Sustainable Energy Technologies and Assessments 22 (2017) 55–64

Contents lists available at ScienceDirect

Sustainable Energy Technologies and Assessments

journal homepage: www.elsevier.com/locate/seta

Original article

Dynamic power allocation of battery-supercapacitor hybrid energy

storage for standalone PV microgrid applications
Wenlong Jing a,⇑, Chean Hung Lai a, Wallace S.H. Wong a, M.L. Dennis Wong b
a
Faculty of Engineering, Computing and Science, Swinburne University of Technology Sarawak Campus, Jalan Simpang Tiga, 93350 Kuching, Sarawak, Malaysia
b
School of Engineering and Physical Sciences, Heriot-Watt University (Malaysia Campus), Putrajaya, Malaysia

a r t i c l e i n f o a b s t r a c t

Article history: Standalone photovoltaic-based microgrid with energy storage system could be a promising solution for
Received 21 November 2016 powering up off-grid communities. One of the major issues that hinder the development of standalone
Revised 27 March 2017 microgrids is the poor service life of the batteries. To address this issue, hybrid energy storage systems
Accepted 7 July 2017
(HESS) and novel power management strategies have been proposed by researchers to enhance the ser-
vice life of battery bank. This paper presents a novel multi-level hybrid energy storage system topology
and its associated power management strategy to mitigate the charge/discharge stress on battery. Matlab
Keywords:
Simulink model of typical standalone PV microgrid with different HESS topologies are developed to eval-
Photovoltaic
Microgrid
uate the performance of the proposed system. Actual solar irradiance data with different weather condi-
Renewable energy tions and estimated load profile based on site survey results are used to analyze the effectiveness of
Hybrid energy storage different HESS in mitigating stress on batteries. A comprehensive analysis and benchmarking is presented
Energy management to compare the technical and financial viability of the proposed system with existing HESS solutions.
Rural electrification Simulation results show that the proposed HESS can improve the life expectancy of the battery and
reduce the operating cost of the standalone PV-battery microgrid.
Ó 2017 Elsevier Ltd. All rights reserved.

Introduction Fig. 1 illustrates a typical output power profile of solar energy in

Sarawak and an estimated energy consumption pattern of rural
The greenhouse effect is well understood via many research communities. The relatively low capacity load demand and inter-
works and its increasing emissions are leading to a global environ- mittent nature of solar energy cause extreme fluctuation in power
mental deterioration in recent years [1]. To reverse these undesir- generation and load, leading to severe imbalance in the power net-
able scenarios, as a replacement to traditional fossil energy work. Therefore, energy storage system (ESS) is generally required
sources, renewable energy sources could play an important role in renewable-based microgrid system to act as a buffer between
in the future power systems [2]. According to [3], nearly 1.2 billion generation and load [4]. ESS increases energy utilization by miti-
people of the global population lacked electricity till to 2016 and it gating the temporal mismatch between PV power generation and
will be projected to decline to around 12 percent by 2030. More load demand by storing surplus energy during peak power gener-
than 80 percent of these people live in remote rural areas without ation period and provide the stored energy to the load during gen-
reliable electricity supply. Standalone microgrid with renewable eration off-peak [5]. In addition, ESS in PV Microgrid also plays an
energy sources could be an attractive solution to address the elec- important role in compensating the reactive power as well as sup-
tricity needs in these remote communities. Currently, the major pressing the voltage fluctuation and flicker [6].
sources of renewable energy include wind, photovoltaic (PV), Among many ESS devices available on the market, lead acid (LA)
hydrogen fuel cells, tidal, and geothermal energy. Among these batteries have been the mainstream residential energy storage
renewable technologies, PV microgrid is one of the attractive solutions. LA battery is a matured and well understood energy
options for low density off-grid communities. For example, the storage technology which is robust and low cost. However, the
remote longhouses located at the inner region of Sarawak, main drawback of LA batteries is the short cycle life especially
Malaysia. when it is operated in cycling applications [7]. The main life-
limiting factors of LA battery such as high charge/discharge rate,
overcharge, high depth-of discharge and dynamic loading pattern
⇑ Corresponding author. are all commonly experienced in PV-based residential energy
E-mail address: [email protected] (W. Jing).

http://dx.doi.org/10.1016/j.seta.2017.07.001
2213-1388/Ó 2017 Elsevier Ltd. All rights reserved.
56 W. Jing et al. / Sustainable Energy Technologies and Assessments 22 (2017) 55–64

7 SC-battery HESS with multiple SC modules was presented in [16]

PV Power that demonstrated to effectively mitigate battery stress. The indi-
Load Profile vidually controlled SC module allows wide range of power require-
6
ments which increase the flexibility and efficiency. However, the
5 redundant SC modules increase the overall cost and complexity.
Other than SC-battery HESS, dual separate battery pack as the
Power (kW)

4 energy storage in which the batteries are charged by PV panels

separately via two solid-state switches [25,26].
3 This paper presents a novel multi-level HESS topology and its
associated power management system which is developed upon
2 the SC-battery hybridization and dual battery solution that include
SC and two battery modules of different chemistries and character-
1 istics. The low cost LA battery will be the primary ESS that makes
up the system capacity, while the Li-ion battery with higher price
0 tag but robust will be the secondary ESS that absorb medium to
0 4 8 12 16 20 0 high frequency fluctuation. The power distribution among them
Hours of Day is set via a scaling factor which can be varied in different scenarios.
SC on the other hand, is programmed to response to high frequency
Fig. 1. Typical PV output power and load profile in rural Sarawak, Malaysia. power exchange. A typical Standalone PV microgrid for rural elec-
trification is used to demonstrate the performance of the proposed
HESS scheme. Comparison between Battery-Only system and com-
storage system [8]. A typical LA battery may only last for hundreds mon SC-battery HESS topologies is presented and discussed.
of charge/discharge cycles which make it economically unsustain- The rest of the paper is organized as follows: Section II presents
able as the ESS being the main cost component in typical PV-based the configuration of the existing HESSs and the proposed HESS for
Microgrids [9]. standalone PV microgrid and the corresponding power manage-
To address these shortcomings, hybrid energy storage system ment strategies. Section III presents the numerical simulation
(HESS) has been proposed by many researchers [10–14]. HESS and thorough analysis and discussion. Section IV presents the eval-
exploits the strength of different ESS devices to compensate weak- uation and comparison of battery stress mitigation with a novel
nesses of homogenous storage element. Many research studies battery health cost function. Analysis will be extended to evaluate
have demonstrated that HESS could reduce the internal system the economic viability of the different HESS topologies. Lastly, Sec-
losses and relieve the battery charging/discharging stress which tion V concludes the paper.
leads to prolonged battery life [15–18]. Glavin et. al. have verified
that SC-battery HESS performs better than battery-alone energy SC-battery HESS topologies
storage system in standalone PV microgrids [19]. Dougal et al.
mathematically proved that the HESS is capable of extending the In SC-battery HESS, the two ESS devices can be coupled to either
battery lifetime [20]. Anthony et al. used a small-scale wind energy a common DC or AC bus. For standalone renewable microgrid,
prototype system to demonstrate the improvement in battery life- common DC bus is the preferred choice because majority of renew-
time when using SCs [15]. able energy sources operate in DC and no synchronization is
The main idea of HESS is to direct the damaging loading condi- needed which minimizes the system complexity [27]. The follow-
tions such as severe fluctuation in charge/discharge current to a ing sections review the different SC-battery HESS topologies in
more robust ESS device. Ideally, the primary battery bank should common DC bus settings.
be responding to the averaged energy exchange due to the mis-
match of the generation and demand. To achieve the desired power
Battery-only ESS
allocation to different ESS elements, actively controlled bidirec-
tional DC/DC converters are required alongside with the optimally
Conventionally, a typical DC-coupled PV microgrid is configured
designed energy management system (EMS). In general, there are
as shown in Fig. 2. The battery bank sits directly between the gen-
two objectives when designing the EMS for power sharing among
eration and load to create a relatively stable DC bus [28]. This sim-
different ESS elements: (1) to improve power quality and energy
ple structure regulates the imbalance PV generation and load
efficiency and (2) to prolong the lifetime of ESS [21].
demand through charging/discharging the battery bank. This con-
Among many HESS configurations, SC-battery HESS have been
servative system performs efficiently under stable generation and
one of the most popular combinations in microgrid applications
demand, however fluctuation in generation and load are often
[16,22–24]. Bidirectional dual active bridge DC/DC converters were
used to enable dynamic power allocation to the battery bank and
SC respectively [22]. The authors claimed that the HESS permits
online battery replacement and allows changing of the battery con- Controller

figuration without disturbing the system normal operation. How-

ever, this topology requires a large number of individual DC/DC
converters that dramatically increase the system complexity and DC Bus
DC/DC Load
overall cost. Kollimalla et. al. presented an actively controlled SC-
battery HESS topology with novel EMS that monitors the
MPPT
demand-generation mismatch and utilizes the error component PV Arrays
of battery current to control the SC power flow [23]. This control
strategy relies heavily on the characteristics of energy storage Battery
devices and lacks of interactions among the HESS and other com-
ponents in microgrid. Thus, the control strategy tends to applica-
tion oriented which may not be applicable to other systems. A Fig. 2. PV microgrid with battery-only ESS topology.
 W. Jing et al. / Sustainable Energy Technologies and Assessments 22 (2017) 55–64 57

the case for PV microgrid. As a result, the battery has to absorb/- ðVco  Vb Þ ðR þRt c ÞC
ib ðtÞ ¼  e b
supply dynamically at all times, putting additional stress on the Rb þ Rc
battery which could potentially accelerate cycle life. 2  tkT
 3
X
N1 1  R RþR
b
e ðRb þRc ÞC £ðt  kTÞ
6 7
þ Io 4  
b c
tðkþDÞT
ðR þR ÞC 5 ð1Þ
Passive SC-battery HESS 1  R RþR
b
e b c £ðt  ðk þ DÞTÞ
k¼0 c
b

In recent years, researchers have been actively investigating in

the SC-battery hybridization for standalone PV microgrid ic ðtÞ ¼ i0 ðtÞ  ib ðtÞ ð2Þ
[29–31]. It is reported that by hybridizing SC and battery, the
charge/discharge stress on battery can be mitigated by allocating where V co is the SC initial voltage, I0 is the peak current of the
the highly dynamic power exchange to the SC. The simplest form pulsed load, T is the period of the pulsed load and D is the duty cycle
of SC-Battery hybrid energy storage system is to passively connect of the pulsed load and io(t) is the output current. From Eq. (1), the
the SC and battery in parallel as illustrated in Fig. 3. power sharing between SC and battery is determined by the inter-
This passively connected HESS does not require any form of nal resistances Rb and Rc as well as the capacity C of the SC. Thus, the
active control mechanism, thus reducing the implementation cost power allocation in passive HESS will be fixed throughout the entire
and system complexity. However, as the terminal of SC and battery operation of microgrid. Fig. 4(b) shows an example of the power
are sharing the same DC bus, the configuration of HESS must be sharing in passive HESS under periodic pulsed load. As can be seen
carefully designed to avoid terminal voltage mismatch. In addition, from the simulation result in Section IV, the SC in passive HESS is
the distribution of load current is purely determined by the inter- under-utilized as a result of sharing identical terminal voltage with
nal resistances and capacitance of the two ESSs. As presented in battery.
[32], the passively connected SC and battery can be modeled as
an equivalent circuit model as shown in Fig. 4(a). The SC is gener- Parallel active SC-battery HESS
ally modelled as a large capacitance C and an equivalent series
resistance Rc, while the battery is modelled as an ideal voltage To address the low level of controllability in passively con-
source Vb with an equivalent series resistance Rb. nected HESS, interfacing the ESSs using bidirectional DC/DC con-
The branch currents of the battery ib(t) and the SC ic(t) to pulsed verters allow the power exchange of the distinct ESS elements to
load can be derived as [20]: be fully controllable. The isolation of ESS from DC bus offers greater
flexibility in HESS configuration, better volumetric efficiency and
cycle life [33,34]. Unlike passive HESS, the performance and useful-
ness of actively controlled HESS relies heavily on the power alloca-
tion and control strategy [35]. Therefore, a carefully designed
Controller control strategy is essential to justify the extra cost of active
components.
DC Bus Fig. 5 depicts a parallel connected active HESS topology where
DC/DC Load both SC and battery are connected in parallel to DC bus via bidirec-
MPPT
tional DC/DC converters. In general, the battery as high energy den-
PV Arrays sity ESS is programmed to meet the low-frequency power variation
throughout the day, while the SC is controlled to response to the
Supercapacitor Battery high-frequency power exchange or to regulate the DC bus voltage.
Fig. 6 shows a typical power allocation and control strategy of
parallel active HESS. Low-pass filtering (LPF) and moving average
Fig. 3. PV microgrid with passive SC-battery HESS topology. smoothing method (MASM) are commonly used power allocation

100

T D = Ton / T Load
80 D = Ton / T Battery
SC
io IO
60
ic ib
40
Current (A)

Rc Rb 20
vo
0
C vb
-20

-40
Ton
-60
30 32 34 36 38 40 42 44 46 48 50
Time (s)

( a) (b )
Fig. 4. (a) Equivalent circuit model of passive SC-battery HESS topology; (b) Power sharing of passive HESS under periodical pulsed load.
58 W. Jing et al. / Sustainable Energy Technologies and Assessments 22 (2017) 55–64

Controller DC Bus
Controller

DC Bus
DC/DC Load DC/DC Load

PV Arrays MPPT
PV Arrays MPPT
Primary
Bidirectional Bidirectional
Supercapacitor Battery
DC/DC DC/DC
Modules
Bidirectional Bidirectional
Supercapacitor Battery
DC/DC DC/DC
Power
Management
System
Power
Management
Secondary
System Diesel Bidirectional
Diesel Controller Battery
Generator (AC/DC) DC/DC
Modules
Fig. 5. PV microgrid with parallel active SC-battery HESS topology.
Fig. 7. Topology of the proposed multi-level HESS in standalone PV microgrid.

+ PHESS LPF / In addition, it is possible to provide a smoother IBatt (for primary

Ppv Pbatt(ref)
- MASM
battery) with the same capacity of SC compared to conventional
-
Pload PSC(ref) parallel active HESS (see Fig. 5). A diesel generator as backup
+
energy source is integrated to the ESS system in case of extreme
Power Allocation Strategy
conditions such as low battery state-of-charge due to low PV gen-
eration in rainy days or when there is big event going on in the
IBatt(ref) + D Switching Signal
PBatt(ref) 1/VBatt PI PWM community, for example during annual festival. The diesel genera-
- (Battery converter)
tor is controlled automatically by the power management system
IBatt and is designed to charge the primary battery bank when the bat-
tery state-of-charge is below 40 percent to avoid deep discharge.
ISC(ref) + D Switching Signal
PSC(ref) 1/VSC PI PWM Conversely, the diesel generator can also be operated manually
- (SC converter)
by users when a heavy consumption is expected. The analysis in
ISC this work assumes that the community keeps sufficient petrol fuel
in reserve so that the electricity supply is guaranteed.
Fig. 6. Typical power allocation strategy of parallel active SC-battery HESS Fig. 8 shows the proposed multi-level power allocation strategy.
topology. Unlike conventional linear filtering or MASM approaches, the
power requirement of the HESS PHESS is decomposed into three fre-
quency bands using two simple low-pass filters. The low frequency
strategies to decompose the low frequency component of the
part of the power requirement will be used as the reference signal
power demand PHESS to mitigate battery stress [36,37].
Pbatt_1(ref) to command the primary battery, the medium frequency
The power reference signals (PBatt and PSC) will pass through
power exchange Pbatt_2(ref) will be satisfied by the secondary bat-
respective control loops with PI controller to correct the instanta-
tery, and the high frequency fluctuation PSC(ref) will be delivered
neous battery current IBatt and SC current ISC by adjusting the duty
by the SC. Since a certain portion of the battery bank will be
cycle (D) of PWM signals. When selecting the bandwidth of the LPF
relocated from the primary battery, same portion of the average
or window size of the MASM, a tradeoff exists between smooth-
power requirement (low frequency) will also be allocated to
ness of battery current IBatt and SC size. For example, having a
low cut-off frequency in LFP or extremely large window size in
MASM generates smoothed IBatt but requires extremely large SC Weight Factor
+ PHESS LPF
capacity and power rating of DC/DC converter to compensate the Ppv (Primary)
W1 Pbatt_1(ref)
-
sluggish battery response. + -
Pload
LPF
Proposed multi-level HESS (Secondary)
Pbatt_2(ref)
-
PSC(ref)
+
SC-battery HESS has been demonstrated to significantly miti-
Proposed Multi-Layer
gate battery stress under dynamic power exchange in renewable Power Allocation Strategy
microgrid applications. However, technical and financial limita- Switching Signal
IBatt(ref) + D
tions such as SC’s size and converter’s power rating suppress what Pbatt_1(ref) 1/Vbatt_1 PI PWM (Primary Battery
-
the HESS can potentially offer. In this work, a novel multi-level DC/DC Converter)
HESS topology and its power allocation strategy is proposed to Ibatt_1
address the abovementioned issues. IBatt(ref) + D Switching Signal
Fig. 7 illustrates the proposed multi-level SC-battery HESS Pbatt_2(ref) 1/Vbatt_2 PI PWM (Secondary Battery
-
topology for PV based microgrid. Unlike the centralized control DC/DC Converter)

of battery bank in conventional HESS settings, the proposed Ibatt_2

multi-level HESS splits the single battery bank into primary battery ISC(ref) + D Switching Signal
PSC(ref) 1/VSC PI PWM (SC DC/DC
and secondary battery. The secondary battery generally holds a -
Converter)
small portion of the overall battery capacity and it can be make
ISC
up from different type of battery, for example Li-ion battery. The
higher level of hybridization allows greater flexibility in power Fig. 8. Power allocation strategy for the proposed multi-level HESS in atandalone
sharing among different ESS elements within the HESS. PV microgrid.
 W. Jing et al. / Sustainable Energy Technologies and Assessments 22 (2017) 55–64 59

the secondary battery via a scaling factor W1. In different scenarios, 0.95 and the secondary battery of the proposed HESS will con-
the scaling factor can be set manually or varied via advanced con- tribute 5 percent (50Ah) to the overall energy storage capacity.
trol algorithm to face the different specific conditions. Since the main focus of this work is to evaluate the effectiveness
of the proposed HESS in mitigating battery stress, the SOC control
Numerical analysis and case study of the batteries and SC are not implemented and the SOCs of
battery and SC are assumed to operate in acceptable range.
Fig. 9 shows the Matlab Simulink model of the proposed multi- The analysis will be carried out under different scenarios using
level HESS. The system parameters and simulation conditions used actual solar irradiance data and estimated load profiles in rural Sar-
in the analysis are tabulated in Table 1. Three bidirectional awak. Fig. 10 shows the simulated power profile of a 5KW PV array
buck-boost DC/DC converters are employed to actively control for a typical sunny and cloudy day recorded in Sarawak, Malaysia.
the current to/flow from the primary battery, secondary battery On the other hand, the load profiles used in the simulation anal-
and SC to the common DC bus. The time constants of the respective ysis were estimated based on actual survey data collected from a
low pass filter are set to decompose the PHESS to low frequency rural community in inner part of Sarawak (1°140 20.500 N, 112°020 1
Pbatt_1(ref) for primary LA battery bank and medium frequency 0.700 E). The targeted rural site consists of 6 households and the
Pbatt_2(ref) for secondary Li-ion battery bank. The remaining high common electrical appliances are lighting, television, common
frequency power exchange will be used as a reference signal to refrigerator, electric fan and simple electronic devices. Fig. 11(a)
control the SC module Psc(ref). presents normal daily electricity consumption, while Fig. 11(b)
Li-ion battery type was selected to form the secondary battery shows a heavy electricity consumption which could be family
due to its superior performance, such as longer cycle life, higher day or festive season.
c-rate capability and wider operating SOC range as compared to To evaluate and benchmark the performance of the proposed
LA battery. These characteristics is desirable for handling the med- multi-level HESS and existing HESS schemes with conventional
ium frequency fluctuation in power exchange and the higher cost battery-only setting, three scenarios are considered as shown in
of Li-ion battery make it a perfect option for the small capacity sec- Fig. 12: (a) sunny day with normal electricity consumption, (b)
ondary battery proposed. In this paper, the scaling factor is set as sunny day with heavy electricity consumption, and (c) cloudy

Fig.9. Matlab Simulink model of the proposed HESS.

Table 1
System parameters of the Matlab Simulink model.

Parameter Battery-Only Passive HESS Active HESS Proposed HESS

PV array peak power (kWp) 5
Daily load energy consumption (kWh) 27.4
Battery nominal voltage (V) 48
Battery capacity (Primary) (Ah) 1000 1000 1000 950
Battery capacity (Secondary) (Ah) – – – 50
Battery internal resistance (ohm) 0.005 0.005 0.005 0.005
SC capacitance (F) – 1000 1000 1000
SC equivalent series resistance (ohm) – 0.001 0.001 0.001
Scaling factor W1 – – – 0.95
Time constant (Primary LPF) (sec) – – – 600
Time constant (Secondary LPF) (sec) – – – 300
60 W. Jing et al. / Sustainable Energy Technologies and Assessments 22 (2017) 55–64

5 5

4 4

Power (kW)
Power (kW)

3 3

2 2

1 1

0 0
0 4 8 12 16 20 0 0 4 8 12 16 20 0
Hours of Day Hours of Day

(a) Sunny day (b) Cloudy day

Fig. 10. 5 kW PV generation profile of typical weather conditions in Sarawak, Malaysia.

3 3.5
Community Hall Additional Appliances
Total Consumption 3 Community Hall
2.5 Household Use Total Consumption
Common Lighting Common Lighting
2.5 Household Use
2
Power (kW)

Power (kW)

2
1.5
1.5

1
1

0.5
0.5

0 0
0 6 12 18 0 0 6 12 18 0
Hours of Day Hours of Day
(a) Normal electricity consumption (b)Heavy electricity consumption

Fig. 11. Estimated load profiles of the target rural site in Sarawak, Malaysia.

6 6 6
(a) Normal Load Profile (a) Heavy Load Profile (a) Normal Load Profile
(b) Sunny PV Output (b) Sunny PV Output (b) Cloudy PV Output
5 5 5
(b)
(b) (b)
4 4 4
Power (kW)

Power (kW)

Power (kW)

3 3 3

2 2 2

(a) (a)
1 (a)
1 1

0 0 0
0 4 8 12 16 20 0 0 4 8 12 16 20 0 0 4 8 12 16 20 0
Hours of Day Hours of Day Hours of Day

(a) Sunny Day with Normal Load (b) Sunny Day with Heavy Load (c) Cloudy Day with Normal Load

Fig. 12. Simulation conditions used for evaluating the performance of different HESS topologies.

day with normal electricity consumption. To present a fair compar- Fig. 13 shows the variations of the battery current in 24 h for the
ison and benchmarking between the proposed and existing HESS different HESS topologies and power management strategies con-
topologies, the Matlab Simulink model of the standalone PV micro- sidered in this work. As shown in the enlarged view of battery cur-
grid with battery-only ESS, passive SC-battery HESS and active rents (figures on the right), the proposed HESS demonstrated the
SC-battery HESS are constructed with identical battery and SC most stable current to the primary battery as compared to the
parameters. active HESS, while the smoothing capability of the passive HESS is
 W. Jing et al. / Sustainable Energy Technologies and Assessments 22 (2017) 55–64 61

80 Battery-Only Battery-Only
Passive HESS Passive HESS
Active HESS 50 Active HESS
Proposed HESS Proposed HESS
50
Battery Current (A)

20

20
-10

-40

0 4 8 12 16 20 0 16
Hours of Day Hours of Day
(a) Sunny Day with Normal Load

80 Battery-Only Battery-Only
50
Passive HESS Passive HESS
Active HESS Active HESS
Proposed HESS Proposed HESS
50
Battery Current (A)

20
20

-10

-40

-10
0 4 8 12 16 20 0 16
Hours of Day Hours of Day

(b) Sunny Day with Heavy Load

80 Battery-Only Battery-Only
Passive HESS Passive HESS
Active HESS 20 Active HESS
50 Proposed HESS Proposed HESS
Battery Current (A)

20

-10 -10

-40

0 4 8 12 16 20 0
Hours of Day
Hours of Day
(c) Cloudy Day with Normal Load
Fig. 13. Primary battery current of HESS Topologies under consideration and the coresponding enlarged views on the right.

hardly observable because the time constant of the SC response in ence among the different HESS topologies as shown in Fig. 14. This
within the range of seconds. Also observed from the battery cur- phenomenon is expected as the average power of the low frequency
rents is that the peak battery current in the proposed HESS scheme component does not change throughout the day.
is the lowest among the four topologies under consideration (see To evaluate the efficiency in utilization of SC in different HESS
Table 2). As for the cycling performance of primary battery, the topologies, the maximum variations in SC state-of-charge (max
change in battery state-of-charge does not show significant differ- [SoC_SC] – min[SoC_SC]) are compared as shown in Fig. 15 and
62 W. Jing et al. / Sustainable Energy Technologies and Assessments 22 (2017) 55–64

Table 2
Numerical simulation results.

HESS Weather/Load Maximum Battery Current Battery Depth-of-Discharge SC Utilization Battery Health Percentage Cost
Topology Condition (A) (%) (%) Cost (T)
Cost(T) Reduction1 (%)
Battery-Only Sunny/Normal 83.65 30.78 – 1.00 –
Sunny/Heavy 61.14 31.97 – 1.05 –
Cloudy/Normal 79.05 27.51 – 1.98 –
Passive HESS Sunny/Normal 82.02 30.66 9.91 0.92 8.0
Sunny/Heavy 59.62 31.82 9.54 1.03 1.9
Cloudy/Normal 81.21 27.39 9.73 1.92 3.0
Active HESS Sunny/Normal 77.29 30.77 63.97 0.67 33.0
Sunny/Heavy 54.88 31.78 63.69 0.71 32.4
Cloudy/Normal 76.21 27.35 62.09 1.19 39.9
Proposed Sunny/Normal 68.80 30.69(Primary) 60.88 0.58(Primary) 42.0
HESS Sunny/Heavy 48.35 30.92(Primary) 63.45 0.57(Primary) 45.7
Cloudy/Normal 71.37 26.79(Primary) 56.93 0.61(Primary) 69.2

#Note 1 – Percentage Cost(T) reduction is calculated in relative to battery-only system.

90 80 80
Battery-Only Battery-Only Battery-Only
Passive HESS Passive HESS Passive HESS
Active HESS Active HESS
Active HESS
State of Charge (%)

80 70 Proposed HESS Proposed HESS

Proposed HESS 70

State of Charge (%)

State of Charge (%)

70 60
60

60 50

50

50 40
0 4 8 12 16 20 0 0 4 8 12 16 20 0 0 4 8 12 16 20 0
Hours of Day Hours of Day Hours of Day

(a) Sunny Day with Normal Load (b) Sunny Day with Heavy Load (c) Cloudy Day with Normal Load

Fig. 14. Primary battery state-of-charge variation throughout the day in three different scenarios.

90 90 90

(b)
State of Charge (%)
State of Charge (%)

State of Charge (%)

70 70 70

50 50 50

30 30 30
Passive HESS Passive HESS Passive HESS
Active HESS Active HESS Active HESS
Proposed HESS Proposed HESS Proposed HESS
10 10 10
0 4 8 12 16 20 0 0 4 8 12 16 20 0 0 4 8 12 16 20 0
Hours of Day Hours of Day Hours of Day
(a) Sunny Day with Normal Load (b) Sunny Day with Heavy Load (c) Cloudy Day with Normal Load

Fig. 15. SC state-of-charge variation throughout the day in three different scenarios.

Table 2. As expected, the passive HESS has the lowest SC utilization function Cost(T) is formulated to quantify the impact of battery
of less than 10 percent as a result of sharing terminal voltage with current on the cycle life of battery as shown [38]:
battery. While the active and proposed HESS topologies have  
approximately similar SC utilization of about 60 percent. The XT dib ðtÞ
n1 ½ib ðtÞ þ n2  
2
CostðTÞ ¼
higher SC utilization rate suggests that the SC is capable of absorb- t¼0
dt 
ing more high-frequency fluctuation in ESS power exchange, hence 2
mitigating highly dynamic current in battery and also, better cost þ n3 ½maxðbðtÞÞ  minðbðtÞÞ

efficiency.  1; if ½ib ðtÞ  ib ðt  1Þ < 0
þ n4  þ n5 ð3Þ
0; if ½ib ðtÞ  ib ðt  1Þ P 0

Cost analaysis where T is the total operating time, ib(t) is the battery current, b(t) is
the battery SoC, while n1, n2, n3, n4 and n5 are positive constants. Five
In order to validate the effectiveness of the proposed HESS in life-limiting factors are considered: (1) charge/discharge rate, (2)
mitigating charge/discharge stress on battery, a battery health cost dynamicity of battery current, (3) depth-of-discharge (DoD), (4)
 W. Jing et al. / Sustainable Energy Technologies and Assessments 22 (2017) 55–64 63

charge/discharge transition, and (5) calendar life. The first term Although the battery aging process is a rather complex phe-
quantifies the damaging impact of high charge/discharge current; nomenon which cannot be quantified accurately with the simpli-
the second term captures the effect of undesirable fluctuation in bat- fied battery health cost function used in this work. However, the
tery current; the third term penalizes the impact of deep discharge; intention of this analysis is to evaluate and compare relatively
the fourth term considers the effect of cycling; and the last term cap- the effectiveness of different HESS schemes in mitigating detri-
tures calendar life of battery. The health cost of charge/discharge mental loading conditions in battery based on the well-
rate and depth-of-discharge are modeled as convex function [39]. understood life-limiting factors. It is assumed that the impacts of
In order to benchmark the different HESS topologies to battery- life-limiting factors to battery’s health are linear and independent.
only system, the Cost(T) for all systems are normalized based on In order to justify the additional cost of implementing the sec-
the battery health cost of battery-only system. Fig. 16 shows the ondary Li-ion battery module with higher price tag, a financial
normalized battery health cost of the four HESS topologies under analysis is presented in Table 3. Assuming the battery’s life cycle
consideration in the three different scenarios. Table 2 summarizes is linearly proportional to the reciprocal of battery health cost,
the simulation results presented in Section 3 and the associated the estimated life cycle of battery for different topologies are calcu-
battery health cost analysis. The results show that the proposed lated based on the typical life expectancy of 500 cycles for Lead
multi-level HESS demonstrates about 42–46 percent reduction in Acid battery and 2000 cycles for Li-ion battery. The secondary bat-
battery health cost in normal condition (sunny day), while nearly tery of the proposed HESS is estimated to last 50 percent of its nor-
70 percent reduction in battery health cost in harsh condition mal life expectancy due to the fact that this battery will be
(cloudy day) compared to battery-only system. The proposed HESS handling harsh charge/discharge current. Assuming the battery is
performs approximately 11–30 percent more superior than the cycled once a day for the entire year, the estimated battery operat-
conventional active HESS which suggests the slowest deterioration ing cost per annum is estimated. Based on the financial analysis,
rate in battery performance and longest service life under despite the addition of secondary battery module of higher cost,
comparable operating condition. the proposed multi-level HESS still managed to reduce the

1.2 1.2 2.2

Battery-Only Battery-Only Battery-Only Max:1.98
Passive HESS Passvie HESS Max:1.05 Passive HESS
Max : 1.00
1 Active HESS 1 Active HESS 1.8 Active HESS
Max:1.03 Max:1.92
Proposed HESS Proposed HESS Proposed HESS
Max : 0.92
0.8 0.8 1.4
Cost (T)

Cost (T)
Cost (T)

Max : 0.67
Max:0.71
Max:1.19
0.6 0.6 1
Max : 0.58
Max:0.57

0.4 0.4 0.6

Max:0.61

0.2 0.2 0.2

0 4 8 12 16 20 0 0 4 8 12 16 20 0 0 4 8 12 16 20 0
Hours of Day Hours of Day Hours of Day
(a) Sunny Day with Normal Load (b) Sunny Day with Heavy Load (c) Cloudy Day with Normal Load

Fig. 16. Primary battery normalized health cost throughout the day in three different scenarios.

Table 3
Financial analysis of different HESS.

HESS Topology Weather/load Capacity Initial Cost3 Battery Estimated Cost/Cycle Estimated Annual Percentage Cost
condition4 (kWh) ($) Health Cost Life Cycle1 ($) Battery Cost Reduction5
Cost() ($) (%)
Battery-only S/N 48 7200 1.00 500 14.40 5256.00 0
S/H 1.05 476 15.13 5522.45 0
C/N 1.98 252 28.57 10428.05 0
Passive HESS S/N 48 7200 0.92 547 13.16 4803.40 8.6
S/H 1.03 485 14.85 5420.25 1.9
C/N 1.92 260 27.69 10106.85 3.1
Active HESS S/N 48 7200 0.67 747 9.64 3518.60 33.1
S/H 0.71 704 10.23 3733.95 32.4
C/N 1.19 420 17.14 6256.10 40.0
Proposed HESS
Primary S/N 45.6 6840 0.58 865 7.91 2887.15 + 350.40 = 3237.55 38.4
Secondary 2.4 960 – 10002 0.96
Primary S/H 45.6 6840 0.57 877 7.80 2847.00 + 350.40 = 3197.4 42.1
Secondary 2.4 960 – 10002 0.96
Primary C/N 45.6 6840 0.61 820 8.34 3044.10 + 350.40 = 3394.50 67.5
Secondary 2.4 960 – 10002 0.96

#Note 1 – Typical life cycle/Cost of battery utilization; typical life cycle for LA – 500 cycles, and Li-ion – 2000 cycles.
#Note 2 – Estimated to perform 50% of the expected cycle life.
#Note 3 – Initial cost of LA battery ($150/kWh) and Li-ion battery ($400/kWh) is considered.
#Note 4 – S – Sunny, C – Cloudy, N – Normal load, H – Heavy load.
#Note 5 – Percentage cost reduction is calculated based on battery-only system.
64 W. Jing et al. / Sustainable Energy Technologies and Assessments 22 (2017) 55–64

operating cost of energy storage system by a minimum of 38.4% in [14] Jing W, Lai CH, Dennis Wong ML, Wong WSH, Smart hybrid energy storage for
stand-alone PV microgrid: optimization of battery lifespan through dynamic
gentle condition and 67.5% in harsh condition.
power allocation. In: Asia-Pacific Power Energy Eng. Conf. APPEEC, vol. 2016–
Janua, pp. 3–7, 2016.
[15] Gee AM, Robinson FVP, Dunn RW. Analysis of battery lifetime extension in a
Conclusion small-scale wind-energy system using supercapacitors. IEEE Trans Energy
Convers 2013;28(1):24–33.
[16] Choi M, Kim S, Seo S. Energy management optimization in a
Standalone photovoltaic microgrid with energy storage system
battery/supercapacitor hybrid energy storage system. IEEE Trans Smart Grid
has been an attractive solution for off-grid communities. Lead acid 2012;3(1):463–72.
battery as the mainstream energy storage system for renewable [17] Mendis N, Muttaqi KM, Perera S, Management of battery-supercapacitor
microgrid suffers from low life expectancy which results in poor hybrid energy storage and synchronous condenser for isolated operation of
pmsg based variable-speed wind turbine generating systems, 5(2), 944–953,
reliability and high operating cost. Hybridization of energy storage 2014.
devices with different characteristic and its associated energy [18] Adhikari S, Lei Z, Peng W, Tang Y. A battery/supercapacitor hybrid energy
management system have been actively researched in quest of storage system for DC microgrids. ECCE Asia 2016:8–14.
[19] Glavin ME, Hurley WG. Optimisation of a photovoltaic battery ultracapacitor
enhancing the service life of battery. This paper proposed a novel hybrid energy storage system. Sol Energy 2012;86(10):3009–20.
multi-level HESS configuration and the associated control strategy [20] Dougal RA, Member S, Liu S, White RE. Power and life extension of battery –
to mitigate the battery stress under dynamic power exchange con- ultracapacitor hybrids. IEEE Trans Components Packag Technol 2002;25
(1):120–31.
dition that often experienced in standalone renewable microgrid. A [21] Tran D, Khambadkone AM, Member S, Energy Management for Lifetime
Matlab Simulink model of typical standalone photovoltaic micro- Extension of Energy Storage System in Micro-Grid Applications, 4(3), 1289–
grid is developed to analyze the performance of HESS in stress 1296, 2013.
[22] Zhou H, Bhattacharya T, Tran D, Siew TST, Khambadkone AM. Composite
reduction which leads to service life extension. Different condi-
energy storage system involving battery and ultracapacitorwith dynamic
tions of solar energy input and load profiles from a rural site in Sar- energy management in microgrid applications. IEEE Trans Power Electron
awak were used in the simulation and analysis. A battery health 2011;26(3):923–30.
[23] Kollimalla SK, Mishra MK, Narasamma NL. Design and analysis of novel control
cost function is formulated based on selected life-limiting factors
strategy for battery and supercapacitor storage system. IEEE Trans Sustain
to predict the health cost of battery in different HESS topologies. Energy 2014;5(4):1137–44.
The study results show that the proposed multi-level HESS could [24] Dusmez S, Khaligh A. A supervisory power-splitting approach for a new
effectively mitigate stress on battery with lowest impact on the ultracapacitor-battery vehicle deploying two propulsion machines. IEEE Trans
Ind Informatics 2014;10(3):1960–71.
battery health. This suggests that improvement of battery life can [25] Grasso AD, Sapuppo C, Tina GM, Giusto R. MPPT charge regulator for
potentially be achieved by using the proposed system. Financial photovoltaic stand-alone dual battery systems. Electr. Eng. Res. Rep. 2009;2
analysis shows that the proposed system is financially viable with (1).
[26] Barca G, Moschetto A, Sapuppo C, Tina GM, Giusto R, Grasso AD, A novel MPPT
no significant increase in overall cost. charge regulator for a photovoltaic stand-alone telecommunication system,
235–238, 2008.
[27] Jing W, Lai CH, Wong SHW, Wong MLD, Battery-supercapacitor hybrid energy
References storage system in standalone dc microgrids: a review, IET Renew Power Gener.
[28] Bortolini M, Gamberi M, Graziani A. Technical and economic design of
[1] Van Nes EH, Sche M, Brovkin V, Lenton TM, Ye H, Deyle E, Sugihara G. Causal photovoltaic and battery energy storage system. Energy Convers Manage
feedbacks in climate change. Nat Clim Chang 2015:3–6. 2014;86:81–92.
[2] Owusu PA, Asumadu-sarkodie S. A review of renewable energy sources, [29] Kuperman A, Aharon I, Kara A, Malki S. A frequency domain approach to
sustainability issues and climate change mitigation. Cogent Eng 2016;15 analyzing passive battery-ultracapacitor hybrids supplying periodic pulsed
(1):1–14. current loads. Energy Convers Manage 2011;52(12):3433–8.
[3] International Energy Agency, World Energy Outlook 2016, IEA, vol. Chapter 1, [30] Shen J, Khaligh A, Member S. A Supervisory energymanagement control
2016. strategy in a battery/ultracapacitor hybrid energy storage system. IEEE Trans
[4] Tan X, Li Q, Wang H. Advances and trends of energy storage technology in Transp Electrif 2015;1(3):223–31.
Microgrid. Int J Electr Power Energy Syst 2013;44(1):179–91. [31] Shao L, Moshirvaziri M, Malherbe C, Moshirvaziri A, Eski A, Dallas S, Hurzook F,
[5] Gouveia C, Moreira J, Moreira CL, Pecas Lopes JA. Coordinating storage and Trescases O, Ultracapacitor/battery hybrid energy storage system with real-
demand response for microgrid emergency operation. IEEE Trans Smart Grid time power-mix control validated experimentally in a custom electric vehicle.
2013;4(4):1898–908. In: 2015 IEEE Appl. Power Electron. Conf. Expo., pp. 1331–1336, 2015.
[6] Shivashankar S, Mekhilef S, Mokhlis H, Karimi M. Mitigating methods of power [32] Ma T, Yang H, Lu L. Development of hybrid battery-supercapacitor energy
fl uctuation of photovoltaic (PV) sources – a review. Renew Sustain Energy Rev storage for remote area renewable energy systems. Appl Energy
2016;59:1170–84. 2015;153:56–62.
[7] Spanos C, Turney DE, Fthenakis V. Life-cycle analysis of fl ow-assisted nickel [33] Etxeberria A, Vechiu I, Camblong H, Vinassa JM. Comparison of three
zinc-, manganese dioxide-, and valve-regulated lead-acid batteries designed topologies and controls of a hybrid energy storage system for microgrids.
for demand-charge reduction. Renew Sustain Energy Rev 2015;43:478–94. Energy Convers Manage 2012;54(1):113–21.
[8] Dufo-lópez R, Lujano-rojas JM, Bernal-agustín JL, Comparison of different lead [34] Zhang S, Xiong R, Zhou X. Comparison of the topologies for a hybrid energy-
– acid battery lifetime prediction models for use in simulation of stand-alone storage system of electric vehicles via a novel optimization method. Sci China
photovoltaic systems, 115, 242–253, 2014. Technol Sci 2015;58(7):1173–85.
[9] Scuiller F, Tang T, Zhou Z, Benbouzid M, Fre J. A review of energy storage [35] Ultra P, Battery C, Jamshidpour E, Poure P, Saadate S, Energy management and
technologies for marine current energy systems. Renew Sustain Energy Rev control of a stand-alone photovoltaic/ultra capacitor/battery microgrid. In:
2013;18:390–400. 2015 IEEE Jordan Conf. Appl. Electr. Eng. Comput. Technol., vol. 2, no. 1, pp. 1–
[10] Lahyani A, Venet P, Guermazi A, Troudi A. Battery/supercapacitors 12, 2016.
combination in uninterruptible power supply (UPS). IEEE Trans Power [36] Allègre AL, Bouscayrol A, Trigui R, Influence of control strategies on
Electron 2013;28(4):1509–22. battery/supercapacitor hybrid Energy Storage Systems for traction
[11] Yin H, Zhao C, Li M, Ma C. Utility function-based real-time control of a battery applications, 213–220, 2009.
ultracapacitor hybrid energy system. IEEE Trans Ind Informatics 2015;11 [37] Jiang W, Zhang L, Zhao H, Hu R, Huang H. Research on power sharing strategy
(1):220–31. of hybrid energy storage system in photovoltaic power station based on multi-
[12] Yang Lei, Chen Wenjun, Wang Yanzi, Wang Weida, Zhao Yulong. A fuzzy-logic objective optimisation. IET Renew Power Gener 2016;10(5):575–83.
power management strategy based on markov random prediction for hybrid [38] Jing WL, Lai CH, Wong WSH, Wong MLD, Cost Analysis of battery-
energy storage systems. Energies 2016;9(1):25. supercapacitor hybrid energy storage system for standalone Pv systems. In:
[13] Bolborici V, Dawson FP, Lian KeK. Hybrid Energy Storage Systems: connecting 4th IET Int. Conf. Clean Energy Technol., 2016.
batteries in parallel with ultracapacitors for higher power density. IEEE Ind. [39] Li N, Chen L, Low SH, Optimal demand response based on utility maximization
Appl. Mag. 2014:31–40. in power networks. In: Power Energy Soc. Gen. Meet., 1–8, 2011.