4 presents the simulation results for

the optimal BESS capacity when using the proposed BAS.
The algorithm retrieves two augmentations, with batteries of
4136 and 3344 kWh being added in years 6 and 11, respec-
tively. Thus, the total installed battery capacity is the sum of
the initial capacity (24,200 kWh) and the added capacities
(4136 kWh + 3344 kWh), totaling 31,680 kWh. The battery
augmentations increase the NPV from 910, 500 to 985, 649
USD, representing an improvement of 8.5%. This financial
gain is mainly due to the initial cost reductions in lithium-
ion batteries and the time value of money. In fact, the BAS
defers part of the initial investment, and the battery costs are
expected to decrease in the future.
Figure 8 shows the annual cash flows from operating
the solar-plus-storage power plant. In the top graph of
FIGURE 7. NPV, final SoH, total revenue, and total cost per combination of Figure 8(a), the annual cash inflows decrease over time by
PCS capacity and E/P ratio for 10 MW of solar generation when using BAS. the power plant degradation. The revenue obtained from
BESS discharging constitutes approximately 69% of the total,
neglecting battery degradation, as in [42]. In this case, the cal- despite the BESS accounting for only 42% of the generation.
culated BESS size is 4900 kW PCS and 5.2-hour battery This high revenue is due to the high REC multiplier applied
storage, yielding an estimated NPV of 854,633 USD, which to the BESS. As the opportunity cost represents the revenue
is 6.2% lower than that (910,500 USD) obtained when con- obtained from a standalone solar power plant, it decreases
sidering battery degradation. Thus, we can conclude that over time due to the solar panel degradation. In Figure 8(b),
considering battery degradation in sizing produces a higher installing the augmented batteries charge additional 922, 783
economic value in the BESS project. The second benchmark and 593, 894 USD to the cash flows in years 6 and 11, respec-
uses the same BESS sizing algorithm but considers the bat- tively. The initial BESS installation costs are 7, 616, 400 and
tery degradation trend estimated from a different applica- 8, 518, 280 USD with and without the BAS, respectively.
tion, the primary frequency control adopted in the German By applying the BAS, the BESS owner can achieve more
market [50]. In this benchmark, the calculated BESS size is consistent cash inflows throughout the project operation,
4600 kW PCS and 6.7-hour battery storage. The estimated obtaining an increase in total revenue from 44, 096, 358 to
NPV is 784,559 USD, being 13.8% lower than that obtained 44, 782, 068 USD.
considering the battery degradation estimated from BESS The simulation time to produce the solution was
operation established in the South Korean PPA. Therefore, 1072 seconds when BAS was not considered. With the BAS,
the optimal BESS sizing depends on accurate estimation of the simulation time increased to 1280 seconds.
the battery degradation considering the BESS operation.
Figure 7 shows the simulation results when integrating the 2) BESS SIZING AND ECONOMIC ANALYSIS
proposed BAS into BESS sizing. Although the NPVs, final FOR WIND GENERATION
SoHs, total revenues, and total costs are calculated consider- Figures 9 and 10 show the NPVs, final SoHs, total revenues,
ing the BAS, the E/P ratio only represents the initial condi- and total costs of various PCS capacities and E/P ratios for
tion. Figure 7(b) shows that the final SoHs are higher than the wind generation without and with the BAS, respectively.
predefined minimum requirement of 80% regardless of the Unlike solar generation, the NPVs are negative for any BESS
BESS capacity when applying the BAS. The PCS capacity capacity when using wind generation. Although the BAS can
and E/P ratio that provide the highest NPV are 4.4 MW and improve the NPV from -862,341 to -827,723 USD, it remains
5.5 h, respectively. negative. Figure 10(b) shows that the BAS allows achieving

VOLUME 8, 2020 187739

 H. Shin, J. Hur: Optimal Energy Storage Sizing With Battery Augmentation for Renewable-Plus-Storage Power Plants

FIGURE 9. NPV, final SoH, total revenue, and total cost per combination
of PCS capacity and E/P ratio for 10 MW of wind generation.

FIGURE 10. NPV, final SoH, total revenue, and total cost per combination
of PCS capacity and E/P ratio for 10 MW of wind generation when using
BAS.

FIGURE 8. Annual cash flows from solar-plus-storage power plant.

the final SoHs higher than 80% for every combination of

PCS capacities and E/P ratios. Table 4 shows that the optimal
BESS capacity for wind generation is notably smaller than
that for solar generation. When the BAS is not considered,
the algorithm retrieves a 2400 kW PCS and 3.4-hour battery
storage, totaling 8160 kWh. If the BAS is included, the algo-
rithm retrieves two battery augmentations where batteries
of 1418 and 1126 kWh are added in years 6 and 11, respec-
tively. The total installed battery capacity is 10,704 kWh
(8160 kWh + 1418 kWh + 1126 kWh).
This low economic feasibility is attributed to two factors:
1) the probabilistic characteristics of wind generation during FIGURE 11. Empirical probability density functions for solar and wind
the peak generation period and 2) the high variability in the output during peak generation periods.
duration of the peak and non-peak wind generation periods.
Figure 11 shows the histograms of solar and wind power Using a small PCS and a large battery may be economi-
outputs during the corresponding peak generation periods. cally beneficial for wind generation because the wind power

187740 VOLUME 8, 2020

H. Shin, J. Hur: Optimal Energy Storage Sizing With Battery Augmentation for Renewable-Plus-Storage Power Plants

which is approximately 6.3:1. Moreover, the distribution of

solar generation is less skewed than that of wind generation,
as shown in Figure 11. Therefore, installing a medium-sized
PCS is economically advantageous under solar generation.
Regarding the battery size, a long-duration BESS is required
to efficiently collect the solar generation, mostly during the
peak-generation periods.
Overall, the balanced durations of peak and non-peak gen-
eration periods and the balanced solar power distribution
enable BESS efficiency and profitability. In addition, we con-
sider that under the time-variant energy pricing, it is econom-
ically beneficial for the BESS operator to stipulate balanced
durations of the peak and non-peak generation periods in the
PPA.
Figure 12 shows the cash flows for the wind-plus-storage
power plant with the PCS and battery capacities listed
in Table 4. The revenues from BESS discharging are con-
siderably smaller than those from directly supplying wind
power to the grid due to the low BESS capacity. Neverthe-
less, the economic benefits of the BAS for solar generation,
including the reduced initial costs, expected future reduction
of battery costs, and increased revenue, remain valid for wind
generation.

VI. CONCLUSION
We propose an approach to calculate the optimal capacity of a
BESS combined with solar and wind generation considering
the influence of lithium-ion battery degradation on the project
cash flows. For a given BESS capacity, the proposed method
first evaluates the SoH by applying the BESS operation algo-
rithm to then update the battery capacity retention, and finally
evaluate the annual cash flow. This process is repeated annu-
ally in the BESS sizing algorithm. In addition, we propose a
BAS to defer investments and leverage the expected reduction
of battery costs over time. The proposed BAS determines the
timing and capacity of batteries to be added based on the
predicted battery degradation rates.
We simulate the proposed algorithms considering South
FIGURE 12. Annual cash flows from wind-plus-storage power plant. Korean market rules and data. Integrating a BESS with solar
generation is profitable, whereas the NPV of integrating a
BESS with wind generation is expected to be negative. For
distribution is heavily skewed toward low outputs. However, a RES system of a given size, the optimal PCS and battery
the BESS must discharge power during non-peak generation capacities are considerably larger for solar generation than
periods to benefit from a high REC multiplier. As the duration for wind generation. These differences in project profitability
of the non-peak generation period for wind generation is and optimal sizes are attributed to the probabilistic charac-
substantially shorter than that of the peak generation period, teristics of RES power outputs and the different durations
the BESS power cannot be fully discharged during the former. of the peak and non-peak generation periods. The simula-
To prevent this problem, a larger PCS may be installed, but tions demonstrate that the proposed BAS can considerably
it may not provide an economic benefit given the skewness improve the NPV for both solar and wind generation. Nev-
of the wind power distribution. Therefore, both the PCS and ertheless, the NPV remains negative for wind-plus-storage
battery should have low capacity for wind-plus-storage power power plants.
plants. The proposed approach can provide a comprehensive
Regarding solar generation, the durations of the peak and framework for the parties involved in a BESS project,
non-peak generation periods are more balanced. In fact, including project developers, engineering, procurement and
the ratio between these durations is approximately 1:3, unlike construction providers, and independent power producers,
that between the corresponding durations of wind generation, to accurately determine the BESS sizes instead of simply

VOLUME 8, 2020 187741

 H. Shin, J. Hur: Optimal Energy Storage Sizing With Battery Augmentation for Renewable-Plus-Storage Power Plants

considering battery degradation regardless of the BESS oper- [13] F. Zhang, Y. Song, and Z. Hu, ‘‘Mixed-integer linear model for transmis-
ation and capacity. In addition, we introduce a systematic sion expansion planning with line losses and energy storage systems,’’ IET
Gener., Transmiss. Distrib., vol. 7, no. 8, pp. 919–928, Aug. 2013.
BAS that can be easily integrated into the BESS sizing algo- [14] T. Qiu, B. Xu, Y. Wang, Y. Dvorkin, and D. S. Kirschen, ‘‘Stochastic
rithm. The simulation results show that an efficient BESS multistage coplanning of transmission expansion and energy storage,’’
depends on balanced peak and non-peak generation peri- IEEE Trans. Power Syst., vol. 32, no. 1, pp. 643–651, Jan. 2017.
[15] X. Tang, W. Deng, and Z. Qi, ‘‘Investigation of the dynamic stability
ods for the pricing rules of the PPA, leading to improved of microgrid,’’ IEEE Trans. Power Syst., vol. 29, no. 2, pp. 698–706,
NPVs. The proposed sizing approach can be adapted to other Mar. 2014.
applications by, for example, modifying the BESS operation [16] S. X. Chen, H. B. Gooi, and M. Q. Wang, ‘‘Sizing of energy storage
for microgrids,’’ IEEE Trans. Smart Grid, vol. 3, no. 1, pp. 142–151,
algorithm and calculation of revenues and costs. Mar. 2012.
The proposed BAS is a suboptimal method. To achieve [17] H. Khorramdel, J. Aghaei, B. Khorramdel, and P. Siano, ‘‘Optimal battery
optimality, we should consider complex multistage optimiza- sizing in microgrids using probabilistic unit commitment,’’ IEEE Trans.
Ind. Informat., vol. 12, no. 2, pp. 834–843, Apr. 2016.
tion incorporating the effect of augmented batteries on the [18] E. Nasrolahpour, S. J. Kazempour, H. Zareipour, and W. D. Rosehart,
performance of the whole BESS. In future work, we will ‘‘Strategic sizing of energy storage facilities in electricity markets,’’ IEEE
develop an advanced BAS to find approximate solutions to Trans. Sustain. Energy, vol. 7, no. 4, pp. 1462–1472, Oct. 2016.
[19] P. Harsha and M. Dahleh, ‘‘Optimal management and sizing of energy
the optimization problem while maintaining moderate com- storage under dynamic pricing for the efficient integration of renew-
plexity. Additionally, the optimal sizing of BESS for the able energy,’’ IEEE Trans. Power Syst., vol. 30, no. 3, pp. 1164–1181,
provision of multiple services is another interesting topic. For May 2015.
[20] R. Arghandeh, J. Woyak, A. Onen, J. Jung, and R. P. Broadwater, ‘‘Eco-
this purpose, it may be necessary to adopt decision making nomic optimal operation of community energy storage systems in compet-
under uncertainty such as stochastic optimization and Markov itive energy markets,’’ Appl. Energy, vol. 135, pp. 71–80, Dec. 2014.
decision processes. [21] K. Bradbury, L. Pratson, and D. Patiño-Echeverri, ‘‘Economic viability of
energy storage systems based on price arbitrage potential in real-time U.S.
electricity markets,’’ Appl. Energy, vol. 114, pp. 512–519, Feb. 2014.
REFERENCES [22] H. Thu Le and T. Quang Nguyen, ‘‘Sizing energy storage systems for wind
power firming: An analytical approach and a cost-benefit analysis,’’ in
[1] N. Günter and A. Marinopoulos, ‘‘Energy storage for grid services and Proc. IEEE Power Energy Soc. Gen. Meeting-Convers. Del. Electr. Energy
applications: Classification, market review, metrics, and methodology for 21st Century, Jul. 2008, pp. 1–8.
evaluation of deployment cases,’’ J. Energy Storage, vol. 8, pp. 226–234, [23] M. Dicorato, G. Forte, M. Pisani, and M. Trovato, ‘‘Planning and operating
Nov. 2016. combined wind-storage system in electricity market,’’ IEEE Trans. Sustain.
[2] Greentech Media. Just How Much Business Can Batteries Take Energy, vol. 3, no. 2, pp. 209–217, Apr. 2012.
From Gas Peakers?. Accessed: Apr. 23, 2020. [Online]. Available: [24] J. Galtieri and P. T. Krein, ‘‘Solar variability reduction using off-maximum
https://www.greentechmedia.com/articles/read/just-how-much-business- power point tracking and battery storage,’’ in Proc. IEEE 44th Photovoltaic
can-batteries-take-from-gas-peakers Specialist Conf. (PVSC), Jun. 2017, pp. 3214–3219.
[3] P. L. Denholm, R. M. Margolis, and J. D. Eichman, ‘‘Evaluating the [25] N. Müller, S. Kouro, P. Zanchetta, P. Wheeler, G. Bittner, and F. Girardi,
technical and economic performance of PV plus storage power plants,’’ ‘‘Energy storage sizing strategy for grid-tied PV plants under power clip-
Nat. Renew. Energy Lab., Golden, CO, USA, Tech. Rep. NREL/TP-6A20- ping limitations,’’ Energies, vol. 12, no. 9, p. 1812, May 2019.
68737, 2017. [26] M. Yue and X. Wang, ‘‘Grid inertial response-based probabilistic deter-
[4] M. Bolinger, J. Seel, and D. Robson, ‘‘Utility-scale solar: Empirical trends mination of energy storage system capacity under high solar penetration,’’
in project technology, cost, performance, and PPA pricing in the United IEEE Trans. Sustain. Energy, vol. 6, no. 3, pp. 1039–1049, Jul. 2015.
States–2019 edition,’’ Lawrence Berkeley Nat. Lab., Berkeley, CA, USA, [27] P. Li, R. Dargaville, F. Liu, J. Xia, and Y.-D. Song, ‘‘Data-based statis-
Tech. Rep., Dec. 2019. tical property analyzing and storage sizing for hybrid renewable energy
[5] M. R. Palacín, ‘‘Understanding ageing in li-ion batteries: A chemi- systems,’’ IEEE Trans. Ind. Electron., vol. 62, no. 11, pp. 6996–7008,
cal issue,’’ Chem. Soc. Rev., vol. 47, no. 13, pp. 4924–4933, 2018, Nov. 2015.
doi: 10.1039/C7CS00889A. [28] G. Ning and B. N. Popov, ‘‘Cycle life modeling of lithium-ion batteries,’’
[6] A. Barré, B. Deguilhem, S. Grolleau, M. Gérard, F. Suard, and D. Riu, J. Electrochem. Soc., vol. 151, no. 10, p. A1584, 2004.
‘‘A review on lithium-ion battery ageing mechanisms and estimations [29] R. Spotnitz, ‘‘Simulation of capacity fade in lithium-ion batteries,’’
for automotive applications,’’ J. Power Sources, vol. 241, pp. 680–689, J. Power Sources, vol. 113, no. 1, pp. 72–80, Jan. 2003.
Nov. 2013. [30] J. Vetter, P. Novák, M. R. Wagner, C. Veit, K. C. Möller, J. O. Besenhard,
[7] P. Mercier, R. Cherkaoui, and A. Oudalov, ‘‘Optimizing a battery energy M. Winter, M. Wohlfahrt-Mehrens, C. Vogler, and A. Hammouche, ‘‘Age-
storage system for frequency control application in an isolated power sys- ing mechanisms in lithium-ion batteries,’’ J. Power Sources, vol. 147,
tem,’’ IEEE Trans. Power Syst., vol. 24, no. 3, pp. 1469–1477, Aug. 2009. nos. 1–2, pp. 269–281, 2005.
[31] O. Erdinc, B. Vural, and M. Uzunoglu, ‘‘A dynamic lithium-ion battery
[8] V. Knap, S. K. Chaudhary, D.-I. Stroe, M. Swierczynski, B.-I. Craciun,
model considering the effects of temperature and capacity fading,’’ in Proc.
and R. Teodorescu, ‘‘Sizing of an energy storage system for grid inertial
Int. Conf. Clean Electr. Power, Jun. 2009, pp. 383–386.
response and primary frequency reserve,’’ IEEE Trans. Power Syst., vol. 31,
[32] D.-I. Stroe, M. Swierczynski, A.-I. Stroe, R. Laerke, P. C. Kjaer, and
no. 5, pp. 3447–3456, Sep. 2016.
R. Teodorescu, ‘‘Degradation behavior of lithium-ion batteries based on
[9] Y. J. Zhang, C. Zhao, W. Tang, and S. H. Low, ‘‘Profit-maximizing plan- lifetime models and field measured frequency regulation mission profile,’’
ning and control of battery energy storage systems for primary frequency IEEE Trans. Ind. Appl., vol. 52, no. 6, pp. 5009–5018, Nov. 2016.
control,’’ IEEE Trans. Smart Grid, vol. 9, no. 2, pp. 712–723, Mar. 2018. [33] D.-I. Stroe, V. Knap, M. Swierczynski, A.-I. Stroe, and R. Teodorescu,
[10] J. Xiao, L. Bai, Z. Zhang, and H. Liang, ‘‘Determination of the opti- ‘‘Operation of a grid-connected lithium-ion battery energy storage system
mal installation site and capacity of battery energy storage system in for primary frequency regulation: A battery lifetime perspective,’’ IEEE
distribution network integrated with distributed generation,’’ IET Gener., Trans. Ind. Appl., vol. 53, no. 1, pp. 430–438, Jan. 2017.
Transmiss. Distrib., vol. 10, no. 3, pp. 601–607, Feb. 2016. [34] X. Hu, H. Yuan, C. Zou, Z. Li, and L. Zhang, ‘‘Co-estimation of state of
[11] X. Wang, M. Yue, E. Muljadi, and W. Gao, ‘‘Probabilistic approach for charge and state of health for lithium-ion batteries based on fractional-order
power capacity specification of wind energy storage systems,’’ IEEE Trans. calculus,’’ IEEE Trans. Veh. Technol., vol. 67, no. 11, pp. 10319–10329,
Ind. Appl., vol. 50, no. 2, pp. 1215–1224, Mar. 2014. Nov. 2018.
[12] M. Motalleb, E. Reihani, and R. Ghorbani, ‘‘Optimal placement and sizing [35] X. Hu, F. Feng, K. Liu, L. Zhang, J. Xie, and B. Liu, ‘‘State estimation for
of the storage supporting transmission and distribution networks,’’ Renew. advanced battery management: Key challenges and future trends,’’ Renew.
Energy, vol. 94, pp. 651–659, Aug. 2016. Sustain. Energy Rev., vol. 114, Oct. 2019, Art. no. 109334.

187742 VOLUME 8, 2020

H. Shin, J. Hur: Optimal Energy Storage Sizing With Battery Augmentation for Renewable-Plus-Storage Power Plants

[36] B. Xu, A. Oudalov, A. Ulbig, G. Andersson, and D. S. Kirschen, ‘‘Mod- HUNYOUNG SHIN (Member, IEEE) received the
eling of lithium-ion battery degradation for cell life assessment,’’ IEEE B.S. degree in radio and communication engineer-
Trans. Smart Grid, vol. 9, no. 2, pp. 1131–1140, Mar. 2018. ing and the M.S. degree in electrical engineer-
[37] Z. Zhang, C. Guan, and Z. Liu, ‘‘Real-time optimization energy man- ing from Korea University, Seoul, South Korea,
agement strategy for fuel cell hybrid ships considering power sources and the Ph.D. degree from the Department of
degradation,’’ IEEE Access, vol. 8, pp. 87046–87059, 2020. Electrical and Computer Engineering, The Uni-
[38] A. Maheshwari, N. G. Paterakis, M. Santarelli, and M. Gibescu, versity of Texas at Austin, Austin, TX, USA, in
‘‘Optimizing the operation of energy storage using a non-linear
December 2017. In 2018, he joined LG CNS as
lithium-ion battery degradation model,’’ Appl. Energy, vol. 261,
a Managing Consultant and participated in energy
Mar. 2020, Art. no. 114360. [Online]. Available: http://www.sciencedirect.
com/science/article/pii/S0306261919320471 storage system planning projects. He is currently
[39] M. Lee, J. Park, S.-I. Na, H. S. Choi, B.-S. Bu, and J. Kim, ‘‘An analysis an Assistant Professor with Sangmyung University, Seoul. His research
of battery degradation in the integrated energy storage system with solar interests include optimization of power systems, risk-hedging strategies, and
photovoltaic generation,’’ Electronics, vol. 9, no. 4, p. 701, Apr. 2020. pricing rule in electricity markets.
[Online]. Available: https://www.mdpi.com/2079-9292/9/4/701
[40] Y. Zhou, H. Obeid, S. Laghrouche, M. Hilairet, and A. Djerdir, ‘‘A dis-
turbance rejection control strategy of a single converter hybrid electri-
cal system integrating battery degradation,’’ Energies, vol. 13, no. 11,
p. 2781, Jun. 2020. [Online]. Available: https://www.mdpi.com/1996-
1073/13/11/2781
[41] C. Zhang, D. Wang, B. Wang, and F. Tong, ‘‘Battery degradation
minimization-oriented hybrid energy storage system for electric vehi-
cles,’’ Energies, vol. 13, no. 1, p. 246, Jan. 2020. [Online]. Available:
https://www.mdpi.com/1996-1073/13/1/246
[42] J. Kong, S. T. Kim, B. O. Kang, and J. Jung, ‘‘Determining the size of
energy storage system to maximize the economic profit for photovoltaic
and wind turbine generators in South Korea,’’ Renew. Sustain. Energy Rev.,
vol. 116, Dec. 2019, Art. no. 109467.
[43] S. Downing and D. Socie, ‘‘Simple rainflow counting algorithms,’’ Int.
J. Fatigue, vol. 4, no. 1, pp. 31–40, Jan. 1982.
[44] E. Peled, ‘‘The electrochemical behavior of alkali and alkaline earth metals
in nonaqueous battery systems—The solid electrolyte interphase model,’’
J. Electrochem. Soc., vol. 126, no. 12, pp. 2047–2051, 1979.
[45] P. Pinson, G. Papaefthymiou, B. Klockl, and J. Verboomen, ‘‘Dynamic
sizing of energy storage for hedging wind power forecast uncertainty,’’ in
Proc. IEEE Power Energy Soc. Gen. Meeting, Jul. 2009, pp. 1–8.
[46] K. Baker, G. Hug, and X. Li, ‘‘Energy storage sizing taking into account
forecast uncertainties and receding horizon operation,’’ IEEE Trans. Sus-
tain. Energy, vol. 8, no. 1, pp. 331–340, Jan. 2017. JIN HUR (Senior Member, IEEE) received the
[47] K. Mongird, V. Viswanathan, P. Balducci, J. Alam, V. Fotedar, V. Koritarov, B.S. and M.S. degrees in electrical engineer-
and B. Hadjerioua, ‘‘Energy storage technology and cost characteriza- ing from Korea University, Seoul, South Korea,
tion report,’’ Pacific Northwest Nat. Lab., (PNNL), Richland, WA, USA,
in 1997 and 1999, respectively, and the Ph.D.
Tech. Rep. PNNL-28866, 2019.
degree in electrical and computer engineering
[48] J. Leadbetter and L. G. Swan, ‘‘Selection of battery technology to sup-
port grid-integrated renewable electricity,’’ J. Power Sources, vol. 216, from The University of Texas at Austin, in 2012.
pp. 376–386, Oct. 2012. He is currently an Associate Professor with the
[49] ‘‘Energy storage system costs survey 2018,’’ Bloomberg New Energy Department of Climate and Energy Systems Engi-
Finance, New York, NY, USA, Tech. Rep., 2018. neering, Ewha Womans University. His research
[50] B. Xu, ‘‘Degradation-limiting optimization of battery energy storage sys- interest includes integrate high level of variable
tems operation,’’ M.S. thesis, Dept. Inf. Technol. Elect. Eng., ETH Zurich, generating resources into electric power systems.
Zürich, Switzerland, 2013.

VOLUME 8, 2020 187743