Journal of Power Electronics (2020) 20:308–318

https://doi.org/10.1007/s43236-019-00027-0

ORIGINAL ARTICLE

Battery–inductor–supercapacitor hybrid energy storage system for DC

microgrids
Duy‑Hung Dam1 · Hong‑Hee Lee1

Received: 10 September 2019 / Revised: 6 November 2019 / Accepted: 14 November 2019 / Published online: 10 December 2019
© The Korean Institute of Power Electronics 2019

Abstract
This paper presents a new configuration for a hybrid energy storage system (HESS) called a battery–inductor–supercapacitor
HESS (BLSC-HESS). It splits power between a battery and supercapacitor and it can operate in parallel in a DC microgrid.
The power sharing is achieved between the battery and the supercapacitor by combining an internal battery resistor and an
output LSC filter, which consists of a supercapacitor and an inductor. The battery current is smoothened to supply and receive
only the low-frequency current component under any disturbance and load condition through cooperation with the superca-
pacitor. Complete guidelines to design the parameters of the BLSC-HESS are also presented. Simulation and experimental
results demonstrate that the proposed BLSC-HESS configuration achieves a high-stability performance and a lower cost,
and is easily applied to DC microgrids.

Keywords DC microgrid · Batteries · Hybrid energy storage system · Supercapacitor · Pulsed load

1 Introduction pulse current significantly increases the operation tempera-

Due to the increasing number of DC generation sources
and requirements of energy storage devices, DC microgrids
have been attracting a great deal of attention. In a microgrid
system, there are many kinds of renewable energy sources
(RESs) such as photovoltaics, wind turbines, and fuel cells,
and it is necessary to integrate them. To stabilize the distur-
bances caused by changes in the loads and generation in a
microgrid, a battery energy storage system (BESS) is one of
the most promising solutions. An ideal BESS has very high
energy and power densities, which has yet to be achieved.
Fortunately, the combination of a battery and supercapaci-
tor can provide high energy and power densities in a hybrid
energy storage system (HESS) [1].
A typical DC microgrid is composed of different RESs
and HESSs, as illustrated in Fig. 1. Each unit is connected
to a common DC bus through a DC–DC converter to sup-
ply power to various kinds of loads. When the power load
changes, a HESS supports the transient high power and dis-
patches the power fluctuation of the RESs [1]. For instance,
* Hong‑Hee Lee
hhlee@mail.ulsan.ac.kr
1
Department of Electrical Engineering, University of Ulsan,
Ulsan, Korea
ture of batteries, which decreases the system efficiency and
the lifetime of the batteries. Furthermore, fast load changes
and low-frequency current ripples shorten the lifetime of a
fuel cell and affect its long-term performance [2]. To solve
this problem, a supercapacitor can be used to manage pulse
loads and load transitions. The size of the battery can be
reduced by 64% when it is combined with a supercapacitor
in a HESS when compared to an energy storage system with
only a battery [3].
Many HESS configurations have been studied for DC
microgrids. These configurations are classified into three
types, as shown in Fig. 2.
(1) A battery and a supercapacitor are connected in parallel
to the DC bus through a DC–DC converter, as shown
in Fig. 2a [4, 5]. This is the simplest configuration.
It has advantages such as low cost and easy control.
However, current sharing between the battery and the
supercapacitor is not controllable.
(2) A battery and a supercapacitor are connected to a multi-
input topology converter, as shown in Fig. 2b [6–9].
In this method, the line impedance between the bat-
tery and the supercapacitor is neglected, which reduces
the system losses. However, the control scheme of this
method is hard to adapt to a DC microgrid, since it

13
Vol:.(1234567890)
Battery–inductor–supercapacitor hybrid energy storage system for DC microgrids 309
Fig. 1  Typical configuration of a DC microgrid with HESSs
Fig. 2  HESS configurations
requires an additional control method to cooperate with
other units in the microgrid.
(3) A battery and a supercapacitor are connected to the DC
bus through a DC–DC converter, as shown in Fig. 2c
[10, 11]. In this configuration, the control options
are increased. However, a complicated controller is
required to coordinate the operation of the superca-
pacitor and the battery. Moreover, the line resistance
between the supercapacitor and the battery requires a
spike in the battery current when the load changes.
To integrate a HESS into a DC microgrid, a dedicated
control method should be used. Recently, several coordinat-
ing methods have been adapted to integrate a battery and a
supercapacitor into a system based on a high-pass filter or
virtual impedance methods. Two main objectives are con-
sidered: battery current compensation and supercapacitor
state of charge (SoC) regulation [12–18]. A supercapacitor
has been used to handle the high-frequency components of
the HESS output current [12–14]. Thanks to this method,
the battery current can be smoothly compensated. A virtual
impedance method was applied to make the supercapacitor
and the other units cooperate [16–18]. However, none of
these methods can seamlessly restore the SoC of the super-
capacitor, which requires additional controllers.
The use of a bulky and high-cost DC–DC converter to
connect a supercapacitor and a battery is another obstacle
to the integration of the supercapacitor into systems such as
electric vehicles and small residential buildings. A passive
solution, where the battery and supercapacitor are connected
in parallel, is widely used in electric vehicles due to its low
cost, low maintenance and small volume. Moreover, this
passive solution can operate without considering the super-
capacitor voltage or SoC. However, when a HESS is used
as an input source for a DC–DC converter to increase the
power quality of the system, the switching noise and ripple
current from the DC–DC converter cannot be filtered out by
a conventional parallel battery–supercapacitor configuration.
It also has a limitation in that the high-frequency current
component leaks to the battery side, which decreases the
lifetime of the battery.
The authors of [19] proposed a new HESS configuration
to smoothen battery current using a supercapacitor and an
inductor. However, the design and performance were not
sufficiently verified. Furthermore, a combined model for
the HESS and the DC–DC converter was considered for the
theoretical analysis. Therefore, the HESS model was not
generalized for application with any type of DC–DC con-
verter. To solve this problem, a battery inductor supercapaci-
tor HESS (BLSC-HESS), which was developed by analyzing
the HESS and DC–DC converter individually, is proposed.
The proposed BLSC-HESS can be easily connected to any
DC–DC converter in a DC microgrid with a simple droop
controller. To analyze the reliability of the proposed topol-
ogy, the parameters were investigated, including the superca-
pacitance, inductance and internal resistance of the battery.
Design guidelines for the BLSC-HESS parameters are pre-
sented. Finally, the effectiveness of the proposed topology
was verified by simulation and experimental results.
2 System description
A typical DC microgrid including a HESS is shown in
Fig. 1. The sources and loads are connected to a common
DC bus. The HESS units are used to compensate the current
mismatch between the RES and the load to effectively oper-
ate the microgrid system.
2.1 Cooperation in a DC microgrid
To integrate various sources in a DC microgrid, a cooperat-
ing controller such as a master–slave or droop controller is
normally required. In general, droop control is more popular
for DC microgrids due to its advantage in terms of decen-
tralization without a communication interface. The current
sharing in droop control is based on the virtual resistance or
droop coefficient, and the higher the droop coefficient is, the
13
310
Fig. 3  Models of: a impedance-based battery; b simple battery; c
supercapacitor
lower the shared current is. In a DC microgrid, at least one
RES unit is operated with droop control to regulate the DC
bus voltage. In this study, a DC microgrid is controlled by
droop control to share power between sources.
2.2 Battery and supercapacitor models
Figure 3 shows battery and supercapacitor models, which
are used as equivalent models to evaluate the performance of
the HESS. Figure 3a is the impedance-based model, which
is widely used in battery design, SoC estimation or real-time
battery simulation [20]. In this paper, since the battery is
only used for charging or discharging in a HESS applica-
tion, the model in Fig. 3b is used to analyze a battery that
operates at a low frequency [21, 22]. In Fig. 3b, the battery
is regarded as a voltage source Vbat with a series resistor Rb
at a specific SoC:
{ current isc can be obtained as
Vbat = VOC (SoC)
Rb = Rs
, (1)
where VOC (SoC) is the open circuit voltage of the battery
at a specific SoC, and Rs is internal equivalent resistance.
The supercapacitor has a much higher power density and
a lower internal resistance, and can be modeled as shown in
Fig. 3b [4]. The values of Csc and Rsc are from a datasheet
provided by the manufacturer.
3 Proposed BLSC‑HESS configuration
The proposed BLSC-HESS configuration is shown in Fig. 4.
The BLSC-HESS is constructed by inserting an induc-
tor between the battery and the supercapacitor, which are
13
D.-H. Dam, H.-H. Lee
Fig. 4  Proposed HESS configuration
connected to a common DC bus through a DC–DC con-
verter. As can be seen in Fig. 4, the construction of the
inductor and supercapacitor is regarded as a low-pass filter
for the battery. The output current of the BLSC-HESS is
decoupled into two parts based on the frequency domain:
the battery current and the supercapacitor current. The low-
frequency current is handled by the battery, while the high-
frequency current is handled by the supercapacitor. The
BLSC-HESS parameters are investigated to analyze the
performance of this low-pass filter.
3.1 BLSC‑HESS parameters
To design the parameters of the BLSC-HESS properly,
there are two main points to consider: the current coupling
frequency between the battery and the supercapacitor, and
the voltage level of the supercapacitor. The currents on the
BLSC-HESS can be analyzed using the transfer function
GiBLSC (s) of the ratio between the battery current and the
supercapacitor current. The supercapacitor voltage level
can be calculated using the transfer function GvBLSC (s) of the
supercapacitor voltage level over the battery voltage level.
From Fig. 4, the battery current ibat and the supercapacitor
{ ( ) ( )
ib = Vbat(− vsc ∕ R)b + ZL1
isc = vsc ∕ Rsc + ZCsc
, (2)
where ZL1 and ZCSC are the impedances of L1 and CSC, respec-
tively. In the frequency domain, Eq. (2) is rewritten as
⎧
−vsc (s)
⎪ ib = (Rb +L1 s)
⎨ i = � vsc (s) � . (3)
⎪ sc Rsc + C 1 s
⎩ sc
The total current of the HESS is the sum of the battery
current and the supercapacitor current:
i = ib − isc (4)
Then, GiBLSC (s) is obtained as follows:
Battery–inductor–supercapacitor hybrid energy storage system for DC microgrids 311

ib 1 + Rsc Csc s
GiBLSC (s) = = ( ) . (5)
i 1 + Rsc + Rb Csc s + L1 Csc s2

From (5), GiBLSC (s) has one zero and two poles when L1 is
greater than zero:
� �
⎧ z = −1∕ R C
sc sc √
⎪ 1
⎪ p = − (Rsc +Rb )Csc − (Rsc +Rb ) Csc −4L1 Csc
2 2

⎨ 1
√
2L1 Csc . (6)
⎪ (Rsc +Rb )Csc + (Rsc +Rb ) Csc −4L1 Csc
2 2
⎪ P2 = −
⎩ 2L1 Csc

In the case of a conventional battery and supercapacitor

connection, L1 is zero and GiBLSC (s) becomes

| i 1 + Rsc Csc s
GiBSC (s) = GiBLSC (s)| = b = ( ) . (7) Fig. 5  Bode diagrams of GiBLSC (s) (solid lines) and GiBSC (s) (dot-
|L1 =0 i 1 + Rsc + Rb Csc s
ted lines) according to Csc when L1 = 0.5 mH, RSC = 15 mΩ and
Rb = 0.1 Ω
From (7), GiBSC (s) has one zero and one pole:
{ 1
z10 = − R
sc Csc
1 . (8)
p10 = −
(Rsc +Rb )Csc

Since one pole is missing, the conventional configuration

has a weak point in that the gain of GiBSC (s) at high frequen-
cies is high due to its zero. To analyze the effects of the
BLSC-HESS parameters, each parameter was swept, and
Bode diagrams of the BLSC-HESS were shown to examine
the performance.

3.2 Supercapacitor Csc

Figure 5 shows bode diagrams of GiBLSC (s) and GiBSC (s) when
the value of Csc changes from 10 to 30 F. The remaining
parameters are L1 = 0.5 mH, RSC = 15 mΩ and Rb = 0.1 Ω.
As shown in Fig. 5, the cut-off frequency of GiBLSC (s)
increases when Csc increases. Theoretically, the decoupling Fig. 6  Bode diagrams of GiBLSC (s) (solid lines) and GiBSC (s) (dot-
ted lines) according to Rb when L1 = 0.5 mH, RSC = 15 mΩ and
frequency of GiBLSC (s) is determined by the first pole p1 in
Csc = 28 F
(6) as follows:
( ) √
R + R C − (Rsc + Rb )2 Csc
2 − 4L C
p sc b sc 1 sc The capacitance of the supercapacitor does not affect
fc = − 1 = .
2𝜋 4𝜋L1 Csc the gain of GiBLSC (s) at high frequencies, as shown in Fig. 5.
(9) Therefore, increasing the capacitance of the supercapacitor
L1 is very small when compared with Csc. Thus, the value does not help reduce the high-frequency current component
of the first pole p1 is almost the same as p10 in (8) since the on the battery side.
cut-off frequencies of GiBLSC (s) and GiBSC (s) are the same, as
shown in Fig. 5. Finally, the decoupling frequency fc can be
rewritten as 3.3 Internal resistance Rb of the battery
p10 1 The internal resistance of the battery is predefined, and
fc ≈ − = ( ) . (10)
2𝜋 2𝜋 Rsc + Rb Csc its effect was analyzed to examine the performance of the
BLSC-HESS in different systems. Bode diagrams of GiBLSC (s)
and GiBSC (s) are shown in Fig. 6 with Rb changing from

13
312 D.-H. Dam, H.-H. Lee

0.1 to 0.5 Ω The remaining parameters are L1 = 0.5 mH , Therefore, to effectively reduce the high-frequency current
RSC = 15 Ωm and Csc = 28 F. component on the battery side, the value of L1 is calculated
With the conventional configuration, a battery with a high so that the gain of GiBLSC (s) at 100 Hz is less than − 20 dB.
internal resistance can reduce the effect of a high-frequency
|
current component on the battery side since the gain of GiBLSC (s)| < − 20 dB (11)
|f =100 Hz
GiBSC (s) is very small, as shown in Fig. 6. However, when
increasing the battery capacity by increasing the number
of parallel battery cells, the internal resistance becomes
smaller. Therefore, the gain of GiBSC (s) becomes higher. This 3.5 SC voltage
means that the high-frequency current can flow to the bat-
tery side. To analyze the supercapacitor voltage variation, a transfer
function of the supercapacitor voltage over the battery volt-
3.4 Inductance L1 age is obtained from Fig. 4:

Rc + C1 s
To evaluate the effect of the inserted inductor, Bode dia- vsc
(12)
sc
Gvsc (s) = = .
grams of GiBLSC (s) and GiBSC (s) are shown in Fig. 7 when L1 Vbat Rc + C1 s + L1 s
changes from 0.5 to 2.5 mH. The remaining parameters sc

are Rb = 0.1 Ω, RSC = 15 mΩ and Csc = 28 F. The gain of
GiBSC (s) becomes constant at high frequencies and is higher
than − 20 dB when the internal battery resistance is small
( Rb = 0.1). As a result, currents with high frequencies such
as ripple current, harmonic current and switching current
can flow from the supercapacitor to the battery, which
greatly reduces the lifetime of the battery.
On the other hand, the performance of the BLSC-HESS
is the same as that of the conventional configuration in the
low-frequency range. However, in a general DC microgrid,
harmonic current can appear due to various types of loads.
With a load such as a variable motor driver, the harmonic
is double the fundamental frequency (100 Hz or 120 Hz). As
shown in Fig. 7, the gain of GiBSC (s) is constant and higher
than − 20 dB when the frequency is greater than 1 Hz. How-
ever, the gain of GiBLSC (s) becomes significantly smaller
when compared with GiBSC (s) due to the inserted inductor L1.
From (12), Gvsc (s) becomes unity when L1 s becomes
very small since it is difficult for battery voltage to change
quickly. As a result, the output voltage of the supercapaci-
tor is equal to the battery output voltage in the steady state.
By choosing the rated voltage of the supercapacitor to be
higher than the operation voltage of the battery, the task of
controlling the supercapacitor voltage can be neglected in
the BLSC-HESS.
3.6 Settling time
The settling time of the BLSC-HESS is defined as the time
required for the response of the BLSC-HESS to reach and
stay within a range of 10% of its final output current in this
study. To evaluate the settling time, a unit step of R(s) = 1∕s
was used as the input of the GiBLSC (s), and the output is as
follows:

1 1 + Rsc Csc s
F(s) = R(s)GLCS (s) = × ( ) .
s 1 + Rsc + Rb Csc s + L1 Csc s2
(13)
Using the MATLAB symbolic toolbox, the output signal
is expressed in the time domain as follows:
t(Rb +Rsc )

F(t) = 1 − e
− 2L1
Fsub (t). (14)

Due to the complicated Fsub (t), a statistical method was

Fig. 7  Bode diagrams of GiBLSC (s) (solid lines) and GiBSC (s) (dotted
lines) according to L1 when Rb = 0.1 Ω, RSC = 15 mΩ and Csc = 28 F
used to estimate the settling time. The decoupling frequency
is decided by the first pole p1. Thus, the settling time ts was
calculated by MATLAB and compared with the value of the
first pole to determine the relationship between them. The
parameters(( of the BLSC-HESS
) ) were swept to evaluate the
ratio ts ∕ Rsc + Rb Csc . The parameter ranges are from 0.1
to 0.54 Ω for Rb , from 10 to 40 F for Csc(( , and from) 0.1)to
1.5 mH for L1. The calculated ratio of ts ∕ Rsc + Rb Csc is

13
Battery–inductor–supercapacitor hybrid energy storage system for DC microgrids 313

v∗out = Vnom − rd iout , (18)

where Vnom is the nominal DC bus voltage, Vout ∗

and iout are
the reference output voltage and the output current of the
HESS, respectively. In addition, rd is the droop coefficient
or virtual impedance.
The droop coefficient is calculated based on the rated cur-
rent of the HESS as
Vnom − Vmin
rd = , (19)
irated

where Vmin is the minimum allowable voltage of the DC

Fig. 8  Ratio between the settling time and Csc (Rbat + Rsc ) bus, and irated is the rated current of the HESS converter. The
output voltage of the HESS is regulated by the inductor cur-
rent, and the inductor current reference i∗L is obtained from
the PI controller of the voltage control loop:
( )
kIv ( ∗ )
(20)
∗
iL = kPv + vout − vout ,
s

Fig. 9  Control diagram of the BLSC-HESS where kPv and kIv are the PI gains of the voltage control loop.
Another PI controller is then used to control the inductor
current:
plotted in Fig. 8, where the ratio is approximately constant at ( )
kIi ( ∗ )
2.25. Therefore, the settling time can be estimated as d = kPi + iL − iL , (21)
s
( )
ts ≈ 2.25Csc Rb + Rsc . (15)
where kPi and kIi are the PI gains of the current control loop.
Finally, design guidelines for the BLSC-HESS are briefly The cut-off frequency of the current loop is selected to be
summarized as follows: 1∕10 of the switching frequency of the HESS converter, and
Step 1 Select the settling time ts in (15) or (16), which the cut-off frequency of the voltage control loop is chosen
is calculated based on the decoupling frequency fc of the as 1∕10 of the current loop cut-off frequency.
BLSC-HESS:
3.8 HESS size and efficiency discussion
2.25 1
ts =
2𝜋 fc
. (16)
The proposed configuration uses only one additional induc-
tor instead of a DC–DC converter. Even though its volume
Step 2 Calculate the capacitance of the supercapacitor:
becomes slightly larger when compared with the configura-
ts tion in Fig. 2a, it is much smaller than the configuration in
Csc = ( ). (17)
2.25 Rsc + Rb Fig. 2b or c. In addition, the inductor volume is much smaller
with a cheaper price when compared with a DC–DC converter,
Step 3 Calculate the value of the inductor, which helps to because a small inductance is enough to decouple the high
reduce the ripple current on the battery side by reducing the frequency current between the battery and the SC. Since the
gain of GiBLSC (s) at a high frequency. As described in (11), the cut-off decoupling frequency depends on the hardware param-
value of the inductor is chosen so that the gain of GiBLSC (s) at eters, it needs to be changed when the battery or inductor is
100 Hz is smaller than − 20 dB. replaced. Some HESS topologies are compared in Table 1.
From Table 1, it can be said that the proposed configuration
has superior performance in terms of price and volume.
3.7 HESS control scheme

The BLSC-HESS is easily controlled without any concern

about the SoC of the supercapacitor or its voltage restora-
tion. Figure 9 shows a HESS control diagram. From Fig. 9,
the droop controller is described as

13
314 D.-H. Dam, H.-H. Lee

Table 1  Comparison between Configuration Components Volume Cost High-frequency Efficiency

different HESS configurations compensation perfor-
mance

Figure 2a BAT + SC + 1 × converter ++ ++ − ++

Figure 2b BAT + SC + 1 × multiport-converter − − +++ +
Figure 2c BAT + SC + 2 × converter −− −− +++ +
Proposed BAT + SC + L + 1 × converter + + ++ ++

− Negative point, + Positive point

Table 2  Parameters used in the simulation and experiment

Parameters Symbol Value

DC microgrid
DC bus voltage Vnom − Vmin 45–48 V
Droop coefficient of HESS rd 0.3
DGs output current 0–8 A
Load 10, 20 Ω
BLSC-HESS converter
Inductor L2 0.5 mH
Output capacitor C 2200 µF
Switching frequency fs 20 kHz
Battery open circuit voltage VOC 24 V
SC nominal voltage 24 V

4 Simulation results Fig. 10  Case I: simulation results of a single HESS in a system

The DC microgrid shown in Fig. 1 was modeled using

PSIM simulation software to evaluate the proposed BLSC-
HESS with the parameters in Table 2. The nominal DC
bus voltage in the microgrid is 48 V with a load of 20 Ω,
and the two distributed generation sources are controlled
as a current source to imitate renewable energy sources.
The BLSC-HESS is connected to the DC bus through a
bidirectional converter, as shown in Fig. 4.
To evaluate the effectiveness of the BLSC-HESS, three
case studies were investigated. Case I shows normal opera-
tion, case II shows the effect of the BLSC-HESS parame-
ters on settling time, and case III evaluates the cooperation
of two BLSC-HESSs.
Case I Only one BLSC-HESS is activated, and
the parameters of the BLSC-HESS are L1 = 0.47 mH ,
Csc = 4 F , Rsc = 25 mΩ and Rb = 0.1 Ω. In Fig. 10, when
the distributed generation current suddenly increases from
0 to 5 A, the BLSC-HESS changes from the charge mode
to the discharge mode to balance the system power flow
according to the change of the injected current from the
distributed generation. As shown in Fig. 10a, even though
the HESS changes from the charge mode to the discharge
mode, the battery output current iBAT1 changes smoothly,
thanks to the pulse current iSC1 supplied by SC1.
Figure 10b shows the output current iHESS1 of the HESS1
converter to regulate the output voltage, which is calculated
by the droop equation in (18). Therefore, the DC bus voltage
vbus is regulated very well, as shown in Fig. 10c. Moreover,
the battery and supercapacitor voltages are always the same,
as shown in Fig. 10d, which confirms the relationship in (12)
with normal battery voltage variations. Figure 10 shows that
the BLSC-HESS effectively decouples the current from the
battery and the supercapacitor under load variations.
Case II Figure 11 shows the output currents iBAT1 and
iSC1 of the battery and the supercapacitor when changing
the parameters. For the same parameters used in case I,
the output current is shown in Fig. 11a for a performance
comparison. In addition, the settling time is 1 s, which is
calculated from (15). When the supercapacitor capacitance
is doubled to 8 F, the settling time becomes 2 s. Figure 11b
shows that the battery current response becomes slower due
to the decreased decoupling frequency fc of GiBLSC (s) with an
increased capacitance. In Fig. 11c, the internal resistance of

13
Battery–inductor–supercapacitor hybrid energy storage system for DC microgrids 315
Fig. 11  Case II: impacts of the SC capacity Csc and the battery inter-
nal resistance Rb on simulation results
the battery is increased to Rb = 0.2 Ω, while Csc is kept at
4 F. The settling time of the BLSC-HESS is the same (2 s)
as in Fig. 11b. Case II verifies the theoretical analysis to
determine the settling time in (15).
To investigate the effect of inductance in the BLSC-
HESS, the inductance L1 was changed from 0 mH in Fig. 12a
to 0.12 mH in Fig. 12b and to 0.47 mH in Fig. 12c. The other
parameters were kept the same (Csc = 4 F and Rb = 0.1 Ω).
To simulate a high-frequency fluctuation in the system, a
high-frequency current iDG1 of 100 Hz was injected into the
DC bus, which results in a ripple in the output current iHESS1
of the HESS1 converter. With the conventional configuration
( L1 = 0 mH ), the gain of GiBSC (s) at 100 Hz is higher than
− 20 dB, as shown in Fig. 5. The battery current iBAT1 has a
high fluctuation, as shown in Fig. 12a. When L1 = 0.12 mH,
the gain of GiBLSC (s) at 100 Hz is reduced, which results in
the lower current fluctuation in Fig. 12b. With a much higher
Fig. 12  Case II: impacts of the inductance L1 on simulation results: a
without L1; b L1 = 0.12 mH; c L1 = 0.47 mH
inductance of L1 = 0.47 mH , the battery current fluctuation
is reduced, and the gain of GiBLSC (s) is around − 28 dB.
Case III To evaluate the proposed BLSC-HESS configu-
ration in parallel operation, two BLSC-HESSs and two dis-
tributed generators were used. The two distributed genera-
tors inject a total current of 10 A into the DC bus, which is
connected to a load of 10 Ω. The power sharing between the
HESSs is ensured by the droop controller. Good power shar-
ing is achieved with the same battery output currents iBAT1
and iBAT2, as shown in Fig. 13. Moreover, the current sharing
between the battery and supercapacitor in the two HESSs
shows good performance with the same settling time ts. As
a result, the proposed HESS configuration effectively oper-
ates in parallel when using the conventional droop control.
The simulation results show that the proposed BLSC-
HESS configuration is applicable, and that it has advantages
in terms of a lower cost and a simple control scheme. It can
operate with the conventional droop control method or any
advanced coordinating methods in a DC microgrid.
5 Experimental results
To validate the proposed BLSC-HESS configuration, the
DC microgrid in Fig. 1 with two HESSs and one distrib-
uted generator was implemented in the laboratory. The
two BLSC-HESS converters in Fig. 4 were controlled by a
TMS320F28379D DSP controller board, and another DSP
control board was used for the converters of the distributed
generator. The parameters of the DC microgrid are the same
as those used in the simulations. The experimental setup is
Fig. 13  Case III: simulation result of the cooperation of two HESSs
in a system
13
316 D.-H. Dam, H.-H. Lee

shown in Fig. 14. The distributed generation converter is

controlled to inject a current of 4.5 A into a DC bus that is
connected to a load of 8 Ω. The internal resistance of the bat-
tery is 0.3 Ω, and it can be changed by adding a serial resis-
tor to the battery. Three case studies were carried out to eval-
uate the characteristics of the proposed HESS configuration.
Case I Only one BLSC-HESS is activated, and the
parameters of the BLSC-HESS are L1 = 0.47 mH and
Csc = 28 F . When the load is connected to the DC bus, the
HESS changes from the charge mode to the discharge mode.
As shown in Fig. 15a, the battery current iBAT1 smoothly
changes from the charge mode to the discharge mode thanks
to the pulsed supercapacitor current iSC1. Figure 15b, c shows
waveforms of the HESS converter output current iHESS1 and
DC bus voltage vbus, which are regulated by the droop con-
troller. In Fig. 15d, the supercapacitor voltage is equal to the
battery voltage in the charge mode and the discharge mode.
Thus, it is always kept in a safe operation range.
Fig. 15  Case I: experimental results of a single HESS in a system
Case II With the same setup conditions as case I, only
the parameters of the BLSC-HESS are changed to evaluate
its performance. The internal battery resistance normally Under an unbalanced AC load of an inverter in a DC
changes when the battery parameters change. However, in microgrid, harmonic and high-frequency currents are gen-
this experiment, it is changed by adding an external serial erated on the DC bus. To evaluate the effectiveness of the
resistor. Figure 16 shows the impacts of the supercapacitance BLSC-HESS configuration in a high frequency range, a
Csc and the battery internal resistance Rb. Figure 16a shows harmonic current iDG1 of 100 Hz was injected into the DC
the supercapacitor and battery currents of case I, for which microgrid through the distributed generation converter.
the settling time is 20 s. When the internal battery resist- As a result, the output current iHESS1 fluctuated to regu-
ance is changed by adding an external resistance of 0.1𝛺 , late the DC bus. In Fig. 17, all three tested cases have the
the settling time becomes longer in Fig. 16b than it is in same amount of injection current. The inductance L1 was
Fig. 16a. The settling time is increased from 20 to 25 s when increased to investigate the effect on the BLSC-HESS.
the internal resistance is increased from 0.3 to 0.4 Ω, which Figure 17a shows the output currents of the supercapacitor
coincides with the calculation in (16). When the superca- and the battery without an inductor L1. The supercapacitor
pacitance Csc is increased from 28 to 56 F, the settling time
is increased from 20 to 40 s, as shown in Fig. 16c.

Fig. 16  Case II: impacts of the SC capacity Csc and the battery inter-
Fig. 14  Experimental setup nal resistance Rb on experimental results

13
Battery–inductor–supercapacitor hybrid energy storage system for DC microgrids 317
Fig. 17  Case II: impact of the inductance L1 on experimental results:
a without L1; b L1 = 0.47 mH; c L1 = 1.12 mH
cannot help to reduce the fluctuation in the output current
iBAT1 of the battery. By adding an inductance of 0.47 µH, the
fluctuation of the battery current iBAT1 is reduced, as shown
in Fig. 17b. In Fig. 17c, the fluctuation of the battery current
iBAT1 is much smaller then when the inductance is increased
to 1.12 mH. The results in Fig. 17 show that the battery cur-
rent is smoothened by inserting the inductor.
Case III HESS1 and HESS2 were used to observe the
performance of the BLSC-HESS in parallel operation. Two
HESSs are controlled by droop control to share the load
power and to regulate the DC bus voltage. The output cur-
rents of the battery and supercapacitor of the BLSC-HESSs
are shown in Fig. 18a, b for HESS1 and HESS2. Since both
the HESSs have similar system parameters, the output cur-
rents of their batteries ( iBAT1 , iBAT2 ) and supercapacitors
(iSC1 , iSC2 ) are almost the same. Good sharing performance
is shown by the matching of the HESS output currents iHESS1
and iHESS2 in Fig. 18c. A stable DC bus voltage vbus under
load variation is shown in Fig. 18d.
These experimental results show the effectiveness of
the BLSC-HESS in smoothening battery current with both
low- and high-frequency currents. The results also verify
the theoretical analysis and design guidelines. The simplic-
ity of the configuration and controller make it easy to apply
and practical.
6 Conclusion
A simple BLSC-HESS configuration with reduced control
complexity and cost was presented in this paper. By insert-
ing an inductor between the battery and the supercapacitor,
the demand current is effectively decoupled into low- and
Fig. 18  Case III: experimental result of the cooperation of two
HESSs in a system
high-frequency current components. Therefore, the battery
handles the low-frequency current, while the supercapaci-
tor handles the high-frequency current. The battery life is
extended, thanks to the reduced magnitude of the fluctuation
of the battery current. In addition, the lifetime of the super-
capacitor is also increased since its voltage is ensured to be
within a safe operating range. The proposed BLSC-HESS
configuration is easily controlled by a conventional droop
controller. Thus, it can be directly applied to a DC microgrid
with extendable capabilities and is easily adaptable to old
or new systems. This study also offered a design process
for the proposed BLSC-HESS. Finally, the performance and
effectiveness of the system were verified by a simulation and
experimental results.
Acknowledgements This work was supported in part by the National
Research Foundation of Korea Grant funded by the Korean Govern-
ment (NRF-2018R1D1A1A09081779) and in part by the KETEP and
the MOTIE (No. 20194030202310).
References
1. Boicea, V.A.: Energy storage technologies: the past and the pre-
sent. Proc. IEEE 102(11), 1777–1794 (2014)
2. Hung, S.T., Hopkins, D.C., Mosling, C.R.: Extension of battery
life via charge equalization control. IEEE Trans. Ind. Electron.
40(1), 96–104 (1993)
3. Etxeberria, A., Vechiu, I., Camblong, H., Vinassa, J.M., Cam-
blong, H.: Hybrid energy storage systems for renewable energy
13
318 D.-H. Dam, H.-H. Lee

sources integration in microgrids: a review. In: 2010 Conference Industrial Electronics for Sustainable Energy Systems (IESES),
Proceedings IPEC, pp. 532–537 (2010) pp. 314–319 (2018)
4. Dougal, R.A., Liu, S., White, R.E.: Power and life extension of 20. Chen, M., Rincon-Mora, G.A.: Accurate electrical battery model
battery-ultracapacitor hybrids. IEEE Trans. Compon. Packag. capable of predicting runtime and I-V performance. IEEE Trans.
Technol. 25(1), 120–131 (2002) Energy Convers. 21(2), 504–511 (2006)
5. Smith, T.A., Mars, J.P., Turner, G.A.: Using supercapacitors to 21. Ran, L., Junfeng, W., Haiying, W., Gechen, L.: Prediction of state
improve battery performance. In: 2002 IEEE 33rd Annual IEEE of charge of lithium-ion rechargeable battery with electrochemical
Power Electronics Specialists Conference. Proceedings (Cat. impedance spectroscopy theory. In: 2010 5th IEEE Conference on
No.02CH37289), vol. 1, pp. 124–128 (2002) Industrial Electronics and Applications, pp. 684–688 (2010)
6. Kwasinski, A.: Identification of feasible topologies for multiple- 22. Shen, J., Khaligh, A.: A supervisory energy management control
input DC–DC converters. IEEE Trans. Power Electron. 24(3), strategy in a battery/ultracapacitor hybrid energy storage system.
856–861 (2009) IEEE Trans. Transp. Electrif. 1, 223–231 (2015)
7. Liu, Y.C., Chen, Y.M.: A systematic approach to synthesizing
multi-input DC–DC converters. IEEE Trans. Power Electron.
24(1), 116–127 (2009) Duy‑Hung Dam received his B.S.
8. Behjati, H., Davoudi, A.: A multiple-input multiple-output DC– degree from the Ho Chi Minh
DC converter. IEEE Trans. Ind. Appl. 49(3), 1464–1479 (2013) City University of Technology,
9. Hintz, A., Prasanna, U.R., Rajashekara, K.: Novel modular multi- Ho Chi Minh City, Vietnam, in
ple-input bidirectional DC–DC power converter (MIPC) for HEV/ 2012. He is presently working
FCV application. IEEE Trans. Ind. Electron. 62(5), 3163–3172 towards his M.S./Ph.D. degrees
(2015) in the School of Electrical Engi-
10. Xu, Q., et al.: A decentralized dynamic power sharing strategy for neering, University of Ulsan,
hybrid energy storage system in autonomous DC microgrid. IEEE Ulsan, South Korea. His current
Trans. Ind. Electron. 64(99), 1 (2016) research interests include power
11. Xu, Q., Xiao, J., Hu, X., Wang, P., Lee, M.Y.: A decentralized sharing, power quality control
power management strategy for hybrid energy storage system with and the application of super-
autonomous bus voltage restoration and state of charge recovery. capacitors in AC/DC
IEEE Trans. Ind. Electron. 0046(9), 7098–7108 (2017) microgrids.
12. Kollimalla, S.K., Mishra, M.K., Ukil, A., Gooi, H.B.: DC grid
voltage regulation using new HESS control strategy. IEEE Trans.
Sustain. Energy 3029(c), 1 (2016) Hong‑Hee Lee received his B.S.,
13. Yuhimenko, V., Lerman, C., Kuperman, A.: DC active power M.S. and Ph.D. degrees in Elec-
filter-based hybrid energy source for pulsed power loads. IEEE trical Engineering from Seoul
J. Emerg. Sel. Top. Power Electron. 3(4), 1001–1010 (2015) National University, Seoul,
14. Hamzeh, M., Ghafouri, M., Karimi, H., Sheshyekani, K., Guer- South Korea, in 1980, 1982 and
rero, J.M.: Power oscillations damping in DC microgrids. IEEE 1990, respectively. From 1994 to
Trans. Energy Convers. 31(3), 970–980 (2016) 1995, he was a Visiting Professor
15. Zhao, Xin, et al.: Energy management strategy of multiple super- at Texas A&M University, Col-
capacitors in a DC microgrid using adaptive virtual impedance. lege Station, TX, USA. Since
IEEE J. Emerg. Sel. Top. Power Electron. 4(4), 1–8 (2016) 1985, he has been with the
16. Gu, Y., Li, W., He, X.: Frequency-coordinating virtual impedance Department of Electrical Engi-
for autonomous power management of DC microgrid. IEEE Trans. neering, University of Ulsan,
Power Electron. 30(4), 2328–2337 (2015) Ulsan, South Korea, where he is
17. Zhang, Y., Li, Y.: Energy management strategy for supercapacitor presently working as a Professor
in autonomous DC microgrid using virtual impedance. In: IEEE in the School of Electrical Engi-
Applied Power Electronics Conference and Exposition (APEC), neering. He is also the Director
pp. 725–730 (2015) of the Network-Based Automation Research Center, which is sponsored
18. Zhao, X., Wu, X., Li, Y., Tian, H.: Energy management strategy by the Ministry of Trade, Industry and Energy. His current research
of multiple supercapacitors in an autonomous DC microgrid using interests include power electronics, network-based motor control and
adaptive virtual impedance. In 2016 IEEE 7th International Sym- renewable energy. Dr. Lee is a Member of the Korean Institute of Power
posium on Power Electronics for Distributed Generation Systems Electronics (KIPE), the Korean Institute of Electrical Engineers, and
(PEDG), pp. 1–8 (2016) the Institute of Control, Robotics and Systems. He was the President
19. Dam, D.-H., Hoang, D.-K., Chun, T.-W., Lee, H.-H.: A hybrid of KIPE in 2014.
energy storage system for transient load and its multiple opera-
tion in DC microgrid. In: 2018 IEEE International Conference on

13