Electric Power Systems Research 143 (2017) 503–512

Contents lists available at ScienceDirect

Electric Power Systems Research

journal homepage: www.elsevier.com/locate/epsr

Design of a command scheme for grid connected PV systems using

classical controllers
Nejib Hamrouni a,∗ , Moncef Jraidi b , Ahmed Dhouib c , Adnen Cherif d
a
National Engineering School of Gabes, University of Gabes, Gabes, Tunisia
b
National Engineering School of Carthage, University of Carthage, Tunis, Tunisia
c
National Engineering School of Tunis, University El Manar, Tunis, Tunisia
d
Faculty of Sciences, University El Manar, Tunis, Tunisia

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

Article history: This paper proposes an approach of a grid connected PV control scheme, that provides optimal PV power
Received 30 November 2014 and high quality of current injected into the grid and, therefore, high power quality. The control scheme
Received in revised form 19 July 2016 combines current control and power control loops. To analyze the dynamic behavior and the design of
Accepted 29 October 2016
the control strategy, a conventional two-stage energy conversion is selected and briefly modeled. The
Available online 17 November 2016
command developed around the first stage will raise the relatively low solar voltage to a suitable level
corresponds to the optimal photovoltaic power. Around the second stage a voltage and current loops
Keywords:
have been designed to regulate respectively the dc link voltage and the output inverter current. The
PV generator
Inverter
aim of this command is to inject energy into the grid at unity power factor and a current with low
PLL harmonic distortion. All command loops of the grid connected photovoltaic system is based on classical
Grid controllers which are composed of proportional-integral (PI) regulators. The choice of control variables
MPPT and the design of command loops are discussed. The proposed global model is simulated with the help
PI controllers of Matlab–Simulink, by considering a 3 kWc solar generator.
Current controller © 2016 Published by Elsevier B.V.

1. Introduction grid. Different output filter topologies are presented in literature,

namely the LCL, LC and L filter [3,4].
Recently, there is an increasing interest in renewable energy The LCL filter is currently the most frequency used topology
sources for stand alone or grid connected systems [1]. Photovoltaic [5,6]. This filter offers a good attenuation with a relatively small
solar power is one of the most promising renewable energy sources component value and a good power quality. However, it can make
in the world among of its advantages, reliability, and availability. instability of the system caused by the zero impedance that the LCL-
The solar energy has been used as the main sources for electri- filter offers at its resonance frequency [7]. Moreover, it increases
cal loads in rural areas away from the grid or grid connected sensor numbers and command complexity [8]. Typically, simple
in distributed energy production. The connection of the photo- series inductors are used as the filter interface between the inverter
voltaic generator (PVG) to the electrical network, which was not and the grid network [9].
conceived to accommodate them, constitutes a technological rise. In order to capture the optimal PV power, several maximum
This connection constitutes a solution to overcome the problem of power point tracking (MPPT) algorithms are designed around the
exhaustion of fossil energies and the increase of power demand in dc–dc converter [10–12]. These algorithms differed by their com-
the world [2]. plexities and performances but they required the measure of the
The number of stage involved in grid connected PV systems is an PV current and voltage. The principal algorithms are perturbation
important issue, as it determines the reliability and control com- and observation (P&O), incremental inductance, constant voltage
plexity in grid connected PV systems. These systems typically have and fuzzy logic algorithms [13]. In this work a P&O algorithm has
two or more stages and filter which is necessary to reduce harmon- been chosen because of its simplicity and less costly implemen-
ics injected by the inverter and to neutralize spikes coming from the tation. Two loops in cascade with the MPPT algorithm have been
added. They permit the control of the current and voltage injected
by the PVG.
In the grid side, different control schemes are developed in the
∗ Corresponding author.
literature [14–17]. The first control scheme is the ac current con-
E-mail address: hamrouni [email protected] (N. Hamrouni).

http://dx.doi.org/10.1016/j.epsr.2016.10.064
0378-7796/© 2016 Published by Elsevier B.V.
504 N. Hamrouni et al. / Electric Power Systems Research 143 (2017) 503–512

Lpv DC link 2.1. Modeling of the photovoltaic source

Cdc PV Low
PV Cpv Boost Inverter Filter Voltage
Generator Grid The mathematical model of PVG is deduced from the mono-
diode model of the PV cell. It is given by Eq. (1) [32]:

      
q 
MPPT command Inverter command
Vpv Ipv Rs Np Vpv Ipv Rs
Ipv = Np Iph − Np Is exp + −1 − + (1)
nkT Ns Np Rp Ns Np
Measures References Measures

Fig. 1. Scheme of the grid connected PV system. The model inputs are the solar irradiance, the ambient temper-
ature, the PV voltage and the wind speed, while the only output is
the PV current which is supplied by the panel.
trol strategy. It is usually adopted because it exhibits better safety,
stability and fast response. It is composed of two-cascaded loops;
an inner current and an outer voltage loops which regulate respec- 2.2. Modeling of the dc–dc converter
tively the output inverter current and the dc-link voltage [18–21].
The output of the dc-link controller sets the reference for the The coupling of the load at constant power to the PV gen-
active current. However, active and reactive power controller pro- erator without the use of a battery does not ensure an optimal
vided the reactive current to be injected into the grid [22]. In case of operation. This is due to the instantaneous fluctuation of the opti-
rotating dq-frame, the reference current d-component is controlled mal power with the variation of the climatic conditions (sunning,
to manage the active power and the reference current q-component temperature. . .). It is then essential to envisage a dc–dc converter
is controlled to manage the reactive power. To obtain the line cur- which ensures an extraction of the optimal PV power and to main-
rent in phase with the grid voltage, reference current q-component tain the dc-link voltage as constant. The boost reveals two power
should be zero. The second control scheme studied in the literature accumulation elements and thus two variables to control [11]. They
use a power controller instead of dc-link voltage controller. The are the PV voltage (Vpv ) and the inductor current (il ). They are given
references for the active and reactive current are calculated from by Eq. (2):
the active and reactive power references [23–25]. A third command
⎧
scheme which does not needs internal current loops because the ⎪ dvpv 1  
inverter switching states are appropriately selected by a switch-
⎪
⎨ = ipv − il
dt Cpv
ing table. This command became the most interesting technique (2)
⎪
⎪ di 1   Rpv
in the last years [20]. All previous studies demonstrate that these ⎩ l = vm − vpv − il
various control strategies are well performing under normal grid dt Lpv Lpv
conditions.
In our approach, a combination between the first and the sec- The internal structure of the dc–dc converter is appeared as a
ond inverter control scheme is developed. The reference current matrix function [33]. It is described by the following relation (3):
components have been calculated from reference active and reac- 
tive powers. To have the grid current vector in phase with the grid f11 f12
voltage (i.e., high power factor), reference reactive power should [f ] = (3)
be zero. The active power control is achieved through the con- f21 f22
trol of the dc-link voltage. The reference current components in
the dq-frame are dc quantities. Thus, the command of the overall
- f12 and f22 are equal respectively to 0 and 1.
system should be made in synchronous reference dq-frame using
proportional-integral (PI) regulators [26–30]. For grid connected
PV system, PI controllers have a very good harmonic rejection for Variables modulated by the boost, which are the voltage vm and
low order harmonics and ensure good performance characteristics the current idc , can be expressed according to [33], by Eq. (4):
such as stability and zero state error [31]. Moreover, those con-
trollers are used due to their characteristics of better regulating dc
 
vm vdc
signals and simple design procedure. = mh (4)
In this paper, a conventional two-stage energy conversion sys- idc il
tem is studied. In the second part, elementary and global models
of the grid connected PV system are briefly developed. The third
part discusses the command scheme designed around converters. - mh is equal to the connection function f11 . It has a binary value
Linear control approach, using PI regulators, are detailed. Simula- {0,1}.
tion results are discussed in Section 4. Finally, the conclusion is
presented in Section 5. 2.3. Modeling of the dc link bus

2. Modeling of the grid connected PV system The dc link voltage is governed by Eq. (5):

The grid connected PV system is composed of a PV generator 1

vdc = (i − ie ) (5)
(3 kWp), an MPPT power adapter, a dc-link voltage capacitor, a Cdc s dc
three-phase inverter (3 kVA) and a simple filter. All these compo-
nents are connected to the LV network. The configuration of the Generally, the dc link voltage oscillates between two levels
connected system is presented by Fig. 1. depending on the operating climatic conditions; ambient temper-
This structure, allowing an energy exchange between PV plant ature and solar radiation. The choice of storage system parameters
and grid, ensures uninterrupted service and decreases the energy influences the dynamic of the filter placed between the inverter
extracted from the grid. and the utility network.
 N. Hamrouni et al. / Electric Power Systems Research 143 (2017) 503–512 505

Lr, Rr dc-link Inverter L-Filter Grid

ir1 er1
~ ie
v1 L, R
ir1
Lr, Rr er1
vr1 u1 Lr, Rr vrl1 ~
ir2 er2 vr1
~ vdc
L, R ir2
Lr, Rr
er2
~
vr2 u2 vrl2
ir3 Lr, Rr vr2
er3 L, R ir3 Lr, Rr er3
~ vrl3 vr3
~
f11* f12* f13*
vr3
Modulator
Fig. 2. Equivalent diagram of the LV network.
m1g* m2g*

2.4. Modeling of the inverter i1j, vdc, vrj

θrest
Current loop
According to [33], the inverter is a double modulator of electric i2q*
i2d*
quantities. The modulated voltage and current vector are given by
Eq. (6): vrj, irj / j={1,2,3}
Voltage loop PLL
⎧  
⎪
⎪   ir1
⎪
⎪
⎨ [ie ] = m1 m2 . i Vdc* Vdc Qr idc vrj
  r2 (6)
⎪
⎪ u1 m1 Fig. 3. Diagram of dc–ac converter command.
⎪
⎪ = .vdc
⎩ u m2 2
first order filter. The choice of the filter depends on a compromise
where ie is given by (7): between the dynamic and the effectiveness of harmonic attenua-
tion of the current injected by the inverter. The current injected
Pr
ie ≈ (7) into the grid can be expressed in the synchronous dq- frame by Eq.
vdc
(10):
Pr is the active power provided by the PVG when we neglect    
the switching losses in the inverter, while u1 and u2 are the com-
 d  ird vrld irq
L +R = + Lw (10)
posed inverter output voltage. m1 and m2 are obtained from the dt irq vrlq −ird
subtraction of two connection functions having a binary value {0,1}.
Their fields of definition are {−1, 0, 1}. They are described by this where filter voltage d-component and q-component are given by
following relation (8): Eq. (11):
 
m1 = f11 − f13 vrld = v1d − vrd
(8) (11)
m2 = f12 − f13 vrlq = v1q − vrq

2.5. Model of the LV grid 3. Control of the grid connected PV system

To adapt the network model to the elementary models devel- The main objective of the control is to convert dc power from the
oped previously and to facilitate the exchange with the industrial PVG to ac power feeding to the grid. The proposed control scheme
environment, we will approach the model used in Power System is composed of two parts. The first block is designed around the
Block set (PSB) of the MATLAB/SIMULINK. Fig. 2 presents the equiv- boost. Its goal is to track the PV’s maximum power point (MPP). The
alent scheme of the network chosen in this paper. This model keeps second block is calculated around the dc–ac converter. Its objective
the natural approach which considers the network as a voltage is to obtain unity power factor and low harmonic distortion at the
source. The part formed by the energy transport line and all the con- inverter output current. This controller consists of an inner current
nection transformers between the various voltage levels is modeled loop and an outer voltage loop, as indicated in Fig. 3.
by a Thevenin equivalent circuit [11]. It is composed of a sinusoidal
voltage source in series with impedance. 3.1. Control of the dc–dc converter
According to Fig. 2, the line current injected into the grid
expressed in the synchronous dq-frame, is given by Eq. (9): In the present paper a MPPT algorithm based on P&O is devel-
    
 d
 ird vrd erd irq oped. Its objective is to find the adequate voltage Vpv * imposed by
Lr + Rr = − + Lr w (9) the dc–dc converter to the system. If the operating voltage of the
dt irq vrq erq −ird PVG is perturbed in a given direction and if the power drawn from
the generator increases, one can infer that the operating point has
2.6. Modeling of the filter moved toward the MPPT and therefore the operating voltage must
be further perturb in the same direction. If the power extracted
The line filter reduces the high frequency harmonics generated from the PVG decrease, the operating point has then moved from
by the switched operation of the inverter. Usually, the line fil- the maximum power point, and therefore, the direction of the oper-
ter consists of inductors but other combinations of capacitors and ating voltage perturbation must be revised. The reference voltage
inductors such us LC or LCL filters can be used. The L-filter is a Vpv *, given by the MPPT algorithm, constitutes the input of the volt-
506 N. Hamrouni et al. / Electric Power Systems Research 143 (2017) 503–512

ipv il Lpv,Rpv idc

   
vpv vm
D
Inverter X= , U1 =, U2 = ipv
il idc
&
ki L-filter By supposing vm appreciably equal to the reference voltage vm *,
vm
PV vpv & the control vector becomes Eq. (16):
Generator vdc
Cpv LV grid
U1 = Kh X + Lv (16)
     
0 mh 0 0 ipv
With: Kh = L= v=
ipv PWM 1 − Rc Rv −Rc Rc Rv Rc vpv ∗
.. vdc
The substitution of Eq. (16) in Eq. (15) gives Eq. (17):
Sensor of Ta .
vm*
(°C) and Φ
vpv X = (A + B1 Kh )X + B1 Lv + B2 U2 (17)
(W/m²) ++ Rc
With:
⎛ 1
⎞
vpv*
MPPT (P&Ob)
-
Rv + + ipv 0 −
+ - ⎜ Cpv ⎟
Ac = A + B1 Kh = ⎝ ⎠
Rc Rv (Rc + Rpv )
il − −
Lpv Lpv

Fig. 4. Diagram of dc–dc converter command. The choice of matrix coefficients Kh allows the design of the
controller gains Rc and Rv.

age and current controllers. The addition of two loops in cascade

3.2. Control of the PV inverter
with the MPPT algorithm allows the control of the energy in the
PV output filter (Lpv Cpv ). The command scheme of the boost is
The system to be controlled is a three phase inverter connecting
developed as illustrated in Fig. 4.
the PVG to the grid. The main goal of the control is to transfer all
The inversion of the previous equations representing the PV
PV power produced to the grid, to regulate the dc-link voltage and
voltage and current variations, respectively, in the capacitor Cpv
to set a unity power factor for grid currents. The proposed control
and the inductance Lpv , gives current and voltage optimal control.
scheme is two loops based. The outer loop is the dc-link voltage
The voltage control loop with the PV current compensation pro-
and the inner one is around inductor currents.
vides the reference current Il *, while the current control loop with
In grid connected control mode, all the available power that can
the PV voltage compensation gives the reference voltage Vm * for
be extracted from the PVG is transferred to the grid. The compen-
PWM, as indicated in Eq. (12). The controller parameters are cho-
sation of the reactive energy required by the local load is ensured
sen to maintain constant the PV voltage and to reduce the current
by the electrical network. The general structure of control, relat-
ripple.
ing to the connected mode, is shown by Fig. 5. The inverter control
  
il ∗ = Rv vpv ∗ −vpv + ipv is deduced by inverting the process model of the dc-link capacitor,
(12) the three phase inverter, the L-filter and the LV grid network. Under
vm ∗ = Rc (il ∗ −il ) + vpv these conditions, three elementary blocks of command around the
inverter appear; the dc link voltage controller, the current loop and
The dc–dc converter command signal is given, according to [34], the phase locked loop (PLL).
by Eq. (13):
vm ∗ 3.2.1. Voltage loop
mh ∗ = (13)
vdc This loop is composed of two modules. They ensure the control
of dc-link voltage and the calculation of the reference currents. The
3.1.1. Design of the current and voltage control loops first module supply a dc current allowing the calculation of the
The main objective of the boost command is to find the reference reference active power Pr *. The second module is devoted to the
voltage vm * which must be imposed to the dc–dc converter. This calculation of reference currents ird * and irq * which are used in the
voltage is obtained from Eq. (12). It is given by Eq. (14): current loops.

vm ∗ = Rc Rv vpv ∗ +(1 − Rc Rv )vpv + Rc ipv − Rc il (14) 3.2.1.1. Dc-link voltage control. The main of the voltage control loop
is to regulate the dc-link voltage at a specified value and to provide
RC and Rv represent respectively the PV current and the voltage
the reference current.
regulators. When the boost switch frequency is higher, the time
The dc-link voltage is set to 550 V while the line voltage is set
constant is very small and its influence to the whole system may be
to 220 V. The closed loop control of the dc-link voltage is needed
ignored, so proportional correctors are sufficient. They are parame-
because the output power of the PV is variable with irradiance
terized according to the value of the capacitor (Cpv ), the inductance
and temperature. The use of the informational graph of causality
(Lpv ) and the dynamic of the regulation loops.
(IGC) as well equations developed in Section 2 permit the design
Eq. (2), representing the model of the PV and boost association,
of the control loop [32]. The inverter input reference current ie * is
can be as follows Eq. (15):
described by the following relation (18) [34]:
.
X = AX + B1 U1 + B2 U2 (15)
ie ∗ = idc − PI(vdc ∗ −vdc ) (18)
With:
⎛ ⎞
1   To control the dc-link voltage, a proportional-integral corrector
0 − 0 1 has been used to reduce the steady state error of the dc bus voltage.
A=⎝ ⎠ , B1 =
Cpv
Rpv
1 , B2 = Cpv It is parameterized according to the capacitor and the dynamic of
1
− − Lpv 0 the regulation loop.
Lpv Lpv
 N. Hamrouni et al. / Electric Power Systems Research 143 (2017) 503–512 507

Fig. 5. Block diagram of the PV inverter command.

Fig. 6. Structure of the dc link voltage controller.

3.2.1.2. Design of the dc-link voltage regulator. The dc-link voltage characteristic polynomial, we deduce parameters of the dc-link
regulator consists of a PI controller which is adopted to avoid sta- voltage controller. They are given by Eq. (20):
tionary dc bus voltage. The low pass filter in the feedback loop limits ⎧
the bandwidth of the dc-link voltage regulator. It implies that the ⎪
⎨ Kdc =
2fc Cdc
3
dynamic of the line current regulator can be neglected in the dc-link (20)
voltage controller design. Therefore, the voltage and the current ⎪
⎩ dc = 9
loops can be considered decoupled, and there by the actual grid 2fc
current components can be considered equal to their references
when designing the dc-link voltage controller. Thus, we can cal- 3.2.1.3. Calculator of reference currents. Reference currents ird * and
culate, from Fig. 6, the closed loop transfer function of the dc-link irq * are calculated from the grid voltage and the reference active and
voltage regulator. It is given by Eq. (19): reactive powers. These currents, according to [36], are expressed in
the synchronous reference dq-frame by Eq. (21):
    
ird ∗ 1 Pr ∗ −Qr ∗ vrd
1
= (21)
 dc C dc
Kdc [sdc + 1] [s + 2fc ] irq ∗ v2rd + v2rq Qr ∗ Pr ∗ vrq
H(s) = (19)
2fc Kdc 2fc Kdc
s3 + 2f c s2 + s+
Cdc dc Cdc Qr * is the reference reactive power. To obtain the line current in
phase with the grid voltage, Qr * should be zero. Pr * is the reference
active power injected into the grid. It is given by Eq. (22):

fc is the cut frequency of the low pass filter F(s). It is set to 30 Hz Pr ∗ ≈ ie ∗ .vdc (22)
[35].
The characteristic equation of the closed loop system is a third 3.2.2. Synchronization
order having three poles. To avoid the oscillation in the dc-link According to Eq. (21), reference currents injected into the grid,
voltage, we supposed that the characteristic equation has three are deduced from the measurement of grid voltages. Indeed, the
identical poles. By identification of the coefficients of the charac- network voltages at the connection point can contain various faults
teristic equation given by the denominator of H(s) with the desired (voltage dips, harmonics, short interruptions. . .) which can pass
508 N. Hamrouni et al. / Electric Power Systems Research 143 (2017) 503–512

Fig. 7. Structure of linear PLL model.

Fig. 8. Block diagram of the direct grid current controller.

into the current provided by the inverter. To mitigate this problem,

several methods are developed in the literature [32]. Among those
methods we are interested in phase locked loop (PLL) expressed in
dq-frame. This method is able to provide the grid frequency and
phase information. The output phase provided by the PLL will be
used as an input for the PI controller in order to have the adaptive
tuning in respect to the frequency grid.
The regulator should be designed to respond with minimum of
overshoot to the grid frequency and voltage variations. When the
difference between grid phase angle ( r ) and inverter angle ( ond ) is
reduced to zero,
√ the PLL became active [32] and we obtain; vrd = 0
and vrq = − 3Vr . The model of the PLL is strongly non-linear. How-
ever, to synthesize the PLL corrector, we considered a linear model
for weak variations of  r . According to Fig. 7, the closed loop transfer
function of the linear PLL model is equivalent to a transfer function
of second order system. The dynamic of this loop is obtained accord-
ing to the desired response time Tr (i.e., Tr = 10 ms). To avoid any
oscillation type of the PLL response, the damping factor should be
equal to 1 ( = 1). This choice of Tr and  permits consequently the
deducing of the various regulator parameters k and   .

3.2.3. Current loop

The ac current control is considered particularly suitable for
active inverters for its safety and stability performance. The cur-
rent control is implemented in a synchronously rotating reference
dq-frame. Thus, individual controllers for the active and reactive
Fig. 9. Simulation results under various climatic conditions: (a): maximal power
current components are developed. The reference currents ird * supplied by the PVG, (b): PV current (reference and measure), (c): PV voltage (ref-
and irq * are controlled to perform, respectively, the reference volt- erence and measure).
age components vrld * and vrlq *. Additional compensation for the
cross-coupling of the d and q-terms are adopted to generate the grid connected current is given by Fig. 8. To simplify the synthesis of
reference voltages v1d * and v1q * for PWM. To improve the response the corrector Cid (s) and Ciq (s), we suppose the following assumption
time and to eliminate the steady state errors of the fundamental (24):
current components, two classic PI controllers are employed as  
indicated by Fig. 8. The reference voltage components v1d * and v1q * vrld ∗ vrld
are expressed in the synchronous reference dq-frame as follows ≈ (24)
vrlq ∗ vrlq
(23):
 The controller parameters of Cid (s) and Ciq (s) can be obtained
v1d ∗ = vrd − Lωrest irq + Cid (s)(ird ∗ −ird )
(23) by the pole compensation method. Solved from Eq. (10) through
v1q ∗ = vrq + Lωrest ird + Ciq (s)(irq ∗ −irq ) Laplace conversion, the filter voltage dq-components are given by
Eq. (25):
The pulsation wrest is provided by the PLL.  
vrld 1 ird
= (25)
3.2.3.1. Design of the current regulator. The current loops should vrlq Ls + R irq
drive the output currents ird and irq to the desired steady state cur-
rents, namely to the shape and phase of the utility grid voltage by PI controllers are used to control the inverter current dq-
generating control signals m1 * and m2 *. The design of current con- components, since they have acceptable performances while
trollers depends of the regulation objectives and filter impedance. regulating the dc variables. The average value of current error can
According to Eq. (23), the closed loop control scheme of the direct be reduced to zero by changing the integral components. Moreover,
 N. Hamrouni et al. / Electric Power Systems Research 143 (2017) 503–512 509

Fig. 10. Characteristics of the input and output inverter voltages (a): dc link voltage
(measure and reference), (b): output inverter voltage (measure and reference).
Fig. 12. Reactive and active powers injected into the grid (reference and measure).

the transient response of PI controller is limited by the proportional

gain. According to the property of the IGC, command signals of the
By using the PI controller and taking account of simplifications three phase inverter are given by Eq. (28):
made in Eq. (24), the open loop transfer function (FTBO ) becomes
   
m1g ∗ 1 2 1 v1 ∗
Eq. (26): = . (28)
m2g ∗ vdc 1 2 v2 ∗
1
Ts + 1 R
FTBO (s) = k L
(26) where reference voltages, expressed in the phase reference abc-
Ts s+1
R
frame, are given by Eq. (29):
When we select a time constant T equal to RL , the closed loop ⎛ ⎞
v1 ∗ 
transfer function (FTBF ) of current controller is given by Eq. (27): ⎜ ⎟ v1d ∗
⎝ v2 ∗ ⎠ = P −1 3/2 (rest ) (29)
FTBF (s) =
1
(27)
v1q ∗
TR
s+1 v3 ∗
k

We calculate k in order to decrease the response time of the 4. Simulation

closed loop compared to the open loop transfer function. The choice
of k is such as the cut-off frequency in closed loop is one decade The most important waveforms simulated are shown in
higher than the cut-off frequency in open loop. In this case k is Figs. 9–13. These simulations have been obtained under various
equal to R and  is a real higher than 10. values of irradiance and constant temperature (25 ◦ C). They are

Fig. 11. Inverter output current (-*40) and grid voltage under various illuminations.
510 N. Hamrouni et al. / Electric Power Systems Research 143 (2017) 503–512

Table 1
Principal parameters of the PI regulators and the grid connected PV system.

Parameter Value Parameter Value

Cell series resistance Rs 0.011  Output PV filter resistance Rpv 1

Cell parallel resistance Rp 10.35  Output PV filter capacitance Cpv 0.005 F
Ideal constant of diode n 1.25 DC link capacitance Cdc 0.009 F
Cell reverse saturation current Is 13.89 × 103 T3 exp(−13.2 × 103 /T) A Filter inductance L 0.02 H
Photo current of PVG Iph [2.807 + 1.4 × 10−3 (T − Tréf )]Ø A Filter resistance R 0.5 
Cell reference temperature Tref 298.15 K Grid inductance Lr 0.005 H
Series cell ns 40 Grid resistance Rr 1
Parallel cell np 1 Gain of PV voltage regulator Rv 10
Series modules Ns 20 Gain of PV current regulator Rc 34
Parallel modules Np 3 Gain of dc link voltage regulator kdc 0.001
Short circuit current of PVG Isc 8,37 A Const. time of dc link voltage regulator ␶dc 0.1
Open circuit voltage of PVG Vpv0 428 V Gain of PLL regulator K 10
Maximal PV current Ipvmax 7.6 A Constant time of PLL regulator ␶ 0.1
Maximal PV voltage Vpvmax 360 V Gain of current regulator k 1
Output PV filter inductance Lpv 24.9e–4 H Constant time of current regulator T 0.05

seen that, the P&O program developed around the dc–dc converter
presents a good dynamic performance. During each change in solar
radiation, the maximal PV power was established in delay time
equal to 0.5 s. This disadvantage characterizes limits of the P&O pro-
gram. In steady state and for each solar radiation, the PVG provides
the maximal power. Fig. 9b and c show respectively the current and
voltage variations of the PVG.
Fig. 9b presents the variation of the measure and reference of the
dc–dc converter input current. It is maintained, in steady state, at a
constant and maximal level in each case of solar irradiance (1000,
500, 800 and 300 W/m2 ). The error in steady state has been caused
by the approximation made during the design of the PV current
controller.
According to Fig. 9c, the measure of the capacitor voltage follows
the reference voltage imposed by the current regulator.
Fig. 10a presents the dc side voltage. It can be seen that the dc
measured voltage follows instantaneously its deadbeat reference
voltage Vdc *. An important steady state error is noticed. The dc-
link voltage oscillates between 530 V and 542 V according to the
solar radiation. The PI controller encounters problems and limits in
the voltage regulation. Fig. 10b represents inverter output voltage
before filtering and the reference voltage provided by the inverter
control system.
Fig. 11 shows the ac side voltage and the inverter output current.
As it might be noticed in this figure, there is a phase leading/lagging
between grid side line current and voltage. A low displacement
can be noticed between the synthesized current, current effectively
injected into the grid, and the line voltage. A current THD, lower
than 5%, around the rated inverter output has been achieved. As a
result, a unit power factor is achieved.
Fig. 12a and b present respectively the reactive and active power
injected into the grid. The reactive power oscillates around its ref-
erence (Q* = 0VAR). It shows the unity power factor operation and
the control performance of the grid current. The measured active
Fig. 13. Simulation results under various climatic conditions: (a) inverter current
power supplied by the inverter oscillates around its maximal ref-
d-component (measure and reference), (b) inverter current q-component (measure
and reference). erence. It can be seen that the power output of the inverter follows
its reference and that average power control has been established.
The measure current dq-components as depicted in Fig. 13a and
selected to demonstrate the most significant aspects of system
b follow their references provided by the current control. These
behavior and the performance of the designed control loops. The
figures showed the control performance of the grid current.
reference and measure are respectively represented by dotted and
As can be noticed from Figs. 13 and 14, the current injected into
continuous line. Salient parameters of the system and regulators
the grid contains high and low frequency harmonics. They are due
are presented in Table 1. The PV generator is a polycrystalline
to the inverter switching, the design and the choice of the filter and
24V/45Wc-AEGPQ40D.
finally the deficiency of the inverter current controller. In order to
Fig. 9a shows characteristics of the PV power provided during
reduce the low order harmonics, the L-filter must be replaced by
various step transients in the solar radiation. The irradiance fluctu-
other filters such us LC or LCL filter.
ates at 2.1 s, 3 s and 4 s. It changes from 1000 to 500 W/m2 , and then
it increases to 800 W/m2 and finishes at 300 W/m2 . It can be clearly,
 N. Hamrouni et al. / Electric Power Systems Research 143 (2017) 503–512 511

Fig. 14. Line current under various climatic conditions (measure and reference).

5. Conclusion [7] T.C.Y. Wang, Z. Ye, G. Sinha, X. Yuan, Output filter design for a grid
interconnected three phase inverter, Power Electronics Specialist Conference
2 (2003) 779–784.
The goal of this paper is to design a control scheme for the [8] M. Liserre, A. De, ll’Aquila, F. Blaabjerg, Stability improvements of an LCL filter
grid connected PV system and to study the performance and the based three phase active rectifier, Power Electronics Specialist Conference 3
dynamic of this command using linear controllers. The command (2002) 1195–1201.
[9] T. Erika, G.H. Donald, Grid current regulation of a three phase voltage source
designed around the dc–dc converter is composed of a conven- inverter with an LCL input filter, IEEE Trans. Power Electron. 18 (3) (2003)
tional MPPT in series with current and voltage controllers. Its main 1189–1194.
objective is to extract the optimal PV power which varies with the [10] A. Saadi, A. Moussi, Etude de l’effet des fluctuations de température sur les
techniques d’optimisation des systèmes de pompage photovoltaïque, CEE
climatic conditions. The current control scheme designed around
(2002) 232–237.
the inverter permits an alternative current and synchronization [11] N. Hamrouni, M. Jraidi, A. Chérif, New control strategy for 2-stage
with the grid using a PLL loop. Furthermore, it permits the con- grid-connected photovoltaic system, Renew. Energy 33 (2008) 2212–2221.
[12] N.S. D’souza, L.A.C. Lopes, X. Liu, Comparative study of variable size
trol of the dc-link voltage in order to transfer the PV power to the
perturbation and observation maximum power point trackers for PV systems,
utility network. Electr. Power Syst. Res. 80 (2010) 296–305.
It has been demonstrated that the proposed control scheme, [13] S. Zheng, P. Wang, L. Ge, Study on improving output current waveform of
based on classical controllers, presented a good performances in photovoltaic grid connected system, Power Electronics and Motion Control
Conference 3 (2006) 1–5.
variable climatic conditions. The variation of the climatic condi- [14] R. Caldon, F. Rossetto, R. Turri, Analysis of dynamic performance of dispersed
tion does not have an influence on the dc-link voltage. The dc-link generation connected through inverter distribution networks, in: Proceedings
capacitor acts as a buffer zone which isolates the PVG from the grid. of the CIRED, Spain, 2003.
[15] S. Sishuba, M.A. Redfern, An adaptative control strategy for parallel and
Moreover, simulation results showed that the control scheme per- isolated operation of dispersed generators, in: Proceedings of UPEC, Greece,
mits to provide the optimal PV power and to inject the energy into 2003.
the grid at unity power factor and a current with low harmonic dis- [16] F. Ruz, A. Rey, J.M. Torrelo, A. Nieto, F.J. Cánovas, Real time test benchmark
design for photovoltaic grid-connected control systems, Electr. Power Syst.
tortion. It has been concluded that, during various step transients Res. 81 (4) (2011) 907–914.
in solar radiation, stability and high dynamic of the grid connected [17] G. Tsengenes, G. Adamidis, Investigation of the behavior of a three phase
PV system are obtained. In spite of those advantages, simulation grid-connected photovoltaic system to control active and reactive power,
Electr. Power Syst. Res. 81 (1) (2011) 177–184.
showed that the current injected into the grid contains some high
[18] F. Atunes, A.M. Torres, A three phase grid connected PV system, Industrial
frequency harmonics. Electronics Society, 26th Annual Conference of the IEEE 1 (2000) 723–728.
As a consequence, it is required to use other filter topologies [19] M. Carlos, B. Domingo, M. Juan, G. Francesc, Boost-buck inverter variable
structure control for grid connected photovoltaic systems, IEEE International
such us LC or LCL filters ensuring the elimination of the switching
Symposium on Circuits and Systems 2 (2005) 1318–1321.
frequency emitted in the grid. The future work will extend the anal- [20] M. Kazmierkowski, R. Krishnan, F. Blaabjerg, Control in Power
ysis to investigate the performance of the control scheme in faulty Electronics—Selected Problems, Academic Press, 2002.
grid operation mode. [21] Q. Zeng, L. Chang, P. Song, SVPWM-based current controller with grid
harmonic compensation for three-phase grid-connected VSI, Proceedings of
PESC04 4 (2004) 3494–3500.
[22] F. Blaabierg, R. Teodorescu, Z. Chen, M. Liserre, Power converters and control
of renewable energy, Proceedings of ICPE’04 (2004) 2–20.
References [23] M. Prodanovic, T.C. Green, Control of power quality in inverter based
distributed generation, Proceedings of IECON’02 2 (2002) 1185–1189.
[1] A. Salvador, B.M. Sergio, B. Josef, G. Javier, G. David, Interfacing renewable [24] D. Candusso, L. Valero, A. Walter, Modeling, control and simulation of a fuel
energy sources to the utility grid using a three level inverter, IEEE Trans. Ind. cell based power supply system with energy management, Proceeding of
Electron. 53 (5) (2006) 1504–1511. IECON’02 2 (2002) 1294–1299.
[2] V.T. Adrian, T. Remus, F. Belaabjerg, M. Liserre, D. Antonio, Independent [25] M. Marei, E. El-Saadany, M. Salama, A novel control algorithm for the DG
synchronization and control of three phase grid converters, International interface to mitigate power quality problems, IEEE Trans. Power Deliv. 19 (3)
Symposium on Power Electronics, Electrical Drives, Automation and (2004) 1384–1392.
Motion-ISPEEDAM (2006) 1246–1251. [26] A. Mohamed, Z. Zhengming, Grid-connected photovoltaic power systems:
[3] P. Milan, C.G. Timothy, Control and filter design of three-phase inverters for technical and potential problems, Renew. Sustain. Energy Rev. 14 (2010)
high power quality grid connection, IEEE Trans. Ind. Electron. 18 (1) (2003) 112–129.
373–380. [27] D.N. Zmood, D.G. Holmes, Stationary frame current regulation of PWM
[4] M. Liserre, A. Dell’Aquila, F. Belaabjerg, Generic algorithm based design of the inverters with zero steady-state error, IEEE Trans. Power Electron. 18 (3)
active damping for an LCL-filter three phase active rectifier, IEEE Trans. Ind. (2003) 814–822.
Electron. 19 (2004) 76–86. [28] M. Liserre, A. Dell’Aquila, F. Blaabjerg, Genetic algorithm-based design of the
[5] M. Liserre, F. Blaabjerg, S. Hansen, Design and control of an LCL filter based active damping for an LCLfilter three-phase active rectifier, IEEE Trans. Power
three phase active rectifier, Industry Applications Conference, Conference Electron. 19 (1) (2004) 76–86.
Record of the IEEE 1 (2001) 299–307. [29] E. Twining, D.G. Holmes, Grid current regulation of a three-phase voltage
[6] F. Blaabjerg, E. Chiarontoni, A. Dell’Aquila, M. Liserre, S. Vergura, Analysis of source inverter with an LCL input filter, IEEE Trans. Power Electron. 18 (3)
the grid side behavior of an LCL filter based three phase active rectifier, IEEE (2003) 888–895.
Ind. Electron. 2 (2003) 775–780.
512 N. Hamrouni et al. / Electric Power Systems Research 143 (2017) 503–512

[30] B. Bolsens, K. De Brabandere, J. Van den Keybus, J. Driesen, R. Belmans, l’insertion de production d’énergie dispersée dans les réseaux électriques,
Model-based generation of low distortion currents in grid-coupled Habilitation à diriger des recherches, UST-Lille, 2003.
PWM-inverters using an LCL output filter, IEEE Trans. Power Electron. 21 (4) [34] E.F. Mogos, Production dans les réseaux de distribution, étude
(2006) 1032–1040. pluridisciplinaire de la modélisation pour le contrôle des sources, Thèse de
[31] R. Teodorescu, F. Blaabjerg, M. Liserre, P.C. Loh, Proportional-resonant doctorat, ENSAM, Lille, 2004.
controllers and filters for grid connected voltage-source converters, IEEE [35] M. Bojrup, Advanced Control of Active Filters in a Battery Charger Application,
Proceedings—Electric Power Applications 153 (5) (2006) 750–762. Doctorat Thesis, Lund Institute of Technology, 1999.
[32] Y. Pankow, Etude de l’intégration de la production décentralisée dans un [36] H. Akagi, A. Nabae, Control strategy of active power filters using multiple
réseau basse tension, application au générateur photovoltaïque Thèse de voltage source PWM converters, IEEE Trans. Ind. Electron. 22 (3) (1986)
Doctorat, ENSAM, Lille, 2005. 460–465.
[33] B. François, Conception des dispositifs de commandes des convertisseurs de
puissance par modulation directe des conversions. Perspectives pour