IEEE PEDS 2019, Toulouse, France

9 – 12 July 2019

Current Sharing Control of Parallel Bidirectional

DC-DC Converter for Energy Storage Applications
in Islanded Microgrid
Monica P M. Kowsalya
School of Electrical Engineering School of Electrical Engineering
VIT University VIT University
Vellore, TamilNadu, India-632014 Vellore, TamilNadu, India-632014
[email protected] [email protected]

Abstract— The paper explores parallel connected bidirectional storage, less loss and power conversion problems [3]. To
DC-DC converters for energy storage applications in microgrid. meet the discrepancies in the load demand during intermittent
The parallel operated bidirectional DC converter confronts with power generation, the standalone DC microgrids badly rely
inaccurate current sharing due to the voltage imbalance and the upon the energy storage systems (ESS) that acts as backup
circulating current. The parallel DC–DC power stage and its source. A hybrid microgrid model with energy storage is
control in the micro grid proves to be more prominent and its
depicted in Fig.1.
efficiency is predetermined by the factors like the minimum
circulating current and voltage balance. This paper aims at
achieving accurate current sharing with a power based droop DG1 DG2 DGn
mechanism. The cynosure of the current work is carried out in
two stages involving the proposed droop mechanism and the
cascaded control. Due to the self-regulating voltage gain, = = =
~ ~ ~
bidirectional capability, the dual active bridge bidirectional DC- AC
Load
AC
Load
AC
Load
DC converter is chosen for the current work. The proposed
droop strategy is deployed for battery fed converters to share
AC BUS
power irrespective of instantaneous estimation of state of charge AC BUS AC BUS

of batteries. This can be achieved by deriving a relation between

the output power and the state of charge of batteries and ~ ~ ~
= = =
formulating a droop equation based on power. The system is
simulated in the MATLAB Simulink environment and the COMMON DC BUS

results were analyzed experimentally.

DC
Keywords—Bidirectional DC converter, Droop control, State Load

of charge, Battery, Distributed energy

I. INTRODUCTION
The need for global environmental conservation and
damage control due to emissions of fossil fuel, invoked
interest in the production of clean and green energy in the Storage subgrid
power sector [1]. The depletion rate of fossil fuels and the Fig. 1. Structure of hybrid Microgrid
demand of green energy drove the power industry to focus on The literature [4] reviews various categories of energy
the power generation with renewable sources such as wind, storage system designs and their advantages. Among the
solar etc. Democratization of power at domestic level through various storage systems like batteries, flywheel, super
the effective utilization of renewables led to the concept of capacitors; the battery energy storage systems (BESS) have
distributed energy generation (DER) with Microgrid / Mini- become prominent and most preferable due to its efficiency,
grid. reliability and flexibility.
A Microgrid is defined by the ability of micro sources to The localized energy exchange between the energy storage
meet the power demand of the load irrespective of its device and the load requires an efficient bidirectional
connection with the utility grid [2]. The network of converter. The BESS integrated to bidirectional DC- DC
distributed generators supplying the load autonomously is converters (BDC) connected to a common bus forms a
termed as islanded Microgrid. The islanded operation of storage sub grid. The modular approach of using battery
Microgrid typically requires a power electronic interface, energy storage system increases the reliability and efficiency
which is a prime decisive factor in determining the type of of the storage sub grid [5]. It also makes the system
Microgrid. Microgrid may be of AC, DC, or hybrid based on upgradable and reconfigurable for future higher power level
the output of the power electronic interface. Capability to requirements of micro grids. Hence a system with batteries
supply the DC power electronic devices which represent 50% interconnected via BDC at the common bus as shown in Fig.
of the load in the domestic house hold is one of the main 2 is adopted for study in the current research work.
factors that provoked research interest into DC microgrids. It During erratic conditions, due to inconsistent generation
also holds the merits like superior compatibility of power and technical failure, the DGs could not supply power to the

978-1-5386-6499-5/19/$31.00 ©2019 IEEE

load connected at the common bus. So as to meet the load the proposed control strategy. Section IV discusses the
demand, the batteries discharge via BDC whose output simulation results and Section V concludes the research
terminals are connected to the common bus. The BDC work.
operate in parallel while supplying modular power from the
battery to load [6]. II. CHOICE OF CONVERTER AND DESIGN
The parallel operated BDC poses problem of unequal The converter topology chosen for the proposed work is
output current at the common bus due to following reasons: depicted in Fig. 3. The evolution of bidirectional DC-DC
• Slight variation in the supply voltage due to state of converters could find its way classified in the literature as
charge of battery. isolated and non-isolated. The isolated bidirectional DC-DC
• Different parasitic resistances connected between converters gained prominence due to its voltage matching and
the output terminals of BDC and the common bus galvanic isolation provided by means of a transformer. The
This tends to create a voltage imbalance at the common dc bidirectional converter proposed in the research work [15]
bus and circulating current flows between the modules of features series resonance and boost capability with simple
battery storage system. The most possible solution is to PWM. The isolated bidirectional converter proves to
reduce the circulating current and to maintain equal voltage
at the output terminals of the BDC. This can be achieved by
the proper selection of power converter and appropriate
control strategy. The dual active bridge BDC is chosen due to
its self-regulating bidirectional voltage gain capability [7].
Regarding the control strategy for power sharing,
the wireless droop control is widely employed. Although the
droop control strategy is simple and communication free, the
power sharing is achieved at an expense of voltage drop [8]. Fig 3. Isolated Bidirectional DC-DC Converter[15]
The coordinated control method proposed in [9] is a A. Converter design
combination of droop and feed forward control which The hybrid converter resembles an isolated resonant
requires tedious calculations. Very few literatures investigate converter of [14], except for the active bidirectional switch
the control of BDC for energy storage in islanded microgrid. (S5) at the secondary side of the galvanic isolation
In the literature [10-11] for energy storage applications, the transformer. The design of transformer, resonant inductor Lr,
droop control is employed to nullify the difference of the state resonant capacitors Cr1, Cr2, and the gating scheme is adapted
of charge (SOC) of parallel connected battery. The virtual from the literature [15]. The design procedure involves
resistance-based droop control of parallel bidirectional selection of transformer turns ratio and the design of the
converter was proposed in the literature [12]. The SOC based resonant tank. Turns ratio of transformer is given by
adaptive droop inclusive of virtual resistance is proposed in Vo (1)
n =
[13]. The research work suggested in the literature, so far 2 V in
requires estimation of SOC. The novel droop strategy In the resonant tank design, the resonant inductor Lr is
proposed in the current work holds the advantage of independent of the conversion ratio since it acts as boost
independence of state of charge estimation. This paper inductor. A large enough inductor was chosen based on the
reports power-based droop function for reference voltage longer switching state of PWM boost mode and the resonant
generation used in processing the voltage and current error of mode. The length of switching time is less than switching
the BDC. The developed droop control strategy, drastically periods Ts of all operating ranges. The Lr, Cr, Lm should be
reduces circulating current and improves the current sharing chosen such that they satisfy the equations below.
between the converters irrespective of distinctness of the PT
supply side parameters. The research work is carried out C r > o 2s (2)
Vo
assuming that the DGs are inoperable due to a technical
failure and the dc load is being supplied by the batteries V o2 (3)
Lr >
during discharge condition from storage sub grid. 2 ω r2 Po T s
n 2 T DT (4)
Lm ≤
4C o f s
The resonant capacitance can also be chosen based on the
resonant frequency ω r ,
DC bus

1 (5)
ω r = 2π f r =
Lr (C r 1 + C r 2 )
Load
During resonant mode of operation, the equivalent circuit of
the converter is as shown in Fig 4.
Lr
iLr Lr
Battery Bidirectional DC iLr 2Cr
converter
nVin Vcr1 2Cr
Fig 2. Structure of system for study nVin
Vo
The paper is organized as follows: Section II describes the
choice of converter and its design and section III contributes Fig. 4. The equivalent circuit of the converter during resonant state [15]
 TABLE I. CHARACTERISTIC COMPARISON TO ILLUSTRATE THE SUPERIOR PERFORMANCE OF CONVERTER
Characteristics Isolated Converter with active full bridge DC-DC converter
Traditional active bridge Series resonant LLC type converter Bidirectional DC-DC Converter
DC-DC converter [8] converter [10] [15] Chosen for the current work [16]
Soft switching range Narrow Narrow wide wide
Voltage stress high Low high low
Active and Passive elements moderate High High Moderate
count
Efficiency Low High Low high
Control complexity moderate Complex Simple Simple
Circulating current due to yes No yes No
leakage inductance and active
full bridge

At the primary side of the transformer, the switches of full balance at source side, droop control at the load side is
bridge achieves zero voltage switching and the diodes at the proposed for accurate current sharing and to regulate the DC
secondary side achieve zero current switching by means of bus at 220V.
magnetizing current for all ranges of load.
A. Mathematical analysis of control
The duty cycle of the bidirectional switch is the
control input, whose switching frequency is equal to twice the An indirect relation between SOC and output power is
resonant frequency. When this switch is activated, the derived. The SOC of the battery is in direct proportional to
leakage inductor Lr acts as boost inductor and resonates the the power, the proposed droop equation is based on power.
capacitance Cr1, Cr2. The duty cycle of the bidirectional Since the energy storage units are identical, the battery
switch is as follows. voltage Vin1, Vin2 at the input of BDC 1 and 2 are as follows
1 2 Lr Po 2 nV in V in1 = V in 2 (7)
D= (1 − ) (6)
nV in Ts V0 According to conservation of energy
Pin = Pout (8)
The proper design of Lm and Lr results in large Lm/Lr ratio that
The SOC of the battery is given by
yields highly efficient power conversion. The converter
SOC final = SOCinitial −  out 
P
efficiency further depends on the duty cycle and gating
sequence that determines the resonant operation. The design
 
(Vdc * Cb ) (9)

parameters considered for simulation is shown in Table I. In Hence according to literature [6]
contrast, the trajectory path of state variables during Vdci = Vref −  Rdroop * I dci  (10)
resonance can also be utilized to determine the duty cycle.
The gating sequence in the current work is adapted (Vdci * I dci ) = I dci * (Vref −  Rdroop * I dci  ) (11)
from the literature [15]. The pulse generated triggering the
full bridge at the primary side of the isolation transformer. By (
Pdroopi = I dci * Vref −  Rdroop * I dci  ) (12)
incorporating the required phase delay to the pulses generated
During discharge mode of battery.
for the full bridge at the primary of the transformer, the
bidirectional switch at the secondary of the transformer is  1 
n
 1 
Rdroop ∝   ∝  (13)
triggered. The Fig. 5 demonstrates the gating pattern of the  SOC   Pdroop 
bidirectional DC-DC converter
Where n is the convergence factor that decides the rate of
discharge, i represents the number of converters, Vdci
represents output voltage of BDC,
Thus from Eq. (13), it can be inferred that
 1 
n
 1 
 soc  ∝   (14)
   Pdroop 
(
Hence, Pdroop )i ∝ (SOCi )n (15)

[SOCi ] ∝ [ Pdroop  ]

1/ n
(16)
i

[SOCi ] ∝ [Pref − Pdc ]i1/ n (17)

[SOCi ] = k * [Pref − Pdc ]i1/ n

Fig. 5. Gating pulse pattern of Bidirectional DC-DC converter
(18)
III. PROPOSED CONTROL STRATEGY
Without estimating the instantaneous SOC of the battery,
The control method proposed in the current work is carried approximate droop equation is proposed.
out assuming the energy storage units are identical and load The droop equation is modelled as follows.
as constant. Irrespective of characteristics of battery like   
1   P  (19)
SOC, charging current, a simple and efficient droop control V ref = V dci − K    ln  dci 
 n  i   P ref 
method is proposed in the current work. Instead of the charge    
The optimal value of convergence factor is chosen to speed 1.2
up the rate of discharge. The droop factor K is determined
1
based on the droop characteristics. The proposed droop

Factor [k/n]
0.8
mechanism generates the voltage reference based on the
proportionate power. The voltage error corresponding to the 0.6
droop reference and the instantaneous output voltage of the 0.4
converter is processed via proportional-Integral (PI) 0.2
controller and the current error is processed with PI control. 0
The control parameters are tabulated in table III. The overall 0 2 4 6 8 10 12 15
control diagram is as shown in Fig. 6. Voltage deviation

Fig.7. Variation of factor [k/n] with respect to voltage deviation

Control loop
TABLE I. SIMULATION PARAMETERS
Battery 1 Simulation Parameters
S.NO
Parameter Value
BDC
Supply
VDC =48v
voltage
Switching
20KHz
LOAD frequency Fs
Power P=300W;
Bidirectional frequency ft=130KHz
BDC DC-DC Transformer V1=40V, R1=0.002Ω
Converter rating ,L1= 0.008H
Battery 2 Vref, Iref 1. V2=200V,R2=0.002Ω
,L2=0.008H
PWM Lr 33.5 μH
Vdc1,Idc1
Cr1, Cr2 100nF
PI Controller C0 1.2 μF
Idc1 2. Battery 1 C1 12Ah
_
+ Vdc1 C2 12Ah
_ Proposed 3. Battery 2
PI Power based 10 ohms (equal)
+ Line
Controller droop control 4. R 1, R 2 10 ohms, 12 ohms
resistance
Control loop (unequal)
Fig 6. Over all control diagram Load
5. RL 120 ohms
resistance
B. Effect of selection of parameters k, n.
IV. SIMULATION RESULTS AND DISCUSSION
a) Convergence factor ‘n’: The selection of
convergence factor ‘n’ determines the speed of the A. Simulation Study
equalization of state of charge. From the mathematical
analysis, it is determined that the convergence factor ‘n’ is The proposed droop control for power sharing in dc
inversely proportional to the voltage deviation. For identical micro grids effectively reduces the circulating current. The
batteries, with different SOC, higher the convergence factor, simulation parameters are tabulated in table II and the
faster the rate of convergence and lower the deviation. For simulation is carried out based on the variation with SOC and
identical batteries with same SOC, the convergence factor is line resistance as depicted in the table III. The control
assumed as equity. parameters are put forth in table IV.
b) Droop factor ‘k: The droop factor is determined TABLE III. CASE STUDY OF THE SIMULATION
based on the fact that the droop power is directly proportional
to voltage deviation. The tangent through the curve of the S. SOC of Line resistance connected to load
graph of droop power versus the voltage approximates the ‘k’ No batteries
1. Equal Different
factor.
2. Different Equal
c) The factor (k/n): The ratio k/n has to be optimum
to avoid the problem of instability. The factor k/n is Case (1): SOC of batteries are equal and the line resistance
approximated as follows are different
k 
(
ΔVd = −   ln Pdroop ) (20)
The SOC of batteries are maintained at 80% for batteries 1
and 2. The line resistance being connected to the load bus are
n
The maximum allowable deviation is 5% of the load bus different and are 10 and 15 ohms respectively of BDC1 and
voltage BDC2. The non-identical line resistances tend to create
voltage imbalance at the common bus connected. Hence the
e (Δ V d )  k  circulating current flows due to the difference of the output
= ≤1 (21)
Pdroop  n  voltages at the converter output. For the ease of simulation
The Fig. 7 depicts the characteristics of the factor (k/n) with battery response time is given as 10 seconds.
respect to voltage deviation, ( ΔVd ).
 TABLE V SIMULATION RESULTS
Case Idc1, Idc2, ILoad Vdc1, Vdc2, VLoad Current
sharing
error
Case (1) 1.45,1.4,2.85 227,225, 216 1.7%
Case (2) 1.5,1.35,2.85 228,231,219 5.2%

B. Experimental Validation
The experimental validation is carried out with the hardware
in loop simulation of the parallel converters in digital signal
processing environment using TMS320F28335. The battery
banks of 12V, 12Ah of 4 each, are connected in parallel to
obtain a supply voltage of 48V. The state of charge of
batteries are maintained at 80%. The proposed control
strategy has been investigated with 48V source voltage fed
to BDC1 and BDC2, assuming the SOC of batteries are the
same. The experimental setup is as shown in Fig. 9. The result
of the current sharing is as tabulated in Table VI as follows.

Fig 7. Simulated results of parallel BDC with identical SOC

TABLE IV. CONTROL PARAMETERS

S. No Control Parameters
1. Droop factor K 0.016
2. Convergence factor n 1
3. Voltage controller Kp, Ki 1.2, 0.9
4. Current controller Kp, Ki 0.56, 0.04

Case (2): Line resistances being connected to the common

load are equal and SOC are different
The initial SOC of the batteries 1 and 2 are maintained at 80%
and 60% respectively. Since the SOC are different the
discharge current being fed to the converter varies. The
dissimilar output currents of the BDC1 and BDC2 creates an
imbalance at the common DC bus. The circulating current
flows between the converters. The simulation results are Fig 9. Experimental setup of the Parallel BDC connected to Battery
tabulated in Table V. The current sharing error with SOC of
both batteries equal and line resistance unequal is 1.7% and
5.2% for the vice versa case. TABLE VI EXPERIMENTAL RESULTS

Case Idc1, Idc2, ILoad Vdc1, Vdc2, VLoad Current

sharing
error
Case (1) 1.5,1.4,2.85 235,232,225 3.5%
Case (2) 1.6 ,1.4,2.85 238,236,224 7.8%

Fig. 10. Current sharing waveform without controller. (x-axis: 10ms/div; y-

axis: 2v/div)
Fig 8. Simulated results of non-identical SOC
The current output of the BDC1 and BDC2 hauls an
irregular current sharing pattern which range from 35% to
68% without the controller. The Fig.10 represents the current REFERENCES
sharing error of 35%. [1] S. R. Bull, “Renewable Energy Today and Tomorrow,” vol. 89, no. 8,
2001.
[2] Dragičević, Tomislav, Xiaonan Lu, Juan C. Vasquez, and Josep M.
Guerrero, "DC microgrids—Part II: A review of power architectures,
applications, and standardization issues," IEEE Trans. on power
electronics, vol. 31, pp. 3528-3549, 2016.
[3] Monica, P., and M. Kowsalya. "Control strategies of parallel operated
inverters in renewable energy application: A review," Renewable and
Sustainable Energy Reviews, vol. 65, pp. 885-901, 2016.
[4] Jing W, Lai CH, Wong SH and Wong ML, "Battery-supercapacitor
hybrid energy storage system in standalone DC microgrids: a
review," IET Renewable Power Generation, vol. 11, no.5, pp. 461-469,
2016.
[5] Rehman, M.M.U., Zhang, F., Zane, R. and Maksimovic, D.,” Control
of bidirectional DC/DC converters in reconfigurable, modular battery
Fig 11. Current sharing waveform of parallel BDC for different SOC and systems,” In Applied Power Electronics Conference and Exposition
uniform line resistance (APEC), pp. 1277-1283, IEEE. March, 2017.
[6] P.C. Loh, D. Li, Y.K. Chai, and F. Blaabjerg, "Autonomous operation
of hybrid microgrid with AC and DC subgrids," IEEE Trans. on power
electronics, vol. 28, no. 5, pp. 2214-2223, 2013.
[7] Tian, J., Ren, C., Jia, Y., Han, X. and Wang, P., “Control strategy of
ILoad (2.85A) HESS with dual active bridge bidirectional DC/DC converter in DC
microgrid,” In Industrial Electronics and Applications (ICIEA), 2017
Idc2(1.5A) 12th IEEE Conference ,IEEE, pp. 1449-1453, June, 2017.
[8] Monica, P., M. Kowsalya, and P. C. Tejaswi. "Load sharing control of
Idc1(1.4A) parallel operated single phase inverters." Energy Procedia, vol. 117,
pp. 600-606, June, 2017.
[9] Xia, Y., Peng, Y., Yang, P., Yu, M.,and Wei, W.,. "Distributed
coordination control for multiple bidirectional power converters in a
hybrid AC/DC microgrid." IEEE Trans. on Power Electronics, vol. 32,
no.6 , pp. 4949-4959. 2017.
[10] Lu, X., Sun, K., Guerrero, J. M., Vasquez, J. C., & Huang, L, "state-of-
charge balance using adaptive droop control for distributed energy
storage systems in DC Microgrid applications." IEEE Trans. Industrial
Electronics, vol. 61, no. 6, pp.2804-2815, 2014.
Fig 12. Current sharing waveform of parallel BDC for uniform SOC and
different line resistance [11] Hu, Rui, and Wayne W. Weaver.,"Dc microgrid droop control based
on battery state of charge balancing." Power and Energy Conference at
Illinois (PECI), IEEE, pp.1-8, Feb, 2016.
Current sharing error analysis carried out clearly illustrates
[12] Dragicevic, T., Guerrero, J. M., Vasquez, J. C., & Skrlec, D,
that current sharing error is 3.54% for BDC fed from batteries "Supervisory control of an adaptive-droop regulated DC microgrid
with uniform SOC and 7.8% for BDC fed from batteries with with battery management capability." IEEE Trans. Power Electron ,
different SOC. The Fig 11. And Fig 12. Represents the vol. 29, no. 2, pp. 695-706, 2014.
current sharing pattern for case (1) and case (2). There exists [13] H.D.Nguyen, and H.T. Nam. "Adaptive coordinated droop control for
a slight depression in load voltage of about 0.05A, due to the multi-battery storage," EUROCON 2015-International Conference on
Computer as a Tool (EUROCON), IEEE. IEEE, pp.1-6, Sep. 2015.
losses incurred by the current transducers which can be
[14] H. Hu, X. Fang, S. Member, F. Chen, and S. Member, “A Modified
neglected. High-Efficiency LLC Converter With Two Transformers for Wide
Input-Voltage Range Applications,” vol. 28, no. 4, pp. 1946–1960,
V. CONCLUSION 2013.
[15] T. Labella, S. Member, W. Yu, J. J. Lai, M. Senesky, D. Anderson,
In this paper, an independent droop control strategy and A. Modular, “A Bidirectional-Switch-Based Wide-Input Range
irrespective of charge balancing at the input source is High-Efficiency Isolated Resonant Converter for Photovoltaic
proposed to address the problem of load sharing. Applications,” vol. 29, no. 7, pp. 3473–3484, 2014.
Mathematical approximation illustrated in the paper [16] K. Lüthje, M. Albert, M. Carmen, A. E. Michael, K. L. Jørgensen, M.
approximates the droop equation independent of state of C. Mira, Z. Zhang, and M. A. E. Andersen, “Review of High Efficiency
Bidirectional dc-dc Topologies with High Voltage Gain,” 2017.
charge of the batteries. As a result, the estimation of the state
of charge of batteries is avoided. The effect of selection of the
droop factor and the convergence factor on the voltage
regulation is explained in the paper. It suggests that, higher
the droop factor, higher will be the voltage deviation and vice
versa. The paper achieves load sharing with the logarithmic
power droop being proportional to the voltage deviation. This
tends to approximate the nonlinear characteristics of battery
to linear form. The proper selection of parameters like K, n
helps to restore the voltage deviation in acceptable range. The
simulation results and experimental validation suggest that
the proposed droop method achieves accurate power sharing.