This article has been accepted for publication in a future issue of this journal, but has not been

fully edited. Content may change prior to final publication. Citation information: DOI 10.1109/TIA.2017.2756032, IEEE
Transactions on Industry Applications

Dynamic Power Management and Control of PV

PEM fuel Cell based Standalone AC/DC Microgrid
Using Hybrid Energy Storage
Rishi Kant Sharma, Student Member IEEE and Sukumar Mishra, Senior Member IEEE

Abstract – In this paper, dynamic power management scheme - intermittency of renewable energy sources and uncertain
is proposed for standalone hybrid AC/DC microgrid which nature of load variation [1]. The PEM fuel cell (FC) is an
constitutes photovoltaic (PV) based renewable energy source, electrochemical device which provides a reliable steady state
proton exchange membrane (PEM) fuel cell (FC) as a secondary
power however, is unable to meet the power transients owing
power source and battery-supercapacitor as hybrid energy
storage. The power management algorithm accounts for seamless to its slow response of internal thermodynamic and
operation of microgrid under various modes and state of charge electrochemical process [3]-[6]. The system is often subjected
(SoC) limit conditions of hybrid energy storage, when all the to sudden change in load and source powers (PV) which leads
sources, storages and loads are connected directly at the dc link. to high fluctuation at dc link voltage and may affect the MG
The power management scheme (PMS) generates current performance [1]. To avoid such contingencies, hybrid energy
references for dc converter current controllers of fuel cell, battery
storage (HES) which is combination of high energy density
and supercapacitor. The average and fluctuating power
components are separated using moving average filter. The dc and high power density needs to be introduced to
link voltage regulation under dynamic changes in load and source supply/absorb steady state and transient power components in
power variation is proposed. Also, PV power curtailment through MG [8]-[9]. Thus, hybrid power source comprising PV as a
control is formulated. The proposed power management is primary source and fuel cell as a secondary source along with
modified and extended to multiple photovoltaic generation battery-supercapacitor (SC) as HES is a promising
system and batteries with all the sources and storages
combination to operate the system in standalone mode [7].
geographically distributed operating under multi-time scale
adaptive droop based control with supervisory control for mode The dynamics of variation in ac/dc load as well PV
transition. The proposed power management scheme is validated power are reflected at dc link voltage. Thus, its control plays
using simulation results. Also, FPGA/Labview based laboratory determined role in dynamic power management of MG. In [6],
scale experimental results are presented to validate the power fuzzy logic controller based on flatness property for dc voltage
management scheme under various critical conditions. regulation is used for PV, FC and SC standalone system. The
power reference of FC is obtained using low pass filter (LPF)
Index Terms—PV, PEM fuel cell, power management,
and does not consider the over and under-utilization of H2. The
supercapacitor, voltage source converter, standalone AC/DC
microgrid, Moving average filter, Multi-time scale, droop. LPF introduces lag in the reference power generation of FC so
SC has to supply both transient and constant power. Due to this
I. INTRODUCTION SC voltage will hit its high/low limit frequently. In [7], with
application of PV, FC, and battery-SC for DC system, has
The demand of Hybrid AC/DC microgrid (MG) has
reduced the burden on SC by using battery-SC combination.
increased significantly with increased penetration of
The operation of PV at off maximum power point (MPPT)
renewable energy sources such as photovoltaic array (PV) at
based on power management scheme is the key issue which
low voltage ac distribution sector. Increased number of dc
should be considered carefully. In [10], operation of PV, FC
loads such as plug-in vehicle, telecom load, central computer
and SC is presented for islanded microgrid under unbalanced
center, emphasizes the need of evolving microgrid suitable for
and nonlinear load condition. The paper lacks the effective
both ac and dc loads [2]. The key challenges involves
energy management.
regulation of voltage and frequency of ac microgrid, dc link
The separation of average and transient power
voltage regulation for both ac/dc system, unbalanced load
component for battery and SC is also a key issue. In [14],
operation as well as dynamic power balance due to-
wind/load power fluctuations are mitigated using battery-SC
This work is partially supported by DST, Govt. of India under its project “UK combination where the average current reference is obtained
India Clean Energy Research Institute” with project no. RP03413G and by passing load current through low pass filter (LPF). Also, in
partially supported by Department of Electronics and Information Technology [11], dc link voltage controller generates average component
(DEITY), Govt. of India under Visvesvaraya PhD Scheme. of current references through LPF. This LPF introduces
The authors are with Department of Electrical Engineering, Indian Institute of
Technology, Delhi, 110016, India. (email-rishikant.iitd@gmail.com, significant time lag and dominant pole near to origin which
sukumar@iitd.ac.in) may hamper system stability. In [12] and [13], moving average
filter (MAF) computes time average value and provides

0093-9994 (c) 2017 IEEE. Personal use is permitted, but republication/redistribution requires IEEE permission. See http://www.ieee.org/publications_standards/publications/rights/index.html for more information.
This article has been accepted for publication in a future issue of this journal, but has not been fully edited. Content may change prior to final publication. Citation information: DOI 10.1109/TIA.2017.2756032, IEEE
Transactions on Industry Applications

average current reference without much lag and instability. control with input current controller to operate the
The multi-time scale control presented in [20]-[21], DGs in distributed way is proposed. A novel MAF
depends on hierarchical control solving economic dispatch based droop is used for time-scaling of fast and slow
problem with exchange of power from grid as well as SoC DGs. The SoC based adaptive droop is proposed for
optimization. While the proposed work does not involves operation of multiple BES.
solving any economic dispatch problem, day ahead scheduling 6. For MG2, supervisory control based mode transition
and power exchange from grid, nor it relies on dedicated signal employing low bandwidth communication
central controller for its operation. For parallel operation and (LBC) is proposed to operate the sources/storages in
proper current sharing of multiple DC sources (DG) with droop mode/ MPPT Mode/ SoC control mode/
converters in a distributed way, the output voltage reference of Voltage control mode.
the converter operates in voltage droop mode defined by The Section II underlines the hybrid AC/DC microgrid
virtual resistance [17]. The multi-time scaling is categorized configuration 1, Section III explains the proposed power
as (i) slow time scale DGs and (ii) fast time scale DGs. The management scheme for MG1, Section IV and V deals with
conventional method to obtain muti-time scale droop for DGs hybrid AC/DC microgrid configuration 2 and its power
with different dynamic responses utilizes the concept of virtual management scheme followed by simulation/ experimental
output impedances where the droop constant is multiplied results in Section VI and Section VII concludes the work.
with low pass filter for slow time scale DGs and multiplied by
high pass filter for fast time scale DGs [15], [16] and [22]. II. HYBRID AC/DC MICROGRID CONFIGURATION 1
In this paper, two dynamic power management schemes The hybrid AC/DC microgrid (MG1) configuration 1 consists
are proposed for two different hybrid AC/DC microgrid of PV with boost converter and PEM fuel cell (FC) with boost
configurations. The first configuration (MG1) consist of single converter, both interfaced directly at dc link, as shown in Fig.
PV and PEM fuel cell based hybrid power sources, single 1. It is supported by hybrid energy storage (HES) comprising
battery and supercapcitor based HES with dc loads and three battery and supercapacitor (SC) interfaced directly to dc link
phase inverter fed ac loads. Here, all the sources and HES are through bidirectional DC-DC converters (BDC). The load
interfaced to dc link directly through their dc-dc converters, consists of dc load and three phase ac load fed by three phase
assuming they are at same geographical location, as presented voltage source converter (VSC). Fuel cell generates only
in Fig. 1. The second configuration (MG2), consist of multiple steady state power to meet the excess power demand in case of
PVs, single PEM FC, multiple batteries (BES) and single SC very less/no PV power is available. The HES supports both the
with dc and ac loads. Here, all the sources and storages are steady state as well as transient power changes in generation
interfaced to dc link through cable connected at their and loads. It also assists the fuel cell to slowly ramp up its
individual dc-dc converters such that they are geographically generation from minimum/zero value to the reference value.
distributed, as shown in Fig. 6.The proposed dynamic power The SC supports transient/fluctuating as well as oscillatory
management scheme (PMS) plays key role in dc link voltage power changes and is insufficient to supply/absorb constant
regulation, current references generation and reference current power changes (for long time duration) due to its low energy
tracking by current controller to drive dc-dc converters of PV, density. While battery support constant power changes due to
FC, SC and battery. The main contributions of the proposed its high energy density, it may also supply transient power
dynamic power management scheme for MG1 and MG2 are (only under crucial circumstances). The VSC operates in
1. For MG1, current reference generation for PEM FC, voltage control mode with fixed frequency obtained from
Battery and SC using single dc link voltage voltage controlled oscillator [1]-[2]. It supplies three phase ac
controller. load or three single phase ac loads. The PV power curtailment
2. For MG1, separation of average and transient current is obtained by operating it on the linear characteristic of PV
references using moving average filter (MAF). curve between MPP (Vmp) point and open circuit (Voc). The
3. For MG1, allocation of different current references to modelling of PV connected to boost converter for MPP,
input current controllers of dc converters of PEM FC Battery and SC connected to dc link through BDC and VSC
and HES by mode based power management control is discussed in detailed in [13].
algorithm to drive the system seamlessly from load
dominating condition to generation dominating Modelling and Control of PEM Fuel Cell
condition while maintaining power supply reliability Some of the important relations that describes the modelling
even if the battery SoC and SC voltage are in limit of PEM fuel cell are as follows [4]-[5]. Fuel cell stack current
condition. and hydrogen flow are related as
4. The operation of PEM FC with effective utilization of = ⁄2 = 2 (1)
H2 is also ensured. Also, control based de-rating Where , is the number of series fuel cells in the stack, is
operation of PV boost converter is presented. stack current (A), F is faraday’s constant (C/Kmol) and is
5. For MG2, muti-time scale adaptive droop based modeling constant (Kmol/(sA)-1).

0093-9994 (c) 2017 IEEE. Personal use is permitted, but republication/redistribution requires IEEE permission. See http://www.ieee.org/publications_standards/publications/rights/index.html for more information.
This article has been accepted for publication in a future issue of this journal, but has not been fully edited. Content may change prior to final publication. Citation information: DOI 10.1109/TIA.2017.2756032, IEEE
Transactions on Industry Applications

DC temperature, (2) can be written as

= − ! ln(% ) − ' (
Bus
Pp (3)
v PL1
LC
Where, B=0.04777 V and C=0.0136 A-1. The Nernst voltage
filter
Load
in terms of gas molarities is given by [4]
*+ / /3.5
= ) + log ) 01 21 77
Boost (4)
PV convert VSC Filter 3 Phase , /613
Array er PL2 Load
Pfc Where, is open cell voltage (V). Hydrogen utilization factor
DC (U) is defined as the ratio of H2 reacted inside the stack to the
Load
injected H2 into the tank and has value between 0 and 1.
8= 9 (
PEM Fuel Boost (5)
Cell converter Pbat
Psc
Values above 0.9 are indicative of over-utilization of hydrogen
leading to fuel starvation and reduces its life and performance.
On the other hand, values below 0.8 corresponds to under-
Battery
Bank
Bidirectional DC-DC
Converter
Bidirectional DC-DC Supercapacitor utilization of hydrogen which indicates the presence of excess
Converter Pack
hydrogen, leading to sharp rise in output voltage and reduction
Fig. 1. Hybrid AC/DC Microgrid configuration 1
in overall power efficiency. Thus, for optimal utilization of H2
Lfc in fuel cell stack, stack current is limited by following criterion
<= <=
Ifc
.:;01 .A;01
≤ ≤
PEM FC Rfc (6)
stack >? >?
The current reference _ obtained from the power
Df management scheme generates the H2 reference given by
2
Vfc Cdc
_ (7)
=
8C
PWM
In accordance to (6), the ramp rate of fuel cell reference current
Fuel cell
processor
is controlled by current slope limiter (Ampere/sec) whose
slope is based on current and power rating of PEM fuel cell
Ifc_ref PI [8]. Hence fuel cell avoids steep changes in load demands and
PM guarantees matching the reactant delivery rate and the usage
Fig. 2. PEM fuel cell with boost converter rate [9].
The PEM FC interfaced to dc link through boost
converters operating in current mode control is shown in Fig.2.
A conventional PI controller is implemented to track the
reference current signal obtained from power management
scheme and generates duty ratio to boost the fuel cell voltage
to reference dc link voltage. The transfer function of fuel cell
boost inductor current to converter duty ratio is given by [8],
% G
,F
M1 + N N O
D̂,F (G) J1 − L J1 − L (8)
=
HI (G) G P,F %N + G',F %N + J1 − L

Where, LFC, RFC, and IFC are the inductance, resistance and
current of boost converter inductor respectively, DF is
converter duty ratio, N QRH %N are dc link voltage and
Fig. 3. Bode plot of compensated loop gain and uncompensated loop gain capacitance respectively. The transfer function (8) is
compensated using proportional integral (PI) controller to
The Polarization curve of the fuel cell gives the fuel cell obtain stable compensated system as well as ensure zero or
voltage which is sum of the Nernst instantaneous voltage E, very less steady state tracking error. The compensated loop
activation overvoltage and ohmic overvoltage gain is stable with phase margin of 700 and bandwidth around
and is expressed mathematically as follows. 2500 Hz, as shown in Fig.3.
= + + (2)
Where, is function of oxygen concentration Co2 and , III. PROPOSED POWER MANAGEMENT SCHEME 1
is function of and and stack internal resistance The proposed dynamic power management scheme
Rint (Ω). Assuming constant O2 concentration and constant (PMS) for MG1 is presented in Fig. 4. It comprises of current

0093-9994 (c) 2017 IEEE. Personal use is permitted, but republication/redistribution requires IEEE permission. See http://www.ieee.org/publications_standards/publications/rights/index.html for more information.
This article has been accepted for publication in a future issue of this journal, but has not been fully edited. Content may change prior to final publication. Citation information: DOI 10.1109/TIA.2017.2756032, IEEE
Transactions on Industry Applications

references generation, mode selection power management Current

slope limiter Ifc_max
algorithm (PM), tracking of current references by current Ifc_r1
controllers and duty ratio generation for PWM switching
pulses for different converters interfaced to sources and HES. Vdc_r
I_avg
Ifc_min
I_ref
PI MAF
A. Reference Current Generation by PMS
Ib_r1
The power balance in hybrid AC/DC microgrid is Isc_r1

ensured by controlling power at the dc link. The total dc link Vdc

power (average and transient) is the summation of power Power Ifc_r
Current d_fc
injected by the primary and secondary source, difference of P Management controllers
based
power absorbed by ac and dc loads and sum/difference of Ib_max Ib_r2 Refrence current
Isc_r
Refrence
d_sc

power supplied/absorbed by the HES. The total power required SOCb_r PI

Generation current
For Ib_r tracking d_b
to maintain the dc link is categorized as (i) average power Ib_min
Fuel cell
component (S TU ) due to slow changes in PV/FC power and SOCb Isc_max
Battery
Isc_r2 And
load power, (ii) oscillatory power component caused due to Vsc_r Supercapacitor
PI Duty
instantaneous ac power (S ) of inverter fed load and (iii) Isc_min ratio
transient power component (S+ ) which arises due to sudden
Vsc
changes in PV power and ac/dc loads. Thus, the power balance
Ppv PL SOCb Vsc
equation of hybrid AC/DC Microgrid at dc link comprising
above three component is given by (9) Fig. 4. Proposed power management scheme for hybrid AC/DC microgrid 1.
SCT (V) + S (V) ± SX (V) ± SY (V) − SZ (V)
= SN (V) = [N \N
- the fuel cell under normal operating condition. The total
(9)
average reference current is to be supplied by secondary source
= S TU (V) + S (V) + S+ (V) FC, however it is unable to meet the demand instantaneously.
Where, SS[ (V),S (V), SX (V) and SY (V) are the PV, FC, SC Thus, the difference of the average reference current and
and battery power respectively, SZ (V) is the sum of ac and dc measured or reference fuel cell current is supplied to/by battery
load power, SN (V) is the total dc link power, [N QRH \N are bank of HES. Hence, the battery reference current is given by
the dc link voltage and current respectively. An \N is the a_`1 = _Q[b − _`1 (12)
effective reference current for PMS so denoted by \_ . It is Therefore, battery supplies/absorb average power and also
the summation of the average current reference \_ TU , assists PEM FC to meet the demand gradually under normal
oscillatory \_] and transient current reference \_ + . operation. The transient and oscillatory component of current
TN
reference (11) is added with proportional dc link voltage error
\_ = \_ TU + \_] + \_ + = ^ CTN + _ ([N − [N )
G (P) to improve dynamics of dc link voltage under transient
(10) [11]. This current serves as the supercapacitor reference
Where, CTN and TN are the proportional and integral gain current ( G _`1).
of the outer dc link voltage controller, [N is the reference G _`1 = _`c − _Q[b + d([N − [N ) (13)
value of dc link voltage. The average component of the To maintain the state of charge (SoC) of battery when it
reference current \_ TU is extracted by filtering \_ through reaches lower SoC limit condition (SoCBL), the battery
moving average filter which computes the time average of the reference current is obtained from SoC control loop which is
signal and does not introduces significant time delay unlike secondary control and operation is initiated by power
low pass filter [13]. The transient and oscillatory component management algorithm (PM). The error of reference SoC and
of reference current is obtained by subtracting average current estimated SoC is minimized using PI controller to generate
reference from total reference current, thus time scaled battery reference current ( a_`2) to charge the battery from the
references are obtained. available power sources.
\_] + \_ + = \_ − \_ TU (11) a_`2 = ^ _ (Ge f − Ge f )
CX +
X

Now, the fuel cell reference current is obtained by limiting the G (14)
ramp rate of \_ TU through current slope limiter as discussed
Where, CX and X are the proportional and integral gain of
SoC control loop of battery respectively, Ge f is the reference
in Section II. Also, the obtained FC current reference must be
SoC value of battery to be maintained. Similarly,
within the maximum/rated value (Ifc_max) and minimum
supercapacitor voltage control loop is considered to maintain
current limit (Ifc_min/zero) value based on rating of PEM fuel
_`1 is the fuel cell reference current obtained
the supercapacitor voltage when it reaches the minimum
cell. Thus,
_`1 varies slowly and remains unaffected
voltage limit (VSCL). The SC voltage control loop uses PI
controller to generate SC reference current G _`2 to charge
from PMS. The
by transients caused by sudden variation in load and generation
the SC to reference voltage.
powers. Now, battery reference current is obtained to assist -

0093-9994 (c) 2017 IEEE. Personal use is permitted, but republication/redistribution requires IEEE permission. See http://www.ieee.org/publications_standards/publications/rights/index.html for more information.
This article has been accepted for publication in a future issue of this journal, but has not been fully edited. Content may change prior to final publication. Citation information: DOI 10.1109/TIA.2017.2756032, IEEE
Transactions on Industry Applications

X
G _`2 = ^ _ ( gF − gF )
CX + Ipv
G
MPPT
(15) Vpv D1
Where, CX and X are the proportional and integral gain of Dpv
0 PM
SC voltage control loop respectively, gF is the reference SC
PI
D2
voltage to be maintained. Isc_r D3

Vdc_r PM
B. Mode Based Power Management Algorithm (PM) PI

The power management algorithm (PM) drives the hybrid Vdc

AC/DC microgrid seamlessly under different modes of Fig. 5. PV power de-rating loop based on power management
operation. It considers following main criterion to operate the
system normally. (i) difference of PV power and combine - is disabled and current duty ratio is held constant. Now, dc
ac/dc load power (PD) (ii) operation of battery under maximum link voltage control loop generates the corresponding duty
(SoCBH) and minimum (SoCBL) SoC limits (iii) operation of ratio (D3) which is subtracted from the current duty ratio, as
SC under high (VSCH) and low (VSCL) voltage limit (iv) shown in Fig. 5.
operation of fuel cell under maximum/rated current and _` = _l\R, a_` = 0, G _` = 0 (20)
minimum/zero current limit (v) operation of PV in MPPT Condition E: SoCB < SoCBL and VSC < VSCL
mode or Off MPPT/de-rate. The excess power is used to charge the battery and
dh = dij − dZ (16) supercapacitor to reference values based on (14) and (15). In
Based on the value of difference in PV power (PPV) and load case, the excess power is insufficient to charge the battery and
power (PL), three modes of operation are as follows: SC to reference value, N decreases. Consequently, FC
(i) Mode I: Generation dominant mode (PD>0) supplies the required power.
(ii) Mode II: Load dominant mode (PD<0) _` = _`1, a_` = a_`1 + a_`2,
(iii) Mode III: Normal mode (PD≅0) G _` = G _`1 + G _`2 (21)
Mode I: Following conditions are formulated based on the
SoC status of battery (SoCB) and SC terminal voltage (VSC). Mode II: In this case, PV always operates at MPPT. Similar
Condition A: SoCBL< SoCB < SoCBH and VSCL < VSC < VSCH to MODE I, important conditions are formulated and are given
Since, PV power is more than the load requirement, the fuel as follows.
cell will supply minimum/zero reference current to meet load Condition F: SoCBL< SoCB < SoCBH and VSCL < VSC < VSCH
demand while battery and supercapacitor will charge Since, load power is greater than the PV power so secondary
themselves as per the current references generated. source FC supplies required steady state power assisted by
_` = _l\R, a_` = a_`1, G _` = G _`1 (17) battery while SC provides transient/oscillatory power in
Condition B: SoCB > SoCBH and VSCL < VSC < VSCH accordance to the reference current generation.
When battery hits high SoC limit and is unable to charge itself _` = _`1, a_` = a_`1, G _` = G _`1 (22)
then its reference current is made zero. Now, SC control the dc Condition G: SoCB < SoCBL and VSCL < VSC < VSCH
link voltage and takes the total reference current \_ to In this case, battery is unable to assist fuel cell hence its
charge itself. Since, SC is not suitable for continuous reference current is zero while, the SC supplies its power
charge/discharge hence PV power is curtailed therefore PV de- demand as well as that of battery.
rating is carried out to make the reference SC current ( G _`) _` = _`1, a_` = 0
zero. Consequently, MPPT mode is disabled and current duty G _` = G _`1 + a_`1 (23)
ratio of PV boost converter (D1) is held constant. Now, the de- Condition H: SoCBL< SoCB < SoCBH and VSC < VSCL
rating controller gives reference duty ratio (D2) which is In this condition, SC is unable to supply transient power so
subtracted from the current duty ratio, as shown in Fig. 5. battery supplies its reference power along with that of SC
_` = _l\R, a_` = 0, G _` = _`c (18) while steady state power is met by FC.
Condition C: SoCBL< SoCB < SoCBH and VSC >VSCH _` = _`1, a_` = a_`1 + G _`1,
When SC voltages becomes greater that its upper threshold G _` =0 (24)
VSCH, then it is unable to absorb power, hence its current Condition I: SoCB < SoCBL and VSC < VSCL
reference is reduced to zero. Now, battery controls the dc link In this condition, FC will supply the average power required to
voltage and total reference current is supplied by it. meet load demand. However, without any of the storage
_` = _l\R, a_` = _`c , G _` = 0 (19) element, it is unable to supply the power instantaneously hence
Condition D: SoCB > SoCBH and VSC >VSCH if dc link voltage decreases and reaches near permissible limit
When battery SoC exceeds its upper threshold as well as SC then load shedding is initiated and microgrid is forced to
voltages exceeds VSCH, then HES is unable to absorb any operate under Mode III.
power. Thus, the dc link voltage will start increasing and make _` = _`1e` 0, a_` = 0, G _` =0 (25)
the system unstable. To counter these effects, PV MPPT mode- Mode III: In this mode, PV and load power are approximately
same so PV operates at MPPT while supercapacitor controls

0093-9994 (c) 2017 IEEE. Personal use is permitted, but republication/redistribution requires IEEE permission. See http://www.ieee.org/publications_standards/publications/rights/index.html for more information.
This article has been accepted for publication in a future issue of this journal, but has not been fully edited. Content may change prior to final publication. Citation information: DOI 10.1109/TIA.2017.2756032, IEEE
Transactions on Industry Applications

the dc link voltage to supply/absorb the small oscillatory DC Bus

power. PL1
_` = _l\R, a_` = 0, G _` = _`c
Ppv1
(26)
Cable LC
Load
IV. HYBRID AC/DC MICROGRID CONFIGURATION 2 filter
Boost
The hybrid AC/DC microgrid configuration 2 (MG2) PV Array converter Pfc
consists of multiple PV sources with their individual boost VSC Filter 3 Phase
Load
converters, multiple battery bank with their individual DC
bidirectional converters, a supercapacitor pack with its PL2
PEM Load
Boost converter
bidirectional converter and a PEM fuel cell with its boost Fuel Cell
Pbat1 Psc
converters, connected to same dc link through cable having
impedance mainly resistive at steady state. The dc loads and
three phase inverter fed ac load are also connected at dc link, Bidirectional DC-DC Bidirectional DC Supercapacitor
Battery
as presented in Fig. 6. Since, the sources and storages are Bank 1
Converter Converter Pack
Pbat2 Ppv2
geographically distributed hence, needs a modified power
management scheme for reliable operation.
Battery Bidirectional DC-DC Boost
V. PROPOSED MULTI-TIME SCALE DROOP BASED Bank 2 Converter converter PV Array
POWER MANAGEMENT SCHEME Fig. 6. Hybrid AC/DC Microgrid configuration 2
The power management scheme for MG 2 is modified and
extension of that proposed in Section III. In this scheme, the PEM FC Rc
stack Lfc
current reference generation and its tracking are executed Ifc
locally in a distributed way based on individual system Iofc
available measurements while mode transition signal is Vofc
Vfc dfc
obtained from supervisory control using low bandwidth
communication channel (LBC)[19], having access to system
critical parameters responsible for mode change. To operate
the microgrid (MG2) having sources with different dynamic Ifc Ifcref1
Ifc_max

response such as PEM FC, batteries and supercapacitor, a MAF Kfc Dfc
Vdcr
muti-time scale adaptive droop constituting MAF is dfc
Ifc_min
Current slope
implemented for proper power sharing and generation of PIIfc limiter

current references while inner current controller tracks these

references to drives the dc-dc converters by generating desired 0
Ifcref2
duty ratio.
LBC Supervisory Control
Mode Transition
A. Reference Current Generation and Tracking
Fig. 7. PEM FC time scale droop based control.
The parallel operation and proper current sharing of multiple
DGs with converters in a distributed way, the output voltage
The reference current for PEM fuel cell must be slow time
reference of the converter operates in voltage droop mode.
scaled while satisfying criterion (6). The FC boost output
Also, multiplication of measured voltage deviation with voltage deviation is multiplied by droop constant Dfc to get
reciprocal of virtual resistance is done for proper current
desired output current sharing which is scaled by factor Kfc to
sharing [15]. obtain the corresponding input current. Now, the obtained
∗
= N − 'N ∗ \ (27) input current reference is slow time scaled by passing it
∗
Where is output voltage reference, N is dc link nominal through MAF and current slope limiter to get the desired FC
voltage, 'N is the virtual resistance or droop and \ is the
boost input current reference Ifcref1. The PI based input current
output current of ith converter. The droop constant is defined
controller tracks the desired current reference and generates
as the reciprocal of virtual resistance and is given by
o q?< the FC boost converter duty ratio dfc, as shown in Fig. 7.
= = (28)
*p< r(osr)j?< The reference dc link voltage of the DGs also considers the
Where and are rated output current and voltage of ith DG voltage drop due to cable resistance and is given by [18]
while t is output voltage tolerance in percentage. N = N ( +' ∗ (29)
Where, N ( is the nominal dc link voltage, ' is cable
resistance and is the output current of ith DG. And, the scale
factor Ki is given by
= ⁄ ( (30)

0093-9994 (c) 2017 IEEE. Personal use is permitted, but republication/redistribution requires IEEE permission. See http://www.ieee.org/publications_standards/publications/rights/index.html for more information.
This article has been accepted for publication in a future issue of this journal, but has not been fully edited. Content may change prior to final publication. Citation information: DOI 10.1109/TIA.2017.2756032, IEEE
Transactions on Industry Applications

Where, Voi and Vini are the output and input terminal voltage Rsc Rc
of the ith DG. Lsc
Isc
The supercapacitor converter output voltage Isco
deviation is multiplied by droop constant DSC and scaled by Csc
Vosc
factor KSC. Now, the obtained current reference ISC1 is passed
through MAF and its error with ISC1 serves as the SC converter
input current reference ISCref1. The ISCref1 is fast time scaled and Isc MAF
shares the transient current using droop control. The obtained
ddroop
input current reference is tracked by PI controller to generate PIIsc Ksc* Dsc
SC converter duty ratio ddroop, as shown in Fig. 8. The SC Isc1 Vdcr
Iscref1
voltage control loop similar to (15) generates ISCref2 and inner dsc
Vscr
PWM
dvsc PIIsc PIvsc
PI current controller generates duty ratio dvsc for SC converter. Iscref2
The battery current reference must be slow time scaled Isc LBC Vsc
as well as the current sharing must depend on its present SoC Mode Transition Supervisory Control
status, hence time scaled adaptive SoC based droop control is Fig. 8. SC time scale droop based control.
proposed. The adaptive SoC droop constant exhibits (i) the
Mode
battery with higher SoC must be discharged at higher rate Transition Ib
while battery with low SoC must be discharged at slow rate. Ibref4
(ii) The charge/discharge current shared must be proportional 0
Ibref2 ddroop
to their capacity (Ah). (iii) The battery with lower SoC must SoCbi
PI Ib
be charged at higher rate [18]-[19]. The proposed adaptive SoC MAF
Dbi
based droop constant senses the output voltage deviation of the Ibref1
Vdcr ebi
converter ebi, and changes the droop constant value dBat

accordingly for the charge and discharge mode. PI soc

PIIb
Fu<
Vob
wcxSJy ∗ (ze%YZ − ze% )L, cf < 0
SoCbr
U =
dSoC
Ibref3
f , Fuv
Fu<
= wcxSJy ∗ (ze% − ze%Y| )L, cf > 0
SoC bi Ib
f ,N X U Fuv
Supervisory Control LBC
Mode Transition
(31)
Where, %f is the capacity of ith battery, %f is the capacity of Fig. 9. BES time scaled adaptive SoC based droop control.
battery with highest Ahr. w and y are the tuning parameter.
ze% is the SoC of the ith battery energy storage (BES). The L1 Rc
obtained output current by multiplying droop constant with ebi Ipv
is now scaled by factor Kbi to get input current reference Ibref2. Cpv Iopv
This current reference is slow time scaled by passing through Vpv dpv Vopv
MAF to get the desired BES converter input current reference
Ibref1. The inner PI based current controller tracks the desired
PV array Ipv Vopv-Rdp*Iopv
current reference and generates the converter duty ratio ddroop.
The SoC control loop for BES similar to (14) generates current ddroop
reference Ibref3 and the inner controller tracks the reference
PIIpv PI pv Vdcr
Ipvref1
current and generates duty ratio dSoC, as shown in Fig. 9. dpv
When multiple photovoltaic generation system MPPT
Vpv
dmppt PIIpv PI vpv Ipv
geographically distributed are connected to dc link through Vpvref
cable, then they may operate either in MPPT mode or droop Ipvref2
Ipv Vpv
mode/OFF MPPT. The MPPT algorithm generates the Mode Transition Supervisory Control
reference PV voltage Vpv. The outer PI based voltage controller LBC
Fig. 10. PV MPPT/droop control.
tracks the reference voltage and generates the inner current
reference for boost converter Ipvref2. The PI based inner current - Ipvref1 and generates duty ratio ddroop, as shown in Fig. 10.
controller tracks Ipvref2 and produces duty ratio dmmpt for boost
converter [18]. For operation of PV at OFF MPPT, droop B. Supervisory Control based Mode transition Algorithm
based control is adopted [19]. The output terminal dc voltage The proposed multi-time scale adaptive droop based PMS
obtained using droop (27) and the error is passed to PI based consists of parallel operation of DGs in a distributed way while
voltage control loop which generates the boost converter input supervisory control generates mode transition signal to drive
current reference Ipvref1. The PI based current controller tracks- system from load dominating to generation dominating mode
using LBC. The supervisory control has access to critical

0093-9994 (c) 2017 IEEE. Personal use is permitted, but republication/redistribution requires IEEE permission. See http://www.ieee.org/publications_standards/publications/rights/index.html for more information.
This article has been accepted for publication in a future issue of this journal, but has not been fully edited. Content may change prior to final publication. Citation information: DOI 10.1109/TIA.2017.2756032, IEEE
Transactions on Industry Applications

parameters of system such as ze% , ze%YZ , ze%Y| , gF , gFZ , VI. RESULTS AND DISCUSSION
gF| , NF , N and ∆ NF = N − N .
Mode I: Generation dominating mode: ∆ NF > 0.
A. Simulation Results 1
Condition A: SoCBLi< SoCBi < SoCBHi and VSCL < VSC < VSCH. The hybrid AC/DC microgrid configuration 1 is simulated
All the PVs operate at MPPT and supply power based on dmmpt. under various modes and conditions discussed above to
The SC absorbs transient power based on ISCref1 and operates validate the proposed dynamic power management scheme 1
under dynamic load and source power variation. All the
in droop mode ddroop. The BESs shares the power based on
components are modelled in Matlab/Simulink in
adaptive droop f , U , with current Ibref1 and duty ratio
= 0.
simpowersystem domain and parameters used for simulation
ddroop. The PEM FC does not supplies any power,
are presented in Appendix.
Condition B: SoCBLi< SoCBi < SoCBHi and VSC > VSCH.
• = 0,
1). Scenario I:
The SC is unable to absorb transient power hence g During t=0 to t=4 s, the system is working in Mode II,
while BES shares both average as well as transient power Condition F as the PV power is less than load power, as shown
therefore, mode transient signal switches to f . The PV in Fig. 12. The deficit power is supplied by fuel cell, battery
operates at MPPT while FC supplies zero current ( ). and SC, according to (22). At t= 2 s, PV power changes from
Condition C: SoCBi > SoCBHi and VSC < VSCH. 6.4 kW to 7.4 kW while at t= 4 s, load power decreases from
When any one of battery hits SoCBH and is unable to absorb 7.6 kW to 5.5 kW. Now, the hybrid microgrid operates in
any power, its reference current is reduced to zero, f • = 0, Mode I Condition A. Consequently, FC power start decreasing
while other BES will supply f o . The SC supplies X o,
while SC and battery absorb power to maintain dc link voltage,
FC supplies while PV remains at MPPT mode. as illustrated in Fig. 13 and Fig. 15. The battery absorbs the
Condition D: SoCBi > SoCBHi and VSC < VSCH. average power while SC compensates the oscillatory and
When both BES hits SoCBH, then both the current reference are transient power as per (17), as shown in Fig. 15. At t=7 s,
• = 0. The SC supplies X
battery SoC reaches its maximum value SoCBH (80%), hence
f o while FC supplies .
system experiences critical Condition B and is presented in
Now, mode transition signal switches PV from MPPT mode to
droop mode with CT o and boost duty ratio HN C.
Fig.16. As a result, the power management scheme generates
current references based on (18) and executes PV de-rating
Condition E: SoCBi > SoCBHi and VSC > VSCH.
loop to match load and PV power, as shown in Fig. 12. At t=8
In this case BES as well as SC are unable to absorb any excess
s, the load power decreases to 4.6 kW. Again, PV de-rating
power, hence both of them supplies zero reference current and
loop decreases the PV boost converter duty ratio to force the
PV operates in droop mode to de-rate its power in accordance
SC current zero, as shown in Fig. 12 and Fig. 14. At this
to load power and maintain dc link voltage under permissible
instant, SC supplies/absorbs a very small oscillatory power. At
tolerance. Therefore, f • = g • = = 0. t =11.45 s, SC voltage reaches its maximum value VSCH, hence
Mode II: Load dominating mode: ∆ NF < 0. system enters into Mode I Condition D, as shown in Fig 16.
In this case, All the PVs always operates in MPPT mode. Now, the dc link voltage is controlled to de-rate the PV power
Condition F: SoCBLi< SoCBi < SoCBHi and VSCL < VSC < VSCH such that it power matches the load power, as presented in Fig.
In this case, BES shares the average power and discharges 12. At this instant, SC, battery and FC supply zero powers
based on adaptive droop constant f ,N X U , with f o based on (20). Three phase line to line voltage of VSC output
current reference and duty ratio HN C. The SC supplies (380 V rms) and load current is shown in Fig. 17.
transient current X o while PEM FC shares power in droop 2). Scenario II:
mode with current reference o. In this scenario, load dominating mode conditions are
Condition G: SoCBi < SoCBLi and VSCL < VSC < VSCH simulated. At t=2 s, PV power reduces from 5.5 kW to 4.5 kW
In this case, BES which hits its low limit operates in SoC and at t=4 s, again reduces to 3.4 kW, as shown in Fig. 18 and
control mode with current f € and duty ratio dSoC, while Fig. 19. During t=0 to t= 6 s, load power of 6.4 kW is more
other BES discharge in adaptive droop mode. The SC supplies than PV power hence the system operates in Condition F. The
average deficit power is supplied by the FC assisted by battery
X o in droop mode while FC supplies o in droop
and transient/oscillatory power being supplied by SC as per
mode.
(22) and is shown in Fig. 19. At t=6.1 s, battery reaches its
Condition H: SoCBLi< SoCBi < SoCBHi and VSC < VSCL
lower SoC limit SoCBL (20%), as presented in Fig. 21. Now,
In this case, SC is unable to supply transient power and mode
the power management algorithm produces current references
transition signal switches to voltage control mode with X
according to (23) as system experiences Condition G. At
and dVsc. The FC supplies o in droop mode. t=9.94 s, the SC voltage reaches its lower voltage limit VSCL,
Condition I: SoCBi < SoCBLi and VSC < VSCL therefore system experiences Condition I. Thus, PM produces
The BES charges though SoC control loop while SC operates zero current references for battery and SC while average
in voltage control mode, hence f € and X while FC current reference is supplied by FC as per (25) and the
supplies o . This condition is common for both the modes. converter current controller track these references effectively,-

0093-9994 (c) 2017 IEEE. Personal use is permitted, but republication/redistribution requires IEEE permission. See http://www.ieee.org/publications_standards/publications/rights/index.html for more information.
This article has been accepted for publication in a future issue of this journal, but has not been fully edited. Content may change prior to final publication. Citation information: DOI 10.1109/TIA.2017.2756032, IEEE
Transactions on Industry Applications

Fig. 17. Load voltage (L-L) and load current scenario I

Fig. 11. Variation of dc link voltage scenario I

Fig. 18. DC link voltage variation scenario II

Fig. 12. Variation of PV power and load power

Fig. 13. Variation of Battery, SC and FC powers scenario I Fig. 19. PV, SC, FC, battery and load power scenario II

Fig. 14. Variation of PV boost converter duty ratio scenario I

Fig. 20. Variations in SC, FC and battery currents scenario II.

Fig. 15. Variation of SC, FC and battery currents scenario I Fig. 21. Variations in % SoC of battery scenario II.

-as presented in Fig. 19 and Fig. 20. The dc link voltage and
input current controllers maintains the dc link almost constant
at 700V reference even when subjected to step change in load
and source powers, as shown in Fig. 11 and Fig. 13.
B. Simulation Results 2
The PMS proposed in Section V is validated for MG 2, -
Fig. 16. Variation of SC voltage and % SoC of Battery scenario I

0093-9994 (c) 2017 IEEE. Personal use is permitted, but republication/redistribution requires IEEE permission. See http://www.ieee.org/publications_standards/publications/rights/index.html for more information.
This article has been accepted for publication in a future issue of this journal, but has not been fully edited. Content may change prior to final publication. Citation information: DOI 10.1109/TIA.2017.2756032, IEEE
Transactions on Industry Applications

Fig. 22. DC link voltage variation

Fig. 27. Experimental setup

- from 1.5 kW to 2.9 kW at t=3 s, to 4.35 kW at 5 s and 6.5 kW

Fig. 23. PV1, PV2 and load powers.
at 7.5 s and hits it’s SoCBH at 8.75 s, shown in Fig. 25 and Fig.
26. Hence, the MG2 enters in Condition D, so the PVs changes
its operation from MPPT mode to droop mode and shares same
power of 2.27 kW, presented in Fig. 23. The SC supplies
transient power while BES shares average power based on
multi-time scale droop. The PEM FC supplies zero power. The
dc link voltage rises from 700 V to 703.5 V till 7.5 s while
Fig. 24. Variations in SC power. decreases to 699.3 V at 8.75 s, as shown in Fig. 22. Thus, the
presented PMS for MG2 is robust to mode transitions without
any oscillations due to use of MAF based multi-time scale
adaptive droop constant as compared to LPF based droop
constant [15]-[16].
C. Experimental Results
The real time operation of the proposed MG1 with HES is
implemented in FPGA platform which consist of Xilinx board
programmed through National Instrument-PXIe chassis and
Fig. 25. Variations in BES powers. analog signal acquisition by NI-PXIe 7853 R series channels.
The proposed PMS is implemented in GUI based LABVIEW
which provides switching pulses to all the converters and VSC.
The hardware setup is presented in Fig. 27, consists of rooftop
PV panels, 48 V lithium- ion battery, 32 V supercapacitor
pack, H-1000 Horizon PEM FC stack with its processor and
dry H2 cylinder as well as boost and bidirectional converters.
It also consists of 1 KVA Semikron inverter with LC filter. Ac
Fig. 26. Variations in % SoC BES bulb and dc Fan and light loads with controllable switch are
used. The parameters for the experimental setup are given in
- presented in Fig.6. Initially MG2 is operating in generation Appendix. Case (1) corresponds to Condition F of Mode II.
dominating MODE I, Condition A. The load power changes Here, PV power is assumed zero (night) and the total ac and dc
from 8.73 kW to 6.57 kW at t=3 s and again decreases to 4.35 load is met by FC and HES. At instant t1, loads are switched
kW at t=5 s. The power supplied by PV1 and PV2 are 6.62 kW ON. Initially, FC current is 0.2 A. At t1, the battery supplies
and 4.52 kW respectively, presented in Fig. 23. The BES1 of 0.25 A while SC supplies 0.4 A instantaneously. The FC
20 Ahr with 79.8% SoC and BES 2 of 10 Ahr with 79.85% current ramp up to 0.6 A in 4 s while, battery current goes to
SoC, absorbs power based on adaptive droop (31). The power zero in same time duration. The SC current reduces to 0.2 A to
absorbed by BES 2 changes from 0.74 kW to 1.45 kW at t=3 supply oscillatory power component in the MG, as shown in
s, and to 2.17 kW at t=5s, and hits its SoCBH at 7.5 s, thus MG2 Fig. 28. Note, that the negative current direction of battery and
enters in Condition C. The BES 1 absorbed power changes - SC shows discharging. Again at t2, the loads are switched ON.

0093-9994 (c) 2017 IEEE. Personal use is permitted, but republication/redistribution requires IEEE permission. See http://www.ieee.org/publications_standards/publications/rights/index.html for more information.
This article has been accepted for publication in a future issue of this journal, but has not been fully edited. Content may change prior to final publication. Citation information: DOI 10.1109/TIA.2017.2756032, IEEE
Transactions on Industry Applications

1 A/Div 20 V/Div
IFC 80 V VDC ISC
1 A/Div
Derating
IPV
Load 1A/Div
2 A/Div Decrease IFC
change 1 A/Div
ISC
IPV 2 A/Div
decreases
2 A/Div ISC
IPV
2 A/Div
2 A/Div
IB 1 A/Div IB
IB
t1 t2 t1 t2 t1 t2

Case (1) Case (2) Case (3)

Fig. 28. Experimental results under various conditions.

The FC current ramps up from 0.6A and reaches 1 A in next with supervisory control for MG2 offers reliable transition
4s. The SC supplies transient 0.3A and then reduces to 0.2A. algorithm for operation of multiple PVs and BES in a
During this period, battery current increases from zero to 0.2A geographically distributed location. It also considers the SoC
and then reduces to zero as FC reaches steady state. Thus, the charging and discharging rates for multiple BESs. Also, the
PMS generates current references as per (22) and PI based PMS considers effective utilization of H2 in FC stack by using
current controllers tracks them accurately. Case (2) refers to current slope limiter. The paper also proposes the control based
critical Condition E in mode I, which is consequent of PV power curtailment under critical conditions. The proposed
Condition B, hence battery and FC current are zero initially. PMS considers all the contingency conditions. The simulation
Since, PV power is more than load power and battery has and experimental results validates the proposed PMS under
already reached SoCBH hence, SC is charging to maintain the normal as well as critical conditions. Thus, PV-PEM fuel cell
power balance. At instant t1, SC voltage reaches maximum with HES and proposed PMS presents a promising scope for
voltage limit (VSCH). The SC current at instant t1 is 0.4 A while operation as a hybrid AC/DC microgrid.
PV is supplying 1 A and dc link voltage is at reference of 80V.
Now, the PV goes into de-rating mode with dc link voltage APPENDIX
controller generating the duty ratio (D3) and subsequently Simulation parameters:
operates PV at off MPPT. The PV current start decreasing at PV: Vmp=350 V, Imp=28.5 A, Pmp=10 KW, Lpv=4.5 mH
instant t1, dc link voltage rises slightly above to 82 V and at t2 De-rating loop parameters: Condition B: C = 0.01, = 0.5
Condition D: C = 0.2, = 2.23. Droop: Rd =0.2.
the PV current reduces to 0.6 A, Isc reduces to zero and dc link
= 300, = 0.06
voltage settles to 80 V reference. Thus, power balance is
PEM FC: R=8314.47, Rint=0.00303,
restored. Case (3), system works in Mode II Condition G.
= 96.487 ∗ 10ˆ ,, ‰ = 4.22 ∗ 10sŠ , ‰ = 7.71 ∗ 10sˆ
Initially at steady state, PV current is 0.2 A while FC supplies
2.5 A, battery supplies 0 A and SC supplies 0.2 A. Now, at ‰ = 2.11 ∗ 10sŠ = 6.74, | = 3.37, | = 18.418,
instant t1, load is decreased so FC current decreases slowly to ,F = 348 V, d,F = 6 KW, P,F = 5 l‹ ',F = 0.01 Ω

2 A. At t2, the battery SoC reaches SoCBL. Hence, IB reduces PI current Control: C = 0.0104 = 65.97. Dfc=1/0.2
to zero while dc link voltage is controlled by SC hence SC Battery: Y = 350 , %Yo = 20 Œℎ`, PY = 4.5 l‹, ze%YZ =
current increases to 0.8 A to maintain the power balance, as 20 % , ze%Y| = 80%, Cf = 0.03 f = 62.3, CB2 =10 Ahr.
shown in Fig. 28. SC: X = 375 , X = 160, C = 1, %X = 10 , gFZ =
180 , gF| = 313.5 , PgF = 4.5 l‹, CX = 0.025,
VII. CONCLUSION X = 36.2. DC link: N = 700 , %N = 4700 • , CN =
The proposed PMS 1 for hybrid AC/DC MG1 successfully 5.29, N = 51.72. Dsc=1/0.2. Rc=0.02 ohm.
drives the MG1 from generation dominating mode to load VSC: = 380 , Rating=25 KVA, L=3mH, C=50 • , Outer
dominating mode with efficient dc link voltage regulation. The voltage and current loop PI control: CT = 0.109, T =
presented PMS is robust to wide variation in operating point. 73.29, C = 18.85, = 628.32.
The use of MAF efficiently separates the average current Experimental parameters: PV: X = 4, C = 3 , C =
reference to be supplied by fuel cell and battery while transient 72.28,d C = 180 •, PCT = 3 l‹. Derating loop: Condition
B- C = 0.035, = 0.43 Condition D: C = 0.8, = 5.35.
and oscillatory component of power to be supplied by SC. The
proposed MAF based multi-time scale adaptive droop PMS

0093-9994 (c) 2017 IEEE. Personal use is permitted, but republication/redistribution requires IEEE permission. See http://www.ieee.org/publications_standards/publications/rights/index.html for more information.
This article has been accepted for publication in a future issue of this journal, but has not been fully edited. Content may change prior to final publication. Citation information: DOI 10.1109/TIA.2017.2756032, IEEE
Transactions on Industry Applications

PEM FC: 1 KW H-1000 Horizon FC Stack, ,F = 45 , = [15] N. Yang, F. Gao, D. Paire, A. Miraoui and W. Liu, "Distributed control
48, ,F = 40 Œ (rated), P,F = 3 l‹, C = 0.0424 =
of multi-time scale DC microgrid based on ADRC," IET Power
Electronics, vol. 10, no. 3, pp. 329-337, 3 10 2017.
17.46. Lithium-Ion Battery: Y = 48 , %Y = 20 Œℎ`, PY = [16] Y. Gu, W. Li and X. He, "Frequency-Coordinating Virtual Impedance
3 l‹, Cf = 0.62 f = 23.7 SC: X = 32 , X = for Autonomous Power Management of DC Microgrid," IEEE

2, C = 1,%X = 9 , gFZ = 15 , gF| = 28 , PgF = 3 l‹,

Transactions on Power Electronics, vol. 30, no. 4, pp. 2328-2337, April
2015.
CX = 4.69 , X = 43.9. Dc link: N = 80 , %N = [17] Q. Shafiee, T. Dragičević, J. C. Vasquez and J. M. Guerrero,

2350 • , CN = 12.56 = 39.76

"Hierarchical Control for Multiple DC-Microgrids Clusters," IEEE
N VSC: 1 KVA, L=3mH, Transactions on Energy Conversion, vol. 29, no. 4, pp. 922-933, Dec.
C=10• , Vo=50V (L-L) CT = 0.087, T = 14.85, C = 2014.
5.78, = 35.78.
[18] Q. Yang; L. Jiang; H. Zhao; H. Zeng, "Autonomous Voltage Regulation
and Current Sharing in Islanded Multi-inverter DC Microgrid," IEEE
Transactions on Smart Grid , vol.PP, no.99, pp.1-1, 2017.
REFERENCES [19] T. Dragičević, J. M. Guerrero, J. C. Vasquez and D. Škrlec, "Supervisory
Control of an Adaptive-Droop Regulated DC Microgrid With Battery
[1] T. Ma, M. H. Cintuglu and O. A. Mohammed, "Control of a Hybrid Management Capability," IEEE Transactions on Power Electronics, vol.
AC/DC Microgrid Involving Energy Storage and Pulsed Loads," IEEE 29, no. 2, pp. 695-706, Feb. 2014.
Transactions on Industry Applications, vol. 53, no. 1, pp. 567-575, Jan.- [20] Bhattarai, B., Myers, K., Bak-Jensen B., Paudyal S., “Multi-time scale
Feb. 2017. control of demand flexibility in smart distribution
[2] Xiong Liu, Peng Wang and P. C. Lon, "A hybrid AC/DC microgrid and networks”, Energies, 10, 37, Jan. 2017.
its coordination control," IEEE Trans. Smart Grid, vol. 2, pp. 567-575, [21] Dinghuan Zhu, “Multi-time scale control of energy storage enabling the
June. 2011. integration of variable generation,” Ph.D. dissertation, Dept. Of Elect.
[3] O.C. Onar, M. Uzunoglu and M. S. Alam, "Dynamic modelling, design And Comp. Eng., Carnegie Mellon Univ., PA, 2014.
and simulation of a wind/fuel cell/ultacapacitor based hybrid power [22] N. Yang, D. Paire, F. Gao and A. Miraoui, "Distributed control of DC
generation system," Journal of Power Sources, vol. 161, pp. 707-722, microgrid considering dynamic responses of multiple generation
Oct. 2006. units," in Proc. IECON 2014 40th Annual Conference of the IEEE
[4] M. Y. El-Sharkh et al, "A dynamic model for standalone fuel cell power Industrial Electronics Society, Dallas, TX, 2014, pp. 2063-2068.
plant for residential applications," Journal of Power Sources, vol. 138,
pp. 199-204, Nov. 2004.
[5] S. Malo and R. Grino, "Design, Construction, and Control of a Stand- Rishi Kant Sharma (S’16) received the
Alone Energy-Conditioning System for PEM-Type Fuel Cells," IEEE B.Tech in Electrical Engineering from UPTU,
Transactions on Power Electronics, vol. 25, no. 10, pp. 2496-2506, Oct. Lucknow, India in 2008 and M.Tech in Control
2010. and Instrumentation from Motilal Nehru
[6] P. Thounthong, A. Luksanasakul, P. Koseeyaporn and B. Davat, National Institute of Technology, Allahabad,
"Intelligent Model-Based Control of a Standalone Photovoltaic/Fuel Cell
India in 2010. He has work experience of both
Power Plant With Supercapacitor Energy Storage," IEEE Transactions
academic as well as Industry. He has served as
on Sustainable Energy, vol. 4, no. 1, pp. 240-249, Jan. 2013
[7] S. Sikkabut et al., "Control of High-Energy High-Power Densities an Assistant Engineer (O&M) in MPPKVVCL,
Storage Devices by Li-ion Battery and Supercapacitor for Fuel Jabalpur, India. Presently, he is working
Cell/Photovoltaic Hybrid Power Plant for Autonomous System towards the Ph.D. degree in Electrical Engineering at Indian Institute of
Applications," IEEE Transactions on Industry Applications, vol. 52, no. Technology Delhi, India.
5, pp. 4395-4407, Sept.-Oct. 2016. His research interest lies in the field of advanced controller
[8] P. Thounthong, S. RaËl and B. Davat, "Control Algorithm of Fuel Cell design, power management and stability analysis of distributed generation in
and Batteries for Distributed Generation System," IEEE Transactions on AC/DC microgrid as well as application of hybrid energy storage.
Energy Conversion, vol. 23, no. 1, pp. 148-155, March 2008.
[9] P. Thounthong, S. Rael and B. Davat, "Control Strategy of Fuel Cell and Sukumar Mishra (M’97-SM’04) is a Professor
Supercapacitors Association for a Distributed Generation System," IEEE in the Department of Electrical Engineering at
Transactions on Industrial Electronics, vol. 54, no. 6, pp. 3225-3233, the Indian Institute of Technology Delhi. His
Dec. 2007. interest lies in the field of Power Systems,
[10] M. Hamzeh, A. Ghazanfari, H. Mokhtari and H. Karimi, "Integrating Power Quality Studies, and Renewable Energy.
Hybrid Power Source Into an Islanded MV Microgrid Using CHB He has published over 100 research articles
Multilevel Inverter Under Unbalanced and Nonlinear Load Conditions," (including papers in international journals,
IEEE Transactions on Energy Conversion, vol. 28, no. 3, pp. 643-651, conferences and book chapters). He is currently
Sept. 2013. holding the position of Vice Chair of Intelligent
[11] N. R. Tummuru, M. K. Mishra and S. Srinivas, "Dynamic Energy System Subcommittee of Power and Energy
Management of Renewable Grid Integrated Hybrid Energy Storage
society of IEEE. He is a recipient of the INSA medal for young scientist, the
System," IEEE Transactions on Industrial Electronics, vol. 62, no. 12,
INAE young engineer award, and the INAE silver jubilee young engineer
pp. 7728-7737, Dec. 2015.
[12] N. R. Tummuru, M. K. Mishra and S. Srinivas, "Dynamic Energy award. He is also a Fellow of IET (UK), NASI (India), INAE (India), IE
Management of Hybrid Energy Storage System With High-Gain PV (India) and IETE (India).
Converter," IEEE Transactions on Energy Conversion, vol. 30, no. 1, pp. He is working as the NTPC Chair professor and has
150-160, March 2015. previously worked as the Power Grid Chair professor. He is also serving as an
[13] S. Mishra and R. K. Sharma, "Dynamic power management of PV based Independent Director of the Cross Border Power Transmission Company Ltd.
islanded microgrid using hybrid energy storage," in Proc. 2016 IEEE 6th and Industry Academic Distinguish Professor. He has handled many research
International Conference on Power Systems (ICPS), New Delhi, 2016, projects and industrial consultancies. He has a vast experience of 25 years and
pp. 1-6. is currently serving as an Editor in IEEE TRANSACTIONS ON SMART
[14] A. Tani, M. B. Camara and B. Dakyo, "Energy Management in the GRID and Associate Editor in IET Generation, Transmission & Distribution.
Decentralized Generation Systems Based on Renewable Energy—
Ultracapacitors and Battery to Compensate the Wind/Load Power
Fluctuations," IEEE Transactions on Industry Applications, vol. 51, no.
2, pp. 1817-1827, March-April 2015.

0093-9994 (c) 2017 IEEE. Personal use is permitted, but republication/redistribution requires IEEE permission. See http://www.ieee.org/publications_standards/publications/rights/index.html for more information.