Energy xxx (2015) 1e13

Contents lists available at ScienceDirect

Energy
journal homepage: www.elsevier.com/locate/energy

A micro-grid ensuring multi-objective control strategy of a power

electrical system for quality improvement
Mouna Rekik, Achraf Abdelkafi, Lotfi Krichen*
National Engineering School of Sfax, Control and Energy Management Laboratory (CEMLab), BP 1173, 3038 Sfax, Tunisia

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

Article history: The increasing integration of fluctuating and NL (non-linear) loads in the main grid can introduce
Received 12 September 2014 problems to the distribution power system quality. This study was interested in a MG (micro-grid) based
Received in revised form on RDG (renewable distributed generator) to participate in system services and to improve the efficiency
7 March 2015
of the power electrical system. The purpose of this paper is to investigate a multi-objective control
Accepted 13 May 2015
strategy for the integration of an MG into electrical network in order to ensure simultaneously active
Available online xxx
power supply, reactive power compensation, harmonic current damping and grid frequency regulation.
This control is mainly composed of two parts: the first one is the “NL Loads currents identification
Keywords:
Micro-grid
system” used to extract the fundamental active current from the disruptive one in order to provide the
Renewable distributed generator required harmonic and reactive currents to the considered NL loads. The second one is the “active power
Power supply transfer and frequency control algorithm” used to manage the MG in six operation modes in order to
Harmonic compensation control the fundamental active power flow exchange between the MG and the electrical network making
Grid frequency regulation the grid frequency in an allowable range of stability.
© 2015 Elsevier Ltd. All rights reserved.

1. Introduction as a local energy system containing various microsources, and

loads, and it can operate either connected or disconnected from the
Recently, the integration of distributed generation systems main grid [6].
based on renewable energy resources like photovoltaic system, The application of this concept provides an opportunity to RDG
wind power conversion system, biomass and geothermal resources, for improving the stability, the reliability of supply and the power
in the future distribution grid, has represented an area of growing quality indices of the future utility grid by controlling different
interest. In fact, electricity generation based on renewable re- aspects of the electrical system such as harmonic and reactive
sources is a reliable, efficient, clean and environmentally friendly power distortions, low power factor, voltage and frequency fluc-
compared to the traditional centralized energy system [1,2]. How- tuations, imbalance voltage, flicker, etc. These aspects can cause
ever, their exploitation was mostly employed for autonomous critical and sometimes several power quality problems such as
application or for reducing electrical network power generation by malfunction protection systems, overheating of conductors and
feeding local consumers [3e5]. electrical equipment, additional losses in power systems and
The need for more flexible power electric systems with good reduction in equipment lifetime [7]. And due to the significant in-
quality generation, leads to the development of the RDG (renew- crease of electricity consumption and to the growth integration of
able distributed generator) applications, that are predicted to play sensitive and NL (non-linear) loads in the electric main grid, con-
an increasing role in the future power systems quality improve- ventional techniques are today insufficient to guarantee the power
ment. New concepts have been elaborated to enhance the visibility electrical system quality in different buses [8]. For that, many
of RDG to participate in system services and guarantee a satisfac- recent works are done for system services using MG based on RDG.
tory operation of the future main grid. From these new concepts, In Ref. [9], a control technique to minimize distribution networks
particular interest focuses on the MG (micro-grid), which is defined nodal voltages deviations by supervising the reactive power
generated by photovoltaic plants inverters has been proposed.
Authors in Ref. [10] have presented a control technique for
* Corresponding author. Tel.: þ216 74 274 418; fax: þ216 74 275 595. distributed generators and single phase distribution static
E-mail address: lotfi[email protected] (L. Krichen). compensator in an MG. This control has been proposed with the

http://dx.doi.org/10.1016/j.energy.2015.05.050
0360-5442/© 2015 Elsevier Ltd. All rights reserved.

Please cite this article in press as: Rekik M, et al., A micro-grid ensuring multi-objective control strategy of a power electrical system for quality
improvement, Energy (2015), http://dx.doi.org/10.1016/j.energy.2015.05.050
2 M. Rekik et al. / Energy xxx (2015) 1e13

objective of reactive power compensation. An energy management first converter is used to regulate the DC bus voltage and to
strategy for a hybrid distributed generators composed by photo- extract the maximum power from the renewable energy source
voltaic, diesel and battery systems has been proposed in Ref. [11]. and the second is needed for loads supplying and system ser-
This strategy focuses on the optimization of the grid frequency vices participating. The considered NL loads are full-wave
profile by acting on the active power produced by the whole system thyristor converters which supply RL loads, and these NL loads
inverters. A study in Ref. [12] has developed a hierarchical method draw harmonic currents from the main grid. The grid is modeled
that enables photovoltaic power inverters to adjust its active power as generation station based on synchronous machine, trans-
outputs in order to ensure the frequency regulation service to po- mission lines operating at different voltage levels and variable
wer systems without association of energy storage systems. In loads.
Ref. [13] authors were interested in the harmonic compensation The functioning of this studied system is as follows. The
service. They have proposed a harmonic impedance synthesis required compensating harmonic and reactive currents are ob-
method based on band-pass filter and resonant integrators for tained by measuring the NL loads currents and passing them
voltage-controlled RDG interfacing converters to ensure harmonic through the “NL loads currents identification system” to remove
voltage damping on the distribution network. For the same the fundamental frequency component. After corrective currents
objective, the idea of using photovoltaic inverters as virtual har- have been calculated, the reference compensating currents can be
monic resistances to perform residential system harmonics generated for each WT inverter command to cancel harmonic and
damping is investigated in Ref. [14]. Similarly, a strategy based on reactive powers at the PCC point. The fundamental active power
integer-order and fractional-order controllers for WT (wind tur- supplied to the NL loads is usually ensured from the WF. In fact, in
bine) () system with different power converter topologies to case of wind availability, the WF which operates in the MPPT
compensate the harmonic current injected into the electrical (maximum power point tracking) supplies the fundamental active
network is explored in Ref. [15]. All these methods have proposed power demanded by NL loads and the excess will be injected into
good solutions to ensure power electrical system quality using MG the grid. But in case of WF lack of generation, it generates the
based on RDG. However, each of them focuses on a specific objec- maximum fundamental active power demanded by NL loads and
tive task, and they don't take into account various problems in the lack will be absorbed from the grid. Considering the inter-
power systems. mittent and fluctuating behavior of wind energy and load de-
In this paper, a multi-objective control strategy for connection of mands, the fundamental active power exchanged between grid
an MG including RDG resources to the main grid to participate in and WF should be assured according to well-defined standards
the system services and to ensure power electrical system quality without disturbing the grid stability and more specifically the grid
improvement is proposed. Compared to present compensation frequency stability. For that, the “active power transfer and fre-
techniques, the proposed MG control technique favors application quency control algorithm” shown in Fig. 1 is proposed. This al-
of different types of renewable energy resources, e.g., WF (wind gorithm provides as an output the various operating modes of the
farm) generation and photovoltaic panels, to solve various prob- studied WF. These modes will be integrated into the voltage-
lems in the electrical network. In fact, the main contribution of this source inverter control of each WT to monitor the powers ex-
control is to monitor the MG in six operation modes in order to change at the PCC point making the grid frequency in an allowable
make it able to ensure, simultaneously, at the PCC (point of com- range of stability.
mon coupling) the harmonic current damping, the reactive power
compensation, the load active power and the grid frequency 2.1. WF mechanical model
regulation.
The description and the modeling of the studied MG are In this paper, the studied WF is composed of similar WT units
developed in Section 2. In Section 3, the control strategy applied to with a short distance between them. The wind speeds profiles
the studied MG was described. An “NL Loads currents identification applied to these units are near. In order to reduce the computation
system” based on the instantaneous powers theory is implemented time and increase the simulation speed when simulating WF with
to calculate the reference currents involved in each RDG inverter large number of identical WT units on power system, the WF can be
control to ensure harmonic and reactive power compensation. represented by an aggregate model that consists of one equivalent
Then, an “active power transfer and frequency control algorithm” is WT [16,17]. This equivalent WT receives as input the mean value of
proposed to supervise the NL loads fundamental active power, to winds applied to the group of WT units as follows [18]:
control the fundamental active power transfer between the MG and PnT
the main grid and to provide the grid frequency service. The i¼1
Vi
Vm ¼ (1)
simulation results and the conclusion of this paper are given in nT
Section 4 and Section 5, respectively.
Such equivalent WT presents as output a rated power equal to nT
2. System description and modeling times the rated power of individual WT unit [18]:

nT 3
The system under analysis is composed of an electrical network Pmec ¼ rSVm Cp (2)
2
and an MG including RDG and NL loads. The type of RDG may be AC
sources like WT generators which may be connected to DC bus
through AC/DC converters, or DC power sources like photovoltaic 2.2. Electrical network model
panels with their DC/DC converters. In any case, the power elec-
tronics converters (DC/DC or AC/DC) are employed to maximize the Currently, the majority of the electrical energy is produced by SG
generated power from renewable sources and the DC/AC converters (synchronous generators) power stations [19]. The synchronous
are focused to improve the AC grid power quality. machine is an electromechanical converter which, from the me-
In this work, the RDG source is considered as a WF (wind chanical energy supplied by a motor, injects into the grid electrical
farm) (WF) power generation as it is sketched in Fig. 1. This WF is energy in three-phase forms. The studied synchronous machine
made up of nT VSWT (variable speed wind turbines). Each one was modeled in this paper in d-q frame axes as the following
includes two power electronic converters AC/DC and DC/AC. The equations [20]:

Please cite this article in press as: Rekik M, et al., A micro-grid ensuring multi-objective control strategy of a power electrical system for quality
improvement, Energy (2015), http://dx.doi.org/10.1016/j.energy.2015.05.050
 M. Rekik et al. / Energy xxx (2015) 1e13 3

Fig. 1. Architecture of the studied system.

2.3. NL loads model

2 3 2 3
Vd rs lq uep 0 0 MsQ uep
6 Vq 7 6 ld uep 7 Modeling and characterization of harmonic sources are impor-
rs Msf uep MsD uep 0
6 7 6 7 tant steps in THD and harmonics studies. Different NL loads are
6 Vf 7 ¼ 6 0 0 rf 0 0 7
6 7 6 7 modeled as a harmonic current source in parallel with the funda-
4 0 5 4 0 0 0 rD 0 5
0 0 0 mental component as follows [22]:
0 0 rQ
2 3 X
id INLL ¼ Ih sinðhut þ fh Þ (7)
6 iq 7
6 7 h
6 6 if 7
7
4 iD 5 The harmonic currents created by balanced NL loads typically
iQ have orders of h ¼ 5, 7, 11,13 … with h ¼ 1 corresponding to the
2 3 2 3 fundamental frequency.
ld 0 Msf MsD 0 id
6 0 lq 0 0 MsQ 7 6 iq 7
6 7d6 7 3. Grid control and ancillary services
þ6 6 Msf
0 lf MfD 0 7 6
7 dt 6 if 7
7 (3)
4 MsD 0 MfD lD 0 5 4 iD 5
0 MsQ 0 0 lQ iQ In this work, the RDG system based on WF distributed genera-
tion is introduced to the grid in order to ensure the following
services:
duep
J ¼ Te  Tm (4)  Compensation of NL loads harmonic currents,
dt
 Compensation of loads reactive power,
The electromagnetic torque 0 Te 0 is given by:  Capture of maximum energy from wind to provide active power,
 Grid frequency regulation to make it at an allowable range of
stability.
3  
Te ¼ np fd iq  fq id (5)
2

with 0 fd 0 and 0 fq 0 are the direct and transverse stator. The following 3.1. NL loads currents identification system
relationships can be written [21]:
As shown in the lower part of Fig. 1, the harmonic and reactive
2 3 compensation is ensured through proper control of each WT
id interface converter. In fact, these inverters are controlled to inject at
   6 iq 7
fd ld 0 Msf MsD 0 6 7 PCC, equal but opposite harmonic component of NL loads in addi-
¼ 6 if 7 (6)
fq 0 lq 0 0 MsQ 6 7 tion to the reactive component current to cancel the original har-
4 iD 5
monic distortion and to eliminate the reactive power flow to the
iQ grid that are responsible for power network pollution.

Please cite this article in press as: Rekik M, et al., A micro-grid ensuring multi-objective control strategy of a power electrical system for quality
improvement, Energy (2015), http://dx.doi.org/10.1016/j.energy.2015.05.050
4 M. Rekik et al. / Energy xxx (2015) 1e13

For each WT inverter reference currents computation, the “NL

Loads currents identification system” presented in Fig. 2 is
employed. The operation of this block system is based on the
instantaneous powers theory which has the advantage of selecting
the disturbance compensation with precision, rapidity and ease of
implementation [23]. In fact, as it can be seen in Fig. 2, the source
voltages at PCC 0 V1 d ; V1 q 0 and the NL loads currents 0 INLL d ; INLL q 0
data are extracted by this block system on the d-q reference frame.
The conventional instantaneous active and reactive powers
0P 0
NLL ; QNLL absorbed by the NL loads are written in matrix form as
shown in this relation.
    
PNLL V1 d V1 q INLL d
¼ (8)
QNLL V1 q V1 d INLL q

Each of 0 PNLL 0 and 0 QNLL 0 powers can be split into continuous (DC) Fig. 2. NL Loads currents identification system.

and alternative (AC) components as illustrated below:

In this paper, the studied WF is used not only as an active power
~
PNLL ¼ P þ P
~ (9) filter to compensate harmonic and reactive power, but also, to
QNLL ¼ Q þ Q capture the maximum of energy from the wind in order to provide
0 0 the total or a portion of NL loads fundamental active power. The
where: 0 P and 0 Q are the active and reactive DC powers related to
~0 excess or lack of fundamental active power generated by the WF
the fundamental frequency of the NL loads current and voltage. 0 P
0 ~ 0 will be injected or absorbed into or from the grid, respectively.
and Q are the active and reactive AC powers related to the har-
However, the uncontrollable exchange of active power between WF
monic components of the NL loads current and voltage.
and electrical power system may disrupt the stability of electrical
The objective of the “NL Loads currents identification system” is
network frequency [6,26]. In fact, it can create derivation which can
to separate alternative and constant components. For this purpose,
exceed the normalized frequency variations mentioned in stan-
2 s-order low pass filters are used. Second-order filters are more
dards. Therefore, in order to ensure an active power transfer with
employed than other filters because they provide good perfor-
the electrical power system and to make the grid frequency in an
mance and can filter the oscillations properly [24,25]. So the AC
~ Q~ 0 can be obtained simply as shown in Fig. 2 allowable range of stability, the “active power transfer and fre-
power components 0 P;
quency control” algorithm is proposed in this work.
by the difference between the input signal and the filtered one.
Considering Equation (9), the NL loads currents in the d-q reference
frame can be divided into four components: active current at the 3.2. Active power transfer and frequency control algorithm
fundamental frequency0 INLL fd 0, reactive current at the fundamental
frequency0 INLL fq 0, active harmonic current 0 INLL hd 0 , and reactive The consequence of a mismatch between the grid supply
harmonic current0 INLL hq 0 . So we can write in matrix form: (generation) and demand (load and network losses) for active

             
INLL d 1 V1 d V1 q P 1 V1 d V1 q 0 1 V1 d V1 q ~
P 1 V1 d V1 q 0
¼ þ þ þ ~ (10)
INLL q V 2 V1 q V1 d 0 V 2 V1 q V1 d Q V 2 V1 q V1 d 0 V 2 V1 q V1 d Q
|fflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflffl{zfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflffl} |fflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflffl{zfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflffl} |fflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflffl{zfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflffl} |fflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflffl{zfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflffl}
INLL fd Eqð10:1Þ INLL fq Eqð10:2Þ INLL hd Eqð10:3Þ INLL hq Eqð10:4Þ

with V 2 ¼ Vd2 þ Vq2 power is a change in the rotational energy stored in the rotating
After NL loads currents identification at d and q-axis, to use the mass of the synchronous electricity generators, and hence, a drift
studied WF as an active power filter, harmonic and reactive com- in the system frequency. As shown in the top part of Fig. 3, the
ponents of NL loads currents should be shared and supplied by the change of the demanded power 0 PD PCC 0 to that of the frequency 0 f 0
inverters of the nT wind power generations. For this purpose, as is as follows: A surplus in demand creates a frequency decrease,
shown in the lower part of Fig. 1, it is sufficient to set d and q- and a lack in demand leads to an increase in the frequency
components of reference currents involved in each WT inverter because when there are more loads, it becomes harder to turn the
control to ensure harmonic and reactive compensation generators at the nominal frequency and vice versa [27]. All the
0I 0
comp ref d ; Icomp ref q equal to the alternative d and q-components generating equipment in an electric system is designed to operate
0I 0 0 0
NLL hd ; INLL hq and to the DC q-component INLL fq of the NL loads within very strict frequency margins. Grid codes specify that all
currents divided by the number of wind power inverters as follows: generating plants should be able to operate continuously between
a frequency range around the nominal frequency of the grid,
usually between 49.5 Hz and 50.5 Hz in Europe [28]. Operation
INLL hd outside these limits would damage the generating plants, so even
Icomp ref d ¼
nT very short duration deviations from the nominal frequency values
  (11) would trip load shedding relays and generation capacity would be
INLL fq þ INLL hq lost. Maintaining the frequency at its target value requires that the
Icomp ref q ¼
nT active power produced and/or consumed be controlled to keep the

Please cite this article in press as: Rekik M, et al., A micro-grid ensuring multi-objective control strategy of a power electrical system for quality
improvement, Energy (2015), http://dx.doi.org/10.1016/j.energy.2015.05.050
 M. Rekik et al. / Energy xxx (2015) 1e13 5

Fig. 3. Description of different selected modes by the “active power transfer and frequency control algorithm”.

Please cite this article in press as: Rekik M, et al., A micro-grid ensuring multi-objective control strategy of a power electrical system for quality
improvement, Energy (2015), http://dx.doi.org/10.1016/j.energy.2015.05.050
6 M. Rekik et al. / Energy xxx (2015) 1e13

loads and grid generation in equilibrium [29]. This equilibrium

must be preserved in the face of fluctuations in loads and un-
controlled generations. Recently, there has been motivation for
RDG to provide grid frequency regulation service by controlling its
real power output in order to assist in balancing total power
generated on the grid with total power consumed [30,31]. In this
paper, the studied WF also focuses on frequency regulation for the
utility grid. It measures simultaneously the frequency of the utility
f and the fundamental active power 0 DP 0 to be transferred between
it and the grid. This power is calculated by subtracting the
fundamental active power demanded by NL loads 0 PNLL f 0 from the
maximum fundamental active power0 PWF MPPT 0 generated by the
WF. And it is considered as excess or lack of the studied WF active
power generation that will be injected or absorbed into or from
Fig. 5. DC/AC converter control.
the grid respectively. The exchange of 0 DP 0 between grid and WF
should be assured according to well-defined standards without
 Control of the fundamental active power transfer between the
disturbing the grid stability and to keep the frequency in the
studied WF and the electrical power system in case of WF excess
standards range. For that, the studied WF can operate, according to
or lack power generation.
the transfer power 0 DP 0 and the frequency variations 0 Df 0 in six
 Regulation of grid frequency by injection or absorption of
operating modes. To monitor the different operating modes of the
fundamental active power at the PCC point.
studied WF, the “active power transfer and frequency control al-
 Ensuring a compromise in specific modes between the
gorithm” presented in Fig. 4 is proposed. This algorithm ensures
exchanged power and the frequency regulation.
the following roles:

The operation of this algorithm is well described in Fig. 3 and

 Supervision of NL loads fundamental active power.
Fig. 4. Indeed, at each instant, the sign of 0 DP 0 (positive or negative)

Fig. 4. Active power transfer and frequency control algorithm.

Please cite this article in press as: Rekik M, et al., A micro-grid ensuring multi-objective control strategy of a power electrical system for quality
improvement, Energy (2015), http://dx.doi.org/10.1016/j.energy.2015.05.050
 M. Rekik et al. / Energy xxx (2015) 1e13 7

Fig. 6. DC Bus and AC/DC converter control.

and the frequency variation0 jDf j0 (included or excluded from the

desired margin) are detected simultaneously. It is noticed that
0 jDP 0 0 0
49:8 j and jDP50:2 j are fundamental active powers correspond-
ing to the 49.8 Hz and 50.2 Hz frequency, respectively. In fact, when
these powers are injected or absorbed into or from the grid, we will
get the limits of the desired frequency range. These powers are
calculated as shown in these relations:

jDP49:8 j ¼ PD PCC  PDs max

(12)
jDP50:2 j ¼ PD PCC  PDs min

This algorithm works as below:

When0 DP_00 , meaning that there is an excess of WF funda-
mental active power production and 0 jDf j  0:2 Hz0 meaning that Fig. 7. Electrical network state before NL loads and WF integration: Demanded power
the frequency variation is included in the desired area. In this case, in PCC (a) e Grid frequency (b) and grid currents (c).
0 DP 0 should be injected into the grid. The injection of active power

into the electrical network causes a reduction of demanded power

0P 0
D PCC and hence an increase of frequency [27,32]. The problem frequency rises, but it does not achieve its optimal range. This is the
here is when injecting 0 DP 0 into the grid, the frequency can be second case of mode 2 (mode 2.2) that allows the injection of
increased without leaving its desired margin 0 f  50:2 Hz0 as it can power with partial frequency regulation. When
exceed it0 f _50:2 Hz0 . For this, it is necessary to compare the power 0 jDP 0 0 0
49:8 j  jDPj  jDP50:2 j , the WF will inject DP into the grid, but
0 DP 0 injected into the grid with the power 0 DP 0
50:2 corresponding to in this situation, the frequency will achieve its optimal mar-
the 50.2 Hz. When0 jDPj  jDP50:2 j0 , the WF can inject 0 DP 0 into the gin0 49:8 Hz  f  50:2 Hz0 . This is the third case of mode 2 (mode
grid without excluding the frequency from its required margins. 2.3) where we will get a total frequency regulation.
This is the first case of mode 2 (mode 2.1). Otherwise, When0 jDP50:2 j3jDPj0 , the WF will operate without MPPT. It will
when0 jDPj_jDP50:2 j0 , the WF will operate without MPPT mode. It meet the demands of NL loads and will inject into the
network0 DP 3jDPj0 for the grid frequency does not exceed
0
will supply the NL loads and it will inject into the network another
fundamental active power 0 DP 3jDPj0 with 0 DP ¼ jDP50:2 j0 . In
0 0
50.2 Hz. This is the second case of mode 3 (mode 3.2).
this situation, the utility frequency never leaves the optimal mar- Now, when 0 DP_00 and0 jDf j_0:20 and 0 Df 300 meaning there is
gins (the frequency will be equal to 50.2 Hz). This is the first case of an excess of production with the case of over
mode 3 (mode 3.1). frequency0 50:2 Hz3f 350:5 Hz0. Under this condition, the injection
When 0 DP_00 and 0 jDf j_0:20 and 0 Df _00 , meaning that there is of 0 DP 0 into the grid increases the frequency more, which can lead to
an excess of WF power generation with the case of under adverse damages in the electrical power and the electrical equip-
frequency0 49:5 Hz3f 349:8 Hz0. In this condition, the injection of ment stabilities. This is mode 5, in which the WF stops feeding
0 DP 0 into the grid will increase the frequency. Three levels of
power to the grid. It operates without MPPT to satisfy only the NL
increasing frequency can be sensed according to 0 DP 0 states: loads demands.
When0 jDPj3jDP49:8 j0 , the WF injects 0 DP 0 into the network and the Now treating the case of WF lack of generation under different
states of frequency variation0 Df 0 :
When 0 DP  00 and the frequency variation is in its optimal
Table 1
margins0 jDf j  0:2 Hz0 , the power0 DP 0 should be absorbed from the
Studied system parameters.
network to meet the fundamental active power required by NL
Wind turbine Number of blades 3 loads. The absorption of active power from the electrical network
blade radius 41 m
causes an increase in power required 0 PD PCC 0 and hence a drop in
r 1.225kgm3
PMSG Nominal power 2 MW
grid frequency 0 f 0 [27,32]. Therefore, it is necessary to compare the
p 120 absorbed power0 jDPj0 with0 jDP49:8 j0 . Because unsupervised ab-
Ls 1.285mH sorption of 0 DP 0 from the network may drop the frequency without
fm 4.813Wb leaving its desired margin0 f  49:8 Hz0 is the first case of mode 1
DC bus Cbus 2200mF
(mode 1.1), as it can exceed it0 f 349:8 Hz0 on the first case of mode 4

Please cite this article in press as: Rekik M, et al., A micro-grid ensuring multi-objective control strategy of a power electrical system for quality
improvement, Energy (2015), http://dx.doi.org/10.1016/j.energy.2015.05.050
8 M. Rekik et al. / Energy xxx (2015) 1e13

never leaves its optimal margins (the frequency will be equal to

49.8 Hz).
When 0 DP  00 and0 jDf j_0:20 and0 Df 300 , meaning that there is
a lack of WF power generation and a rise in grid frequency
0 50:2 Hz3f 350:5 Hz0 , the absorption of 0 DP 0 from the grid will

decrease the frequency that can reach 3 levels according to 0 DP 0

statements indeed: When 0 jDPj3jDP50:2 j0 the frequency decreases
but does not reach 50.2 Hz. This is the second case of mode 1 (mode
1.2), which allows NL loads to absorb0 DP 0 from the network with
partial frequency regulation. When0 jDP49:8 j  jDPj  jDP50:2 j0 in
this situation the frequency decreases and reaches its optimal
margins [49.8, 50.2 Hz], it is the third case of mode 1 (mode 1.3).
The third level is attained when0 jDPj_jDP49:8 j0 . This is the second
case of mode 4 (mode 4.2) which allows disconnection of the minor
priority NL loads for the frequency drop does not exceed 49.8 Hz.
Now, when 0 DP  00 and 0 jDf j_0:20 and 0 Df _00 meaning that
there is a lack of WF power generation with the case of under
frequency 0 49:5 Hz3f 349:8 Hz0, the absorption of 0 DP 0 from the
network will further reduce the frequency level which can disrupt
the grid and the electrical equipments stabilities. This is mode 6,
which disconnects the minor priority NL loads until the cancella-
tion of the lack and consequently the absorption of fundamental
active power from the network is shut down.

3.3. Wind turbine control side

As shown in Fig. 1, each WT is controlled by its own control

block. The control strategy used in each WT is divided into 2 parts.
The first one illustrates the “DC/AC converter control” and the
second represents the “DC Bus and AC/DC converter control” as
shown in Fig. 5 and Fig. 6, respectively.
Fig. 8. Electrical network state after NL loads and before WF integration: Demanded The objective of “AC/DC converter control” is to get each WT to
power in PCC (a) e Grid frequency (b) and grid currents (c). give harmonic and reactive power compensation. In addition, each
WT must deliver an appropriate fundamental active power0 Pref 0,
(mode 4.1). In this last mode, the minor priority NL loads will be MPPT or without MPPT according to the selective operating mode,
disconnected in order to alleviate the demands on the network. to meet NL loads fundamental active power and to participate in
Consequently, the absorbed power from the grid system services. In fact, the WT reference currents applied on each
becomes0 DP 3jDPj0 with0 DP ¼ jDP49:8 j0 and the grid frequency
0 0 WT DC/AC converter on d-q reference frame 0 IWT ref d ; IWT ref q 0 is

Fig. 9. NL loads active and reactive powers. Fig. 10. NL active and reactive load powers identification.

Please cite this article in press as: Rekik M, et al., A micro-grid ensuring multi-objective control strategy of a power electrical system for quality
improvement, Energy (2015), http://dx.doi.org/10.1016/j.energy.2015.05.050
 M. Rekik et al. / Energy xxx (2015) 1e13 9

Fig. 11. NL loads currents identification. Fig. 12. Active power transfer and frequency control algorithm inputs: Average wind
speeds (a) e Grid frequency before regulation (b) and transfer power before regulation
(c).
deduced by adding the compensating reference currents
0I 0
comp ref d ; Icomp ref q calculated by the “NL Loads currents identi-
fication system” to the fundamental WT reference currents situations the wind generator debits only a torque similar to the
0I 0
WTf ref d ; IWTf ref q corresponding to the WF operating mode that asked torque 0 Tem ref 0 to ensure the balance between production
is selected by the “active power transfer and frequency control al- and consumption and hence a continued voltage at DC bus.
gorithm”. Then, the reference currents are expressed in a-b-c
reference frame through the inverse Park transformation
4. Simulation results
0IWTf ref abc 0. These currents are compared with the injected
inverter currents 0 IWT abc 0 using PI control loops. The differences
In order to evaluate the performance of the proposed approach,
between both currents are then compared to a triangular PWM
a model of the AC grid, NL loads and WF generators have been
signal to provide the control command of WT voltage source
performed. The simulated WF is made up of fifteen WT generators
inverter switches.
(nT ¼ 15) of a 2 MW variable-speed PMSG, with a total installed
Fig. 6 illustrates the control applied to the “DC Bus and AC/DC
capacity of 30 MW. The parameters of the proposed control strategy
converter control”. The role of this control is to regulate the DC
used in the simulation conditions are provided in Table 1. The
voltage bus 0 UBus 0 at a fixed value and to extract the necessary po-
following subsections present the process of simulation before and
wer to supply NL loads and to participate in the system services.
after integration of the variable NL loads and the studied multi-
The principle of this control is based on the vector control applied
objective WF in the electrical power system.
to permanent magnet synchronous generator (PMSG), which con-
sists in imposing a reference direct current0 Isd ref 0 equal to zero.
This allows the electromagnetic torque to behave in a similar way 4.1. Initial electrical network state
as in the quadratic component of the stator current 0 Isq ref 0 so:
Initially, neither the NL loads nor the studied WF is connected to
Tem ref the network. In fact, the electrical network is considered just to
Isq ref ¼ (13) feed the constant and the variable balanced linear loads presented
pfm
in Fig. 1. In normal and stable conditions, the network produces
The adjustment of the DC bus voltage is assured by applying a 215 MW in order to supply the demands; in this case, the frequency
reference torque 0 Tem ref 0 at the terminal of PMSG equivalent to the is equal to 50 Hz. But, variations in active power required by the
required torques needed to supply NL loads and to participate in loads generate an electrical network over or under production and
system services. In fact, the “DC Bus and AC/DC converter control” therefore frequency levels variations. Indeed, if the demands are
system receives instantaneous information about the required greater than the stable network generation (215 MW in this paper),
modulated current0 Im G 0 . The imposed reference torque on WT a drop in frequency will be obtained and vice versa. Fig. 7 describes
control generator is obtained by dividing the required WT power the network state before NL loads and studied WF integration.
0P 0 0 0
WT by the rotor mechanical speed Um of the WT PMSG. In such According to Fig. 7a and b, the impact of the active power demand

Please cite this article in press as: Rekik M, et al., A micro-grid ensuring multi-objective control strategy of a power electrical system for quality
improvement, Energy (2015), http://dx.doi.org/10.1016/j.energy.2015.05.050
10 M. Rekik et al. / Energy xxx (2015) 1e13

0P 0
DPCC variations in PCC point at the network frequency is
remarkable. During the intervals [0, 55 s] and [300, 350 s] the fre-
quency is equal to 50 Hz because there is equilibrium between
production and consumption0 PDPCC ¼ 215 MW 0 . In [55, 175 s] the
demand is less than the production, which will lead to an increase
in frequency. Whereas, during [175, 300 s] demand exceeds pro-
duction and hence we will get a frequency drop. It is also notable as
shown in Fig. 7a and Fig. 7b, that when an instantaneous active
power demand 0 PDPCC 0 variation is carried, the frequency needs a
period of time to change from a level to another. This period de-
pends on the amplitude of 0 PDPCC 0 variations. In fact, as shown in
Fig. 7b, active power demand 0 PDPCC 0 variation at 0 t1 ¼ 119s0 is
more important that at 0 t2 ¼ 300s0 for that, 0 Dt1 _Dt2 0 . Fig. 7c rep-
resents the requested network currents in PCC point. It is noticeable
that these required currents are well sinusoidal, but they present
variations due to loads power changes in PCC point.

4.2. Electrical network state after NL loads connection and before

WF integration

In this section, NL loads were connected and supplied by the

studied electrical power system. So the power required from the
network 0 PDPCC 0 becomes as presented in Fig. 8a. It is noticeable
that these NL loads have negative influences on the grid quality and
stability. They disrupt the network frequency on the one hand. In
fact, as discussed in Fig. 8b, frequency exceeds its normalized
margins [49.5, 50.5 Hz]. And they pollute the utility currents (THD
very strong) as mentioned in Fig. 8c on the other hand which can
cause several power quality problems such as malfunction pro-
tection systems, reduction in equipment lifetime and additional
losses in power systems. For this, in the next section, a multi-
objective WF will be integrated in the electrical network to
address these issues.

4.3. Electrical network state after NL loads connection and WF

integration

In this paper, the studied WF is added for the following

objectives:

 To compensate NL loads harmonic power,

 To compensate NL loads reactive power,
 To provide active power,
 To make the grid frequency in an allowable range of stability.

To validate the capabilities of the proposed control strategy

applied on the multi-objective WF, we began by simulating the
behavior of the NL loads. In fact, Fig. 9a and Fig. 9b illustrate the
active and the reactive powers demanded by the NL loads,
respectively. Using “NL Loads currents identification system” block,
these powers are composed of two components as shown in Fig. 10.
The curves in red illustrate the continuous active and reactive
powers components related to the fundamental NL loads current
and voltage and the curves in blue represent the alternative active
and reactive powers component related to the sum of the NL loads
currents and voltages disturbances.
After power identification, the compensating reference currents
that will be shared among different WT inverters is determined by
adding the NL loads harmonic currents presented in Fig. 11c with
NL loads fundamental reactive currents presented in Fig. 11b. The
NL loads fundamental active currents presented in Fig. 11a will be
usually supplied by the WF. But in case of generation lack, the WF
Fig. 13. WF operating modes with its own special cases.
satisfies the maximum of NL loads fundamental active currents and
the lack will be absorbed from the grid.

Please cite this article in press as: Rekik M, et al., A micro-grid ensuring multi-objective control strategy of a power electrical system for quality
improvement, Energy (2015), http://dx.doi.org/10.1016/j.energy.2015.05.050
 M. Rekik et al. / Energy xxx (2015) 1e13 11

Fig. 16. Zoom in case of WF excess of generation.

on the other the power exchanged0 DP 0 with the powers lim-

its0 DP50:2 0 and 0 DP49:8 0 which are shown in Fig. 12c. Then, it presents
as an output the various operating modes of the studied WF. These
modes will be integrated into each WT inverter control to monitor
the powers exchange at the PCC point without disturbing the grid
Fig. 14. “Active power transfer and frequency control” algorithm impact on the grid frequency. Fig. 13 shows the transitions between the six modes
frequency: Transfer power after regulation (a) and grid frequency after regulation (b).
during the whole sequence. Some modes of operation are obtained
from different cases. For example mode 1 will be activated when
one of these cases: mode 1.1 or mode 1.2 or mode 1.3 is selected. It
The excess or lack of WF fundamental active power0 DP 0 is is noticed that only one case and only one mode is active at each
controlled and supervised by the “active power transfer and fre- moment.
quency control algorithm” to be injected or absorbed into or from From Fig. 14a, it is observable that both power 0 DP 0 before and
the grid, respectively, without disturbing the grid frequency and to after regulations are equal in mode 1 and mode 2 for two reasons:
make it the most possible in the optimum range of stability [49.8, because injection or absorption of 0 DP 0 does not exceed the fre-
50.2 Hz]. This algorithm receives as input the grid frequency states quency from its optimal margin, or else, because the transfer of 0 DP 0
shown in Fig. 12b, and the power0 DP 0 that will be exchanged be- regulates the frequency level to be close to its desired margin
tween the WF and the grid as shown in Fig. 12c. This power de- (mode 1.2 and mode 2.2). However, in specific modes, the powers
pends mainly on the NL loads fundamental active power 0 PNLL f 0 transferred 0 DP 0 before and after regulation are not the same. In
shown in Fig. 11a and on the maximum WF active power 0 PWF MPPT 0 fact, 0 DP 0 after regulation is equal to zero in both modes 5 and 6, and
generated according to the average applied wind speed presented hence, no change will be accrued in the frequency. Because, when
in Fig. 12a. both0 DP 0 before and after regulation are equal in these modes, the
The operation of this algorithm is well explained in Section 3.2. frequency can submit strange derivations of maximum levels [49.5,
Indeed, at every moment, it senses on the one hand the frequency 50.5 Hz] which can damage the electrical equipment and the
status (included or excluded from [49.8, 50.2 Hz]) and it compares personnel. Moreover, 0 DP 0 after regulation is equal to one of these
limitations powers 0 DP49:8 0 or 0 DP50:2 0 in modes 3 or 4, respectively
because the injection or the absorption of 0 DP 0 before regulation
exceeds the grid frequency from its desired margin [49.8, 50.2 Hz].
But, 0 DP 0 after regulation limits the frequency variability to 49.8 Hz
or 50.2 Hz.

Fig. 15. Behaviours of NL loads currents, wind turbine currents and grid currents at
PCC after proposed control application. Fig. 17. Zoom in case of WF lack of generation.

Please cite this article in press as: Rekik M, et al., A micro-grid ensuring multi-objective control strategy of a power electrical system for quality
improvement, Energy (2015), http://dx.doi.org/10.1016/j.energy.2015.05.050
12 M. Rekik et al. / Energy xxx (2015) 1e13

Fig. 18. Zoom in case of transition between WF excess of generation to the WF lack of
generation.

Fig. 7b, Fig. 8b and Fig. 14b show respectively the grid frequency
in these different situations: Before NL loads and WF integration, Fig. 20. Magnitude spectrum: After NL loads connection to the grid and before WF
after NL loads connection and before WF integration, and finally, integration (a) e After harmonic and reactive power compensation by the studied WF
(b).
after NL loads and WF integration. It is observable that in the first
situation the frequency exceeds its desired margin during long
periods of time. Moreover, in the second situation the frequency generation, are perfectly on phase. But in case of WF lack of gen-
variation becomes more important and harmful which can damage eration, the grid voltage and the current absorbed from the grid at
the electrical equipment and the personnel. But in the last situa- PCC are perfectly opposite. To evaluate the capabilities of studied
tion, the frequency variations are well minimized which proves the WF to compensate harmonic currents, the spectra of currents
effectiveness of this control strategy. before and after connection of WF are shown in Fig. 20. Fig. 20 a
Fig. 15 illustrates the behaviors of the grid and the WF conver- shows that the THD after connection of the variable NL loads to the
sion chain after application of the proposed control strategy. Fig. 16, distribution network is 62.45%. Fig. 20b shows that the THD of grid
Fig. 17 and Fig. 18 are zooms of Fig. 15 under the respective currents after connection of the studied WF is drastically decreased
following cases: In case of WF excess of generation, in case of WF to 2.81% which demonstrates capabilities of the proposed system to
lack of generation and in case of the transition between them. compensate loads harmonic currents with presented control
According to these figures, it is observable that the grid currents are strategy.
well sinusoidal without any disturbance which proves the capa-
bility of the studied WF to ensure its objectives successfully.
5. Conclusion
Fig. 19 shows the performance of the studied WF to compensate
reactive power. In fact, as depicted in this figure, the grid voltage
A multi-objective control strategy for the integration of RDG
and the current injected into the grid at PCC, in case of WF excess of
system to the distribution grid to eliminate the harmonic distor-
tion, to ensure the reactive power compensation and to provide
active power in addition to the grid frequency regulation has been
proposed in this paper. This control has two main structures. The
first one is the “NL Loads currents identification system” which is
used to calculate the reference currents applied for each renewable
source inverter control to ensure harmonic and reactive power
damping. The second one is the “active power transfer and fre-
quency control algorithm” which is implemented to control the
fundamental active power exchange between the RDG and the
electrical power system and to provide the grid frequency service.
The proposed method in different operation modes of the studied
RDG has been tested through simulation results. The obtained re-
sults prove that the electrical network does not need to generate
harmonic and reactive currents for NL loads. Similarly, the utility
frequency variation is well minimized into desired range of sta-
bility. This shows that the studied RDG can be considered simul-
taneously as active power filters, reactive power corrector devices,
active power generator and frequency regulator.

References

[1] Stigka KE, Paravantis AJ, Mihalakakou KG. Social acceptance of renewable
energy sources: a review of contingent valuation applications. Renew Sust
Energy Rev 2014;32:100e6.
Fig. 19. Grid voltage and current states after harmonic and reactive power compen- [2] Lin SY, Chen JF. Distributed optimal power flow for smart grid transmission
sation: In case of WF excess of generation (a) e In case of WF lack of generation (b). system with renewable energy sources. Energy 2013;56:184e92.

Please cite this article in press as: Rekik M, et al., A micro-grid ensuring multi-objective control strategy of a power electrical system for quality
improvement, Energy (2015), http://dx.doi.org/10.1016/j.energy.2015.05.050
 M. Rekik et al. / Energy xxx (2015) 1e13 13

[3] Fazelpour F, Soltani N, Rosen AM. Feasibility of satisfying electrical energy [27] Renewables Academy. ReGrid: frequency and voltage regulation in electrical
needs with hybrid systems for a medium-size hotel on Kish Island, Iran. En- grids.
ergy 2014;73:856e65. [28] Power quality application guide: Standard EN 50160- Voltage characteristics
[4] Sarrias R, Fernandez ML, García AC, Jurado F. Coordinate operation of power in public distribution systems; 2004.
sources in a doubly-fed induction generator wind turbine/battery hybrid [29] Xavier P Nelson Vinotha Kumar, Muthukumar S. Frequency regulation by free
power system. J Power Sources 2012;205:354e66. governor mode of operation in power stations. In: International conference on
[5] Stoppato A, Cavazzini G, Ardizzon G, Rossetti A. A PSO (particle swarm computational intelligence and computing research; 2010. p. 1e5.
optimization)-based model for the optimal management of a small PV [30] Australian Energy Market Operator. Wind integration: International experi-
(Photovoltaic)-pump hydro energy storage in a rural dry area. Energy ence WP2: Review of grid codes. 2011.
2014;76:168e74. [31] Camblong H, Nourdine S, Vechiu I, Tapia G. Control of wind turbines for fa-
[6] Rekik M, Abdelkafi A, Krichen L. A novel control strategy of a distributed tigue loads reduction and contribution to the grid primary frequency regu-
generator operating in seven modes for ancillary services under grid faults. Int lation. Energy 2012;48(1):284e91.
J Electr Power Energy Syst 2013;47:100e8. [32] Hernandez-Aramburo CA, Green TC, Mugniot N. Fuel consumption minimi-
[7] Fuchs E, Masoum MAS. Power quality in power systems and electrical ma- zation of a microgrid. IEEE Trans Ind Appl 2005;41(3):673e81.
chines. Academic Press; 2011.
[8] Arrillaga J, Watson NR. Power system harmonics. John Wiley & Sons, Inc;
2004.
Glossary
[9] Cagnano A, De Tuglie E. Centralized voltage control for distribution networks
with embedded PV systems. Renew Energy 2015;76:173e85. Vm: average wind speed applied to the fifteen WT
[10] Majumder R. Reactive power compensation in single phase operation of MfD: mutual inductance between the main field winding and direct damper
microgrid. IEEE Trans Ind Electr 2013;60(4):1403e16. Vi: wind speed applied to each WT
[11] Urtasun A, Sanchis P, Barricarte D, Marroyo L. Energy management strategy Te: SG electromagnetic torque
for a battery-diesel stand-alone system with distributed PV generation based Pmec: WF mechanical power
on grid frequency modulation. Renew Energy 2014;66:325e36. Tm: SG mechanical torque
[12] Xin H, Liu Y, Wang Z, Gan D, Yang T. A new frequency regulation strategy for Cp: power coefficient
photovoltaic systems without energy storage. IEEE Trans Sust Energy J: mechanical inertia of the SG
2013;4(4):985e93. S: WT swept area
[13] Wang X, Blaabjerg F, Chen Z. Synthesis of variable harmonic impedance in fd,fq: SG d-q stator leakage flux
inverter-interfaced distributed generation unit for harmonic damping r: air density
throughout a distribution network. IEEE Trans Ind Appl 2012;48(4):1407e17. np: SG number of pole pairs
[14] Munir S, Li YW. Residential distribution system harmonic compensation using nT: number of WT units
PV interfacing inverter. IEEE Trans Smart Grid 2013;4(2):1e11. Vsd,Vsq: d-q components of the PMSG stator voltages
[15] Melício R, Mendes VMF, Catala ~o JPS. Comparative study of power converter Vd,Vq: SG d-q output voltages
topologies and control strategies for the harmonic performance of variable- Isd,Isq: d-q components of the PMSG stator currents
speed wind turbine generator systems. Energy 2011;36:520e9. Vf: SG main field excitation voltage
[16] García M, Comech MP, Sallan J, Llombart A. Modelling wind farms for grid p: number of pole pairs of PMSG
disturbance studies. Renew Energy 2008;33:2109e21. id,iq: SG d-q armature currents
[17] Ali M, Ilie IS, Milanovic VJ, Chicco G. Wind farm model aggregation using fm: PMSG stator flux
probabilistic clustering. IEEE Trans Power Syst 2013;28(1):309e16. if: SG main field excitation current
[18] Fernandez LM, García CA, Saenz JR, Jurado F. Equivalent models of wind farms Ls: Stator winding inductance of PMSG
by using aggregated wind turbines and equivalent winds. Energy Conv Manag iD,iQ: SG D-Q dampers currents
2009;50:691e704. INNL: NL loads current
[19] Kopse D, Rudez U, Mihalic R. Applying a wide-area measurement system to rs: SG stator resistance
validate the dynamic model of a part of European power system. Electr Power Ih: harmonic currents amplitudes
Syst Res 2015;119:1e10. ld,lq: SG d-q stator main inductances
[20] Bensmaine F, Barakat A, Tnani S, Champenois G, Mouni E. Dual control of u: fundamental angular frequency
synchronous generator for terminal voltage regulation-comparison with a rD,rQ: SG dampers resistances
single control. Electr Power Syst Res 2012;91:78e86. fh: harmonic phase difference
[21] Barakat A, Tnani S, Champenois G, Mouni E. Analysis of synchronous machine lD,lQ: SG D-Q dampers inductances
modeling for simulation and industrial applications. Simul Model Pract The- PWT_MPPT: MPPT WT generated power
ory 2010;18(9):1382e96. rf: main field resistance
[22] De La Rosa FC. Harmonics and power systems. Taylor & Francis Group; 2006. PWF_MPPT: MPPT WF generated power
[23] Murugesan K, Muthu R, Vijayenthiran S, Mervin JB. Prototype hardware lf: main field inductance
realization of the DSTATCOM for reactive power compensation. Int J Electr Rf: filter resistance
Power Energy Syst 2015;65:169e78. uep: SG electrical angular speed
[24] Savaghebi M, Vasquez JC, Jalilian A, Guerrero JM, Lee TL. Selective compen- Lf: filter inductance
sation of voltage harmonics in grid connected microgrids. Mathe Comput MsD: mutual inductance between stator and direct damper
Simul 2013;91:211e28. PWF: WF generated power according to the selective modes (MPPT or without MPPT)
[25] Bilgin HF, Ermis M. Current source converter based STATCOM: operating MsQ: mutual inductance between the stator and quadrature damper
principles, design and field performance. Electr Power Syst Res 2011;8: PNLL_f: NL loads fundamental active power
478e87. Msf: mutual inductance between direct stator winding and main field
0
[26] Li C, Zhan P, Wen J, Yao M, Li N, Lee WJ. Offshore wind farm integration and PNLL f : NL loads fundamental active power after disconnecting the minor priority NL
frequency support control utilizing hybrid multiterminal HVDC transmission. loads
IEEE Trans Ind Appl 2014;50(4):2788e97.

Please cite this article in press as: Rekik M, et al., A micro-grid ensuring multi-objective control strategy of a power electrical system for quality
improvement, Energy (2015), http://dx.doi.org/10.1016/j.energy.2015.05.050