Energy Management System with Equalization

Algorithm for Distributed Energy Storage Systems in

PV-Active Generator Based Low Voltage DC
Microgrids

Nelson L. Daz , Adriana C. Luna , Juan C. Vasquez , and Josep M. Guerrero

Department of Energy Technology, Aalborg University, Aalborg, Denmark
Facultyof Engineering, Universidad Distrital F. J. C., Bogota, Colombia
nda@et.aau.dk, acl@et.aau.dk, juq@et.aau.dk, joz@et.aau.dk
www.microgrids.et.aau.dk
(PV1+ESS1) (PV2+ESS2)
AbstractThis paper presents a centralized strategy for equal- PV1 ESS1 ESS2 PV2
izing the state of charge of distributed energy storage systems in
an islanded DC microgrid. The proposed strategy is based on a
simple algorithm called equalization algorithm, which modies
the charge or discharge rate by weighting the virtual resistor
DC DC DC
of local droop control loops at each distributed energy storage PV1 Primary
Control DC DC
ESS1 Primary
Control
ESS2 Primary DC
DC DC
PV2 Primary
Control
Control
system. The proposed strategy, can be used as an additional
function of the microgrid energy management system where
the state of charge of distributed ESS is equalized within a
determined window of time. Finally, real-time simulation results DC common bus
of a low voltage DC microgrid are presented in order to verify
the performance of the proposed approach. Critical Load

KeywordsDistributed energy storage systems, Droop control,

Equalization algorithm, State of Charge.

I. I NTRODUCTION Fig. 1: Islanded DC microgrid conguration based on

Low voltage DC power distribution systems have been
widely used for supplying critical loads, such as data cen-
ters, remote communication stations or residential applications
[1], [2]. In particular, DC power distribution systems offer
more reliability, and efciency than conventional AC power
distribution systems [3]. In addition, aspects associated with
synchronization, reactive power ow, and harmonic currents
are not a concern in DC power systems [4].
With the fast development of Renewable energy sources
(RES), a microgrid appear as a feasible solution for a co-
ordinated integration of RES into a low voltage DC power
system [5]. However, the stochastic behavior of the RES, such
as photovoltaic (PV) generators, requires the integration of
more energy storage systems (ESS) in order to smooth the
variations at the RES [6]. As a matter of fact, when economic
and environmental issues do not allow interconnection with
the main power grid, the capacity of the ESS needs to be
increased in order to ensure several to many hours of energy
reserve [7]. For that reason, the current trend is oriented
to the integration of distributed RES and its corresponding
ESS as a unit denoted as active generator (PV+ESS) [6], [8].
Fig. 1 shows the basic scheme of an islanded DC microgrid
composed by two (PV+ESS) active generators and a critical
load. Commonly, valve regulated lead-acid (VRLA) batteries
are the most used in islanded microgrids, since they offer
(PV+ESS) active generators.
a good commitment between energy density, deep-cycle life,
transportability, availability, and cost [7], [9].
When distributed ESS are used, it is recommended a
coordinated operation between them in order to avoid deep-
discharge in one of the energy storage unit and over-charge in
the others. Differences at the SoC could limit the life-time of
the ESS with the smallest SoC, since this ESS will be exposed
to bigger deep of discharge [10]. Therefore, when the ESS are
being charged, it is desirable to prioritize the charge of the ESS
with the smallest state of charge (SoC), and on the contrary,
when the ESS are being discharged, the unit with the highest
SoC should provide more power to the microgrid than the
others in order to ensure stored energy balance [11], [12].
Commonly, droop control strategies are used in order to
achieve good power sharing between units [5]. Indeed, con-
ventional control loops for distributed ESS are complemented
with control actions which adjust the droop coefcients in
accordance to the SoC. In this way, it is possible to achieve
equalization at the stored energy. It this sense, several different
approaches have been proposed for equalizing the SoC at
distributed ESS such as in [13][20]. However all of them
assume equal capacity for all the distributed ESS. In [4],
the authors consider some differences between energy storage

978-1-4799-9880-7/15/$31.00 2015 IEEE 293

units. Despite this, the equalization strategy is applied to PV Array
ESS based on electric-double-layer capacitors rather than on Buck Converter DC Bus
L
batteries. Although, the stored energy is balanced, long time
and additional control loops are required. VPV I PV I RES
This paper proposes a simple function for the energy man- C 1
C 2

agement system (EMS) of an islanded DC microgrid, based on

PWM
a centralized strategy denoted as equalization algorithm, which
achieves asymptotic approach of the SoC within an established I ref +
window of time for distributed ESS based on batteries. The MPPT PI
-
proposed equalization algorithm weights the droop coefcients
of the droop control loops, within a dened window of time, Primary Control RES
in order to equalize the SoC for distributed ESS. Real-time
simulation results under charge and discharge conditions and Fig. 2: Control diagram for RES.
considering differences at the batteries capacity are presented
Bidirectional Buck
in order to validate the proposed strategy.
DC Bus Converter Battery Array
L
The paper is organized as follows. Section II explains the
operation of the microgrid and how the droop control loops I ESS I Bat
VDC
should be adjusted in order to achieve equalization. Section C 2
C 1

III explains the proposed equalization algorithm, and nally

+ I ref
PWM
Sections IV and V present real-time simulation results and + + V*
PI PI
conclusions respectively. - - -
VDC
Rd
II. C ONFIGURATION AND O PERATION OF THE DC
M ICROGRID Primary Control ESS

The low voltage DC microgrid consider for this study-case Fig. 3: Control diagram for ESS.
is composed by two PV-based active generators (PV+ESS) and
a critical load (see Fig. 1). The microgrid is formed around a IESSi is the output current of each ESS. In this case, a typical
standard 48Vdc common bus, which is a kind of low voltage double-loop VCM controller is implemented for a bidirectional
DC power distribution system that is widely used, since it non-isolated buck converter, as can be seen in Fig.3. When the
allows working on a live conductor with minimum risk for same virtual resistance (Rd) is applied to each ESS control
personal injury and without special safety requirements [21]. loop, the current is equally shared between ESS (see Fig. 4(a)).
Moreover, non-isolated buck DC/DC converters operating in
continuous mode are used for the conversion stage of each Under the discharge of the battery, for balancing the SoC
RES and ESS as proposed in [15]. between ESS, the ESS with the highest SoC should supply
more power to the microgrid than the other. On the contrary,
In isladed operation, it is expected that the RES operate when the batteries are being charged the ESS with the smallest
by using a maximum power point tracking (MPPT) algorithm, SoC should get more energy from the microgrid than the
in order to obtain from the RES the maximum amount of other. This behavior can be achieved by weighting the virtual
available energy. Fort that reason, RES operate as constant resistance (Rd) by a factor i as is shown in Fig. 4(b), where
power sources following the current reference given by the the largest SoC have been assumed for ESS2. Therefore, (1)
MPPT algorithms. This current reference can be obtained from can now be rewritten as follows:
MPPT methods as the one proposed in [22]. However MPPT
VDC = VDC Rd i IESSi (2)
strategies are out of the scope of this paper, interested readers
may also refer to [23]. To get back to the point, the inductor
VDC (V) V (V)
current of the converter is controlled by typical inner current V*
DC
V*DC *
VDC=VDC-RdIESS2=VDC-RdIESS1
* V*
loops. Fig. 2 shows the scheme of the inner current control for DC DC
*
VDC=VDC-RdD2IESS2
each RES. VDC
VDC VDC
Meanwhile, the ESS operate in voltage control mode
*
VDC=VDC-RdD1IESS1
(VCM) being responsible of regulating the bus voltage. At
this mode, the batteries will be charged or discharged in order
to compensate the unbalance between the energy generated IESS1 = IESS2 I (A) IESS1 IESS2 I (A)
by RES and load consumption [20], [24]. Commonly, The
power unbalance is equally shared between ESS by means (a) (b)
of conventional droop control loop [5]. Therefore, the voltage Fig. 4: Droop characteristics: (a) Equal virtual resistance Rd
at the common bus (VDC ) is established by the following (b) Weighted virtual resistance Rd.
equation:
Finally, the microgrid system is complemented with an en-
VDC = VDC Rd IESSi (1)
ergy management system (EMS) that execute the equalization
where Rd is the virtual resistance of the droop control loop, algorithm in order to obtain the values for 1 and 2 as can

VDC is the voltage reference of the common DC bus, and be seen in Fig. 5.

294
 I RES1
(PV1+ESS1)
DC/DC
DC/
C/DC DC/DC
DC/
C/DC
I ESS1
PV Array
a Converter
C
Co nverte
t r Converter
C
Co nverte
t r Battery
Batt
ttery
ry Array
ay 1
I RES 1 I ESS 1 I Bat1

VPV
P I PV VDC SoC Bat 1
DC

PWM
W W
PWM *
Currentt
Current I ref + + V DC
C
I ref Current
C rrent
Cu contr
t ol
control PI
P
- -
MPPT
MP
MPPT contr
t ol
control loop
l op
lo o
VDC
D D1
l op
loo
loop
Rd EMS
Primary
P im
Pr i ary
r Co
C
Control
ntr
t ol RES
RES 1
RE Primary
P im
Pr i ary Control
r Co
C t ol Battery
ntr Batt
ttery
r 1

I Load
Load I Load
Equalization
DC Common Bus
Algorithm
PV Array
ay DC/DC
C/DC Converter
DC/ C nverte
Co t r DC/DC
C/DC Converter
DC/ C nverte
Co t r Batt
tterryy Array
Battery ay 2
I RES 2 I ESS 2 I Bat 2
VPV
P IWTG SoC Bat 2
VDC
D

Primary
P im
Pr i ary Control
r Co
C t ol RES
ntr RES 2
RE Primary
P im
Pr i ary Control
r Co
C t ol Battery
ntr Batt
ttery
r 2 D2
I ESS 2
(PV2+ESS2)
I RES 2
Fig. 5: Diagram of the DC microgrid with conventional inner control loops.
88
It is important to say that this paper will only consider the
86
operation of the microgrid when the ESS are being charged I Bat1
84
or discharged. Anyhow, the operation of the microgrid should mSoC1

be complemented by appropriate charge strategies that avoid 82

excessive overcharge in the batteries, such as in [20], and [15], 80

I Bat 2
SoC (%)

as well as load-shedding, or actions for limiting the deep of 78

mSoC2
discharge of the batteries as proposed in [25]. Next Section 76

will explain the proposed equalization algorithm in detail. 74

72

III. P ROPOSED E QUALIZATION A LGORITHM FOR S O C 70

68
The algorithm is based on the fact that the rate of change of 0 1 2 3 4
Time (s)
5 6 7

the SoC is directly proportional to the battery current (Ibati

mSoCi ), where mSoCi is the rate of change for the SoC at Fig. 6: Expected behavior of the equalization algorithm.
each ESS. For that reason, by adjusting mSoCi it is possible
to achieve an equalization of the SoC at distributed ESS, as is
battery in (A/h), Bati is the charging/discharging efciency,
shown in Fig. 6, where the dashed lines represent the behavior
and IBati ( ) is the instantaneous current at each battery array
of the SoC without compensation, and the continuous lines
[9]. In addition, the output current of each ESS (IESSi ) is
represent the expected behavior of the equalization algorithm.
inversely proportional to the battery current (IBati ) by a factor
1/D,
In order to derive the proportional relationship between 1
the battery current and the ration of the SoC, we consider the IESSi = IBati (4)
D
ampere-hour (Ah) counting method equation [15],
 t where D is the duty-cycle of the PWM signal that controls the
IBati ( ) buck converter. By considering a constant current charge from
SoC(t)Bati = SoC(0)Bati Bati d (3)
0 CBati (3), the following relationship can be obtained,
 
where SoC(t)Bati is the nal SoC after a period t, SoCBati (%) 3600CBati (A/h)
IESSi = (5)
SoC(0)Bati is the initial SoC, CBati is the capacity of the t(s) Bati (%)D

295
where SoCBati = SoC(t)Bati SoC(0)Bati . Then (5) Meassure
External Data
takes the form:
IESSi = mSoCi KBati (6) Solve
 K Bat1mSoC1  K Bat 2 mSoC 2  I RES1  I RES 2  I Load 0
being, mSoCi the rate of the SoC, and KBati a proportionality SoC (0) Bat1  mSoC1 't SoC (0) Bat 2  mSoC 2 't
constant that depends on the main parameters of the ESS.
In a general case, where n distributed active generators No If Ibat<0 Yes
(PV+ESS) are integrated into the microgrid, it is easy to derive
the Kirchhoff current law equation for the common node as:
n 
n
If SoC (0) Bat1 ! SoC (0) Bat 2 If SoC (0) Bat1 ! SoC (0) Bat 2
IESSi + IRESi ILoad = 0 (7)
Yes No
i=1 i=1

where, (ILoad ) is the load current, (IRESi ) is the power D1  Cmin Cmax 
supplied for each RES, and (IESSi ) is the current at each ESS. D2 D1  k Bat1mSoC1 k Bat 2 mSoC 2 
Wait until
In particular, for the proposed DC microgrid shown in Fig. Time 't
No Yes
5, by combining (6) and (7), we have
D2  Cmin Cmax 
2
 2
 D1 D 2  k Bat 2 mSoC 2 k Bat1mSoC1 
mSoCi KBati + IRESi ILoad = 0 (8)
END
i=1 i=1
Fig. 7: Equalization algorithm.
Moreover, in order to perform the SoC equalization, it is
required that the straight-line equations, that represents the TABLE I: Main Parameters of the microgrid
behavior of the SoC within a dened period of time (t),
Parameter Symbol Value
are equalized between then as: Nominal Bus Voltage
VDC 48 (V)
10 ()
SoC(0)Bat1 + mSoC1 t = SoC(0)Bat2 + mSoC2 t (9) Nominal Load
Maximum (RES) Power Rating
RLoad
PRESmax 300 W
Nominal Voltage V bat 48V
to be more precise SoC(t)Bat1 = SoC(t)Bat2 . Nominal Battery Capacity Cbat 0.02 (Ah)
Period of the Equalization Algorithm t 5 (s)
To get back to the point, the main task of the equalization Duty-cycle D 0.93 (Ah)
algorithm is to solve the equation system composed by (8) Charging/discharging efciency Bati 100
and (9), in order to obtain the adequate values for mSoCi Virtual Resistance Rd 0.5 ()

that ensure the equalization of the SoC within a dened

period (t). Once the value for each mSoCi is obtained, it
is necessary to obtain the adequate values for the weighting
factor i (see (2)).
Firstly, the algorithm determines if the ESS are being
charged of discharged by evaluating the sign of any ESS
current. This step is necessary since under the process of
charge the smallest value of should be assigned to the ESS
whit the smallest SoC in order to charge the ESS with the
smallest SoC faster. On the other hand, under the process of
discharge the smallest value of should be assigned to the ESS
with the biggest SoC. In this way, the ESS with the biggest
SoC will be discharged faster than the others. Consequently, it
is also necessary to determine which ESS has the biggest SoC,
this is possible by simple comparison. As a result, the values
of are bounded to 1. In addition, by considering differences
at the capacity of distributed ESS the maximum value of the
weighting factor (i ) is determined by:
max = Cmin /Cmax (10)
where Cmax and Cmin are the maximum and minimum values
of the ESS capacities. To illustrate, the algorithm is shown in
Fig. 7.
Likewise, under normal operation it is expected a similar
charge/discharge ratio for both ESS (mSoC1 = mSoC2 ) despite
of differences at the capacity of each ESS. To achieve this
behavior, the virtual resistance Rd of the ESS with the biggest
capacity should be weighted by the relationship Cmin /Cmax .
IV. H ARDWARE - IN - THE - LOOP R ESULTS
The proposed equalization algorithm has been tested in
a low voltage DC microgrid model established in a MAT-
LAB/Simulink model that was downloaded into a dSPACE
1006 platform in order to evaluate the performance of the al-
gorithm in real-time. The microgrid composition was presented
in Fig. 5. The power stage and main parameters are presented
in Table I. Particularly, small values of capacity (Cmax = 0.02)
have been selected in order to speed up the simulation time.
Detailed models of the VRLA batteries have been used for
simulation as is proposed in [15].
For the simulation, three main cases have been considered.
That is, CBat1 = CBat2 , CBat2 > CBat1 , and CBat2 < CBat1 .
All the cases were simulated by considering a total generation
from RES of PRES = 260W and PRES = 100W for charging
and discharging respectively.
A. Case CBat1 = CBat2
Fig. 8 shows the performance of the equalization algorithm
when the ESS are being charged. An initial SoC of 65% and
75% have been established for ESS1 and ESS2 respectively.
Fig. 8(a) shows the equalization process for SoCBat1 and
SoCBat2 . Fig. 8(b) shows the way the output current at
each ESS is equally shared between ESS when the algorithm
is not applied, and how the current is adjusted during the
equalization in order to achieve the objective. At the end,

296
 51.0 0 51.0 0
Equalization
hjhkhjlkhjkhjjjj Equalization 0 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 Equalization
Equalization 17 18 19 20 21 22 23 24
Algorithm Algorithm Algorithm Time (Sec.) Algorithm

{
{
{
100 100
90 SoCBat2 90 SoCBat2
SoC (%)

SoC (%)
#1:2 #1:2
80 80 #1:1
#1:1
70 70
60 0
SoCBat1 SoCBat1
60 0
0 1 2 3 4 5 6 7 8 9 10 11 2 22 23 24
1 12 13 14 15 16 17 18 19 20 21 0 1 2 3 4 5 6 7 8 6 17 18 19 20 21
1 12 13 14 15 16
9 10 11 2 22 23 24
Time (s) Time (s)
(a) (a)

0.5 #1:1 IBat2 0.0

#1:2 #1:1 IBat1
0.0
IBat (A)

-0.2

IBat (A)
-0.5 #1:2
-0.4
-1.0 -0.6
IBat1 IBat2
-1.5 0 -0.8 0
0 1 2 3 4 5 6 7 8 1 17 18 19 20 2
11 12 13 14 15 16
9 10 1 21 22 23 24 0 1 2 3 4 5 6 7 8 9 10 1
11 12 13 14 15 16 21 22 23 24
6 17 18 19 20 2
Time (s) Time (s)
(b) (b)

0
0
Diff(SoC)

Diff(SoC)
-5
-5
#1:1 #1:1
-10 0
-10 0
0 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24
0 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24
Time (s)
Time (s)
(C)
(c)

Fig. 8: ESS Charge when CBat1 = CBat2 : (a) SoC, (b) IBat , Fig. 10: ESS Charge when CBat1 < CBat2 : (a) SoC, (b) IBat ,
(c) Diff(SoC). (c) Diff(SoC).
49.0 0
0 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 Equalization
Equalization 17 18 19 20 21 22 23 24 25
48.0 0
Algorithm Time (Sec.) Algorithm 0 1 2 3 4 5 6 Equalization
7 8 9 10 11 12 13 14 15 16 Equalization
17 18 19 20 21 22 23 24 25
SoCBat2
{

#1:2
{

80 #1:1
Algorithm Time (Sec.) Algorithm
SoCBat2

{
{
#1:2
SoC (%)

60 80 #1:1
SoCBat1
SoC (%)

40 60
SoCBat1
20 0 40
0 1 2 3 4 5 6 7 8 9 10 11 2 22 23 24 25
1 12 13 14 15 16 17 18 19 20 21
20 0
Time (s) 0 1 2 3 4 5 6 7 8 9 10 11
1 12 13 14 15 16 17 18 19 20 21
2 22 23 24 25
(a)
Time (s)
(a)
2.0 IBat2
1.5 #1:2
IBat2
IBat (A)

2.0 #1:2
1.0 1.5
IBat (A)

0.5 1.0
#1:1
IBat1
0.0 0 0.5 #1:1
0 1 2 3 4 5 6 7 8 9 10 11 2 22 23 24 25
1 12 13 14 15 16 17 18 19 20 21 IBat1
0.0 0
Time (s) 0 1 2 3 4 5 6 7 8 11 12 13 14 15 16 17 18 19 20 2
9 10 1 21 22 23 24 25
(b)
Time (s)
(b)
5
Diff(SoC)

0 0
-5 #1:1
Diff(SoC)

-5
-10 #1:1
-15 0 -10
0 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25
-15 0
Time (s)
0 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25
(C)
Time (s)

Fig. 9: ESS Discharge when CBat1 = CBat2 : (a) SoC, (b) (c)

IBat , (c) Diff(SoC). Fig. 11: ESS Discharge when CBat1 < CBat2 : (a) SoC, (b)
IBat , (c) Diff(SoC).
Fig. 8(c) shows the difference between SoC Dif f (SoC) =
SoCBat1 SoCBat2 , where it is possible to see that the see clearly from Fig. 12 (from 10s to 11s), that the ESS is able
difference is practically zero after two iterations. Similarly, to smoothly transfer from charge to discharge modes during
Fig. 9 shows the response of the microgrid when the ESS are the equalization process in order to achieve the goal.
being discharged. Comparing Fig. 8 and Fig. 9, we can see
that for the discharging |IBat2 | > |IBat1 |, and for charging
|IBat1 | > |IBat2 | SoC equalization is achieved.
V. C ONCLUSION
B. Case CBat2 > CBat1
Fig. 10 and 11 show the response of the microgrid when The proposed been effective for SoC equalization in dis-
CBat1 = 0.01(A/h). We can see that when the equalization tributed ESS. Nevertheless, at least two iterations are required
is not applied, we have |IBat2 | > |IBat1 |. The reason of this in order to equalize completely the SoC, This is because the
is that ESS2 requires much more current in order to achieve transient and dynamic responses have not been considered
mSoC1 = mSoC2 . However when the algorithm is applied, the by the algorithm. Despite this, and by assuming a linear
current is adjusted in order to equalizing SoCs. behavior, the algorithm is able to equalize the SoC under
few iterations. This algorithm can be complemented by an
optimization process in order to minimize the period of time
C. Case CBat2 < CBat1
t, and by taking into account the power constrains of the
Fig. 12 and 13 show the response of the microgrid when system. Additionally, the algorithm can be easily adapted for
CBat2 = 0.01(A/h). Compared to the previous case |IBat2 | < AC microgrid and grid connected microgrids whit a larger
|IBat1 |, when the equalization is not applied. It is possible to number of interconnected active generators.

297
 51.0 0
0 1 2 3 4 5 6 Equalization Equalization
7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25
Algorithm Algorithm
active generator for smart grid applications, IEEE Transactions on
Time (Sec.)
Industrial Electronics, vol. 58, pp. 45834592, Oct 2011.

{
{
100
90 SoCBat2 [7] J. de Matos, F. e Silva, and L. Ribeiro, Power control in ac isolated
SoC (%)

#1:2
80 microgrids with renewable energy sources and energy storage systems,
#1:1
70
SoCBat1 IEEE Transactions on Industrial Electronics,, vol. PP, no. 99, pp. 11,
60 0
0 1 2 3 4 5 6 7 8 9 10 11 2 22 23 24 25
1 12 13 14 15 16 17 18 19 20 21 2014.
Time (s)
(a)
[8] F. Marra, G. Yang, C. Traeholt, J. Ostergaard, and E. Larsen, A
decentralized storage strategy for residential feeders with photovoltaics,
0.5
#1:1
IBat2
IEEE Transactions on Smart Grid, vol. 5, pp. 974981, March 2014.
IBat (A)

0.0 #1:2
[9] . I. S. C. Committee, Ieee guide for optimizing the performance and
-0.5
IBat1 life of lead-acid batteries in remote hybrid power systems, 2008.
-1.0 0
0 1 2 3 4 5 6 7 8 9 10 11 2 22 23 24 25
1 12 13 14 15 16 17 18 19 20 21 [10] D. Linden and T. Reddy, Handbook of batteries. McGraw-Hill hand-
Time (s) books, McGraw-Hill, 2002.
(b)
[11] Y.-K. Chen, Y.-C. Wu, C.-C. Song, and Y.-S. Chen, Design and
0
implementation of energy management system with fuzzy control for dc
Diff(SoC)

-5 microgrid systems, Power Electronics, IEEE Transactions on, vol. 28,

#1:1
no. 4, pp. 15631570, 2013.
-10 0
0 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25
[12] J. Guerrero, J. Vasquez, J. Matas, M. Castilla, and L. de Vicuna,
Time (s) Control strategy for exible microgrid based on parallel line-interactive
(c)
ups systems, IEEE Transactions on Industrial Electronics, vol. 56,
Fig. 12: ESS Charge when CBat1 > CBat2 : (a) SoC, (b) IBat , no. 3, pp. 726736, 2009.
(c) Diff(SoC). [13] X. Lu, K. Sun, J. Guerrero, J. Vasquez, L. Huang, and R. Teodorescu,
48 0
Soc-based droop method for distributed energy storage in dc microgrid
0 1 2 3 4 5 6 Equalization
7 8 9 10 11 12 13 14 15 16 Equalization
17 18 19 20 21 22 23 24 25 applications, in Industrial Electronics (ISIE), 2012 IEEE International
Algorithm Time (Sec.) Algorithm Symposium on, pp. 16401645, 2012.
#1:2
{
{

80 #1:1
SoCBat2 [14] Y. Zhang, H. J. Jia, and L. Guo, Energy management strategy of
60
SoC (%)

40 SoCBat1 islanded microgrid based on power ow control, in Innovative Smart

20 Grid Technologies (ISGT), 2012 IEEE PES, pp. 18, 2012.
00
0 1 2 3 4 5 6 7 8 9 10 11 2 22 23 24 25
1 12 13 14 15 16 17 18 19 20 21 [15] T. Dragicevic, J. Guerrero, J. Vasquez, and D. Skrlec, Supervisory
Time (s) control of an adaptive-droop regulated dc microgrid with battery man-
(a)
agement capability, IEEE Transactions on Power Electronics, vol. 29,
2.0 IBat1 no. 2, pp. 695706, 2014.
1.5
IBat (A)

1.0 #1:2
[16] C. Li, T. Dragicevic, N. Diaz, J. Vasquez, and J. Guerrero, Voltage
0.5 #1:1
scheduling droop control for state-of-charge balance of distributed en-
IBat2
0.0 0 ergy storage in dc microgrids, in Energy Conference (ENERGYCON),
0 1 2 3 4 5 6 7 8 1 17 18 19 20 21
1 12 13 14 15 16
9 10 11 2 22 23 24 25
2014 IEEE International, pp. 13101314, May 2014.
Time (s)
(b) [17] C. Li, T. Dragicevic, M. G. Plaza, F. Andrade, J. C. Vasquez, and
0
J. M. Guerrero, Multiagent based distributed control for state-of-charge
balance of distributed energy storage in dc microgrids, in Industrial
Diff(SoC)

-5
#1:1
Electronics Society, IECON 2014 - 40th Annual Conference of the IEEE,
-10
pp. 21802184, Oct 2014.
-15 0
0 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 [18] X. Lu, K. Sun, J. Guerrero, J. Vasquez, and L. Huang, State-of-charge
Time (s) balance using adaptive droop control for distributed energy storage
(c)
systems in dc microgrid applications, IEEE Transactions on Industrial
Fig. 13: ESS Discharge when CBat1 > CBat2 : (a) SoC, (b) Electronics, vol. 61, pp. 28042815, June 2014.
IBat , (c) Diff(SoC). [19] Q. Shaee, T. Dragicevic, J. Vasquez, and J. Guerrero, Hierarchical
control for multiple dc-microgrids clusters, Energy Conversion, IEEE
R EFERENCES Transactions on, vol. 29, pp. 922933, Dec 2014.
[20] N. Diaz, T. Dragicevic, J. Vasquez, and J. Guerrero, Intelligent dis-
[1] W. Li, X. Mou, Y. Zhou, and C. Marnay, On voltage standards for dc tributed generation and storage units for dc microgrids - a new concept
home microgrids energized by distributed sources, in Power Electronics on cooperative control without communications beyond droop control,
and Motion Control Conference (IPEMC), 2012 7th International, IEEE Transactions on Smart Grid, vol. 5, pp. 24762485, Sept 2014.
vol. 3, pp. 22822286, 2012. [21] J. Akerlund, -48 v dc computer equipment topology-an emerging tech-
[2] D. McMenamin, Case studies supporting -48 vdc as the power input nology, in Telecommunications Energy Conference, 1998. INTELEC.
of choice for computer equipment deployed in the telecom network, Twentieth International, pp. 1521, 1998.
in Telecommunications Energy Conference, 1998. INTELEC. Twentieth [22] N. Diaz, A. Luna, and O. Duarte, Improved mppt short-circuit current
International, pp. 261265, 1998. method by a fuzzy short-circuit current estimator, in Energy Conversion
[3] C. Wang and P. Jain, A quantitative comparison and evaluation of 48v Congress and Exposition (ECCE), 2011 IEEE, pp. 211218, Sept 2011.
dc and 380v dc distribution systems for datacenters, in Telecommuni- [23] V. Salas, E. Olas, A. Barrado, and A. Lzaro, Review of the maximum
cations Energy Conference (INTELEC), 2014 IEEE 36th International, power point tracking algorithms for stand-alone photovoltaic systems,
pp. 17, Sept 2014. Solar Energy Materials and Solar Cells, vol. 90, no. 11, pp. 1555
[4] H. Kakigano, Y. Miura, and T. Ise, Distribution voltage control for dc 1578, 2006.
microgrids using fuzzy control and gain-scheduling technique, Power [24] Y. Zhang, H. J. Jia, and L. Guo, Energy management strategy of
Electronics, IEEE Transactions on, vol. 28, no. 5, pp. 22462258, 2013. islanded microgrid based on power ow control, in 2012 IEEE PES
[5] J. Guerrero, J. Vasquez, J. Matas, L. de Vicua, and M. Castilla, Innovative Smart Grid Technologies (ISGT), pp. 18, Jan 2012.
Hierarchical control of droop-controlled ac and dc microgrids a general [25] D. Wu, F. Tang, T. Dragicevic, J. Vasquez, and J. Guerrero, Au-
approach toward standardization, IEEE Transactions on Industrial tonomous active power control for islanded ac microgrids with pho-
Electronics, vol. 58, no. 1, pp. 158172, 2011. tovoltaic generation and energy storage system, Energy Conversion,
[6] H. Kanchev, D. Lu, F. Colas, V. Lazarov, and B. Francois, Energy IEEE Transactions on, vol. 29, pp. 882892, Dec 2014.
management and operational planning of a microgrid with a pv-based

298