You are on page 1of 12

Available online at www.sciencedirect.

com

Solar Energy 84 (2010) 11751186


www.elsevier.com/locate/solener

Direct power control of grid connected PV systems


with three level NPC inverter
Jaime Alonso-Martnez *, Joaqun Eloy-Garca, Santiago Arnaltes
Dept. of Electrical Engineering, University Carlos III of Madrid, Avda. Universidad 30, 28911 Leganes, Madrid, Spain
Received 2 December 2009; received in revised form 12 March 2010; accepted 21 March 2010
Available online 27 April 2010
Communicated by: Associate Editor Igor Tyukhov

Abstract
This paper presents the control of a three-level Neutral Point Clamped (NPC) voltage source inverter for grid connected photovoltaic
(PV) systems. The control method used is the Extended Direct Power Control (EDPC), which is a generic approach for Direct Power
Control (DPC) of multilevel inverters based on geometrical considerations. Maximum Power Point Tracking (MPPT) algorithms, that
allow maximal power conversion into the grid, have been included. These methods are capable of extracting maximum power from each
of the independent PV arrays connected to each DC link voltage level. The rst one is a conventional MPPT which outputs DC link
voltage references to EDPC. The second one is based on DPC concept. This new MPPT outputs power increment references to EDPC,
thus avoiding the use of a DC link voltage regulator. The whole control system has been tested on a three-level NPC voltage source
inverter connected to the grid and results conrm the validity of the method.
2010 Elsevier Ltd. All rights reserved.
Keywords: Direct power control; MPPT; Multilevel converter; Photovoltaic power systems

1. Introduction
In the last decades there has been a growing concern
about energy resources, as the energy consumption trend
does not seem to be sustainable. This situation has led to
research in renewable energies. Nowadays, wind energy
can be said to be the only well established renewable
energy. Solar photovoltaic energy is following its path in
terms of installed power increase and applications (grid
connection instead of stand-alone systems).
Higher power usually demands higher voltages, in order
to maintain currents at an acceptable level. Isolation voltage limit of PV panels constrains the number of series-connected panels in a PV array, thus limiting its maximum DC
*
Corresponding author. Address: Carlos III University of Madrid,
Department of Electrical Engineering, Avda. Universidad 30, 28911
Leganes, Madrid, Spain. Tel.: +34 91 624 88 52; fax: +34 91 624 94 30.
E-mail address: jalonsom@ing.uc3m.es (J. Alonso-Martnez).

0038-092X/$ - see front matter 2010 Elsevier Ltd. All rights reserved.
doi:10.1016/j.solener.2010.03.023

voltage level. Therefore, multilevel inverters appear to be a


very good solution for solar applications, as PV arrays concatenation is straight forward to each level of the DC link.
In this scenario, power control of multilevel inverters for
photovoltaic applications is recently being considered.
Several methods for power control of multilevel inverters have already been presented. Most of them rely on
some kind of current control algorithm whose output is
modulated for switching the inverter (Rodriguez et al.,
2002). There are some modulation algorithms for multilevel inverters such as multicarrier PWM (Pulse Width
Modulation), Three Nearest Vectors Space Vector Modulation, modulation based on virtual vectors, harmonic
elimination, etc. (Bouhali et al., 2007; Yao et al., 2008;
Barkati et al., 2008; Rahim et al., 2010). Nevertheless, there
is still a great interest on new modulation techniques to
improve multilevel inverter performance. On the other
hand, direct power control methods are found. All of them
are based on the direct torque control concept for electrical

1176

J. Alonso-Martnez et al. / Solar Energy 84 (2010) 11751186

Nomenclature

grid current complex vector
ia, ib, ic grid phase currents
id, iq
grid current components in the dq axis
middle point current
im
current of PV array PVn, n 2 {1, 2}
iPVn
L
lter inductance
P, Q
active and reactive power delivered to the grid
DP*, DQ* desired variations of P and Q (binary)
power delivered by PV array PVn, n 2 {1,2}
PPVn
state of switch n of branch x, x 2 {a, b, c},
Sxn
n 2 {1, 2, 3, 4}
sample time
Ts
vd, vq grid voltage components in the dq axis

machines (Takahashi and Noguchi, 1986; Depenbrock,


1988) and on the instantaneous power theory (Kim and
Akagi, 1999), developed for two-level inverters (Malinowski et al., 2001; Malinowski and Kazmierkowski, 2003;
Eloy-Garcia and Alves, 2006; Zhou et al., 2007; Chaoui
et al., 2008; Bouaa et al., 2009). The goal is to directly
control active and reactive power in an inverter connected
to the grid, in the same way as it was applied to control torque and ux in induction machines without modulators.
Among them, Extended Direct Power Control (EDPC) is
a generic algorithm applicable to any multilevel inverter,
no matter the number of levels. It is based on geometric
considerations about inverter voltage vectors and their
inuence on active and reactive power changes, as well as
middle points voltage control (Eloy-Garcia et al., 2007).
The use of EDPC together with an MPPT algorithm
makes it feasible to directly control the power of a grid connected PV system in a similar way as with electrical generators in wind energy. Moreover, a new MPPT is presented
that allows directly controlling DC link voltages, thus eliminating the need for the usual separate DC link voltage regulator and active power hysteresis comparator, as it is
shown in following sections.

total DC link voltage


vdc
vg
grid voltage complex vector
vgab, vgbc, vgca grid line voltages
vi
inverter voltage vector
viab, vibc, vica inverter line voltages
middle point voltage
vm
desired variation of vm (trinary)
Dvm
vn
inverter complex voltage corresponding to
switching state n, n 2 [0.27]
voltage across PV array PVn, n 2 {1, 2}
vPVn
DvPVn desired variation of vPVn
x*
reference for variable x
Dx
increment of variable x over a sampling period

In such a system, current increments can be expressed as


D 

Ts
vi  vg ;
L

where vg vgab  vgbc a2 ; vi viab  vibc a2 and  ia


2p
ib a ic a2 (a ej 3 ) are grid voltage, inverter voltage and
current vectors respectively and Ts is the sample time.
Expressing these vectors in a synchronous reference
frame aligned along vg , active and reactive power increments are respectively
DP Did  vd ;
DQ Diq  vd ;

being id and iq direct and quadrature current components


respectively, and vd the direct component of grid voltage.
Thus, given a grid voltage vg , it is possible to determine
the eect of each inverter switching state in active power P

2. Extended direct power control


2.1. Multilevel direct power control
This section contains a very brief summary of DPC and
EDPC algorithms. More detailed explanation and proof of
all statements can be found in (Eloy-Garcia et al., 2007).
Direct Power Control (DPC) of a three-phase inverter is
based on the changes in active and reactive power that follow the application of a certain inverter voltage vector.
Fig. 1 shows a grid connected three-phase three-level inverter with an inductive lter, and Fig. 2 shows the 27 possible
inverter switching states and their respective inverter voltage vi .

Fig. 1. Three-phase two-level inverter connected to the grid.

J. Alonso-Martnez et al. / Solar Energy 84 (2010) 11751186

1177

and small voltage vectors (v8  v13 and v20  v25 ) can
change vm in both directions due to their intrinsic multiplicity (two switching states for the same output voltage). This
is clearly seen in Table 1, where inverter voltage vectors
and corresponding middle point voltage change are shown.
Thus, in EDPC, after estimating P and Q and measuring
vdc and vm, hysteresis comparators are used to calculate
DP*, DQ* and Dvm , that are the signs of the desired variations of P, Q and vm respectively. P and Q are binary variables, and vm is trinary. With these inputs, EDPC decides
which inverter voltage vector to apply among those that
will cause changes in P, Q and vm of the desired sign. If
any of those references cannot be followed, a relaxed decision algorithm chooses a sub-optimal solution. Again,
proof of this and a more detailed explanation can be found
in (Eloy-Garcia et al., 2007).
EDPC scheme is shown later in Section 4.
Fig. 2. Three-level NPC inverter voltage vectors and switching states.

3. Multilevel grid-connected photovoltaic systems


and reactive power Q, and therefore it is possible to choose
the inverter switching state that best suits the desired
changes in P and Q.
Several strategies exist to choose the most appropriate
switching state. A general algorithm for multilevel inverters, Extended Direct Power Control (EDPC), has been proposed (Eloy-Garcia et al., 2007), based on geometrical
considerations.
2.2. DC link middle point voltage control
If an inverter voltage vector switches the middle point of
the DC link, then its voltage vm changes depending on
inverter output currents, as middle point output current
im can be calculated from inverter switching state as
im S a2 S a3 ia S b2 S b3 ib S c2 S c3 ic ;

Several multilevel PV topologies and control strategies


have been proposed (Calais et al., 1999; Shimizu et al.,
2003; Walker and Sernia, 2004; soon Kang et al., 2005; Ozdemir et al., 2008; Busquets-Monge et al., 2008). There is predominance in the literature of single phase inverters based in
cascaded H-bridge topologies, but other options do exist as
well. More recently, large photovoltaic systems have become
common and three-phase inverters have gained attention,
mainly based in NPC topologies, with lower power switch
count compared to cascaded H-bridge inverters.
When connecting the PV generator to the multilevel
inverter, there are three main alternatives. The rst one is
to connect the whole generator to the inverter as shown

where Sx2 and Sx3 are the middle switches in branch


x = {a, b, c}, as shown in Fig. 1. Therefore, choosing an
appropriate switching state allows EDPC to directly control vm. It has to be taken into account that large voltage
vectors v1  v6 and zero voltage vectors (v7 ; v26 and v27 )
cannot change vm without any change in vdc because Sx2Sx3
equals always zero; medium voltage vectors (v14  v19 ) can
change vm only in one direction, depending on currents;
Table 1
Switching states of a three-level NPC inverter and their inuence on
middle point voltage control.
vi

im

sgn(Dvm)

v1 -v7 ; v26 ; v27


v8 ; v15 ; v18 ; v23
v11 ; v20
v10 ; v14 ; v17 ; v25
v13 ; v22
v9 ; v24
v12 ; v16 ; v19 ; v21

0
ia
ia
ib
ib
(ia + ib)
(ia + ib)

0
sgn(ia)
sgn(ia)
sgn(ib)
sgn(ib)
sgn(ia + ib)
sgn(ia + ib)

(a)

(b)

(c)
Fig. 3. Dierent topologies of multilevel grid-connected PV systems.

J. Alonso-Martnez et al. / Solar Energy 84 (2010) 11751186

1178

in Fig. 3a. This option forces the voltage in all PV modules


to be the same, which leads to a sub-optimal operating
point in the case of heterogeneous conditions across the
PV eld such as partial shadowing, dust accumulation,
module mismatches (due to either manufacturing or ageing) or orientation errors in the case of solar tracking
devices (Busquets-Monge et al., 2008).
The other two options consist in connecting a fraction of the
PV array to each DC level of the inverter, either with or without an intermediate DC/DC stage, as shown in Fig. 3b and c.
This allows an independent voltage control of each fraction of
the PV array, therefore allowing to more accurately track the
optimal operating point in case of heterogeneous conditions.
The intermediate DC/DC converter allows the inverter to
operate with equal and constant voltages in all DC link capacitors, but decreases eciency and increases costs.
In this paper, a control strategy is presented for a diode
Neutral Point Clamped three level inverter without intermediate DC/DC conversion stage, as shown in Fig. 3b.
4. Voltage reference based multilevel MPPT
The power delivered by a PV system varies with the voltage across the array, the irradiance and cell temperature
(Fig. 4). Therefore PV systems need to implement a maximum power point tracking algorithm in order to maximize
the PV modules output power.
Several Maximum Power Point Tracking strategies have
been developed (Calais et al., 1998; Hohm and Ropp, 2003;
Esram and Chapman, 2007; Kimball and Krein, 2008; Chu
et al., 2009) such as perturb & observe, incremental conductance, power plant models, periodic sweeps, open circuit voltage/short-circuit current, fuzzy algorithms and
ripple-based methods.
4.1. Voltage reference based MPPT
Most of the MPPT implementations for photovoltaic
inverters output either a dc-link voltage reference to the
120

Current (Ix10)

100

80

Power (W)

60
2

G = 1000 W/m

20

Pmax

G = 300 W/m2

10

15

20

25

30

35

Voltage (V)
Fig. 4. PV module power and current vs. voltage.

being DvPVn the measured variation of the voltage, DPPVn


the measured variation of power, and DvPVn the desired variation of voltage, all of them refered to PV array PVn, with
n = {1, 2}.
The desired voltage variation for vm corresponds to the
desired voltage variation for array PV2, and the desired
voltage variation for vdc is the sum of the desired voltage
variations for both PV arrays. Finally those desired variations are integrated after applying a gain, in order to obtain
the references vdc and vm .
4.2. Simulation results

Pmax

40

inverter, or a duty cycle reference to a DC/DC converter


depending on the system topology.
Such a conventional approach will be used in this section as a rst attempt to integrate a MPP tracking algorithm in the inverter control. The designed MPPT
module outputs a reference voltage for each of the voltages
in the dc-link, vm and vdc . The reference voltage for the middle point vm is fed to a three-level hysteresis regulator, and
the total dc-link voltage reference vdc is fed to a PI voltage
regulator which outputs an active power reference to the
inverter hysteresis regulator, as can be seen in Fig. 5.
For a PV system it is well known (Esram and Chapman,
2007) that dP/dv is positive when the voltage is lower than
the voltage of the maximum power point, negative when it
is higher and zero when it is equal, as can be seen in Fig. 4.
This property is used in the MPPT algorithm as shown in
Fig. 6. It is a simple ripple based algorithm for a multilevel
inverter: no perturbation is introduced in the system, but
instead the natural perturbations in DC voltage caused
by the inverter switching are observed.
For each of the PV arrays, PV1 and PV2, the sign of the
variations of power and voltage over two consecutive samples are computed. The sign of the desired voltage variation
in each PV array that will cause the PV power to increase in
that array is the product of the signs of those derivatives as
in





DP PVn
DvPVn
 sgn
4
sgn DvPVn sgn
Dt
Dt

40

45

The proposed control system has been simulated in


Matlab, and the power section has been co-simulated in
Psim.
Each of the simulated PV arrays, PV1 and PV2, consist
in 2 strings of 11 Isofoton I110/24 modules. The modules
have been simulated after the characteristics provided by
the manufacturer, shown in Section 6. A temperature
dependent single exponential model with series resistance
as described in Hansen et al. (2000) has been used for module simulation.
The dc-link capacitors are 2200 lF each, and the AC lter has an inductance of 20 mH. The system is connected to
a standard 380 V grid. The system has been simulated with
a time step of 20 ls.

J. Alonso-Martnez et al. / Solar Energy 84 (2010) 11751186

1179

Fig. 5. First approach of EDPC with MPPT.

Fig. 6. Voltage reference based MPPT.

Fig. 7 shows the behavior of the proposed control system. In Fig. 7ae, a step in irradiance is applied to array
PV2 at t = 0.3 s, changing from 1000 to 750 W/m2. For
array PV1, the irradiance remains constant and equal to
1000 W/m2.
As can be seen from these gures, the MPPT controller
quickly adjusts voltage references in order to track the
maximum power point. Fig. 7d clearly shows that the maximum power point is tracked with an eciency of virtually
100% in stationary conditions. After the irradiance step,
the voltage of array PV1 remains constant, and the voltage
of array PV2 is increased in order to extract the maximum
power from the array with the new irradiance. After less
than 150 ms, the tracking eciency is 100% again.
Contrary to what happens with experimental data, the
eciency in simulation results is very straightforward to

calculate, since the maximum power for a given irradiance


and temperature can be deduced from the PV model. In
PV
, being PPV
this case, the eciency is computed as g PPMPP
the power delivered by the array and PMPP the maximum
power for the current irradiance and temperature.
Fig. 7ef show the system response to a step in reactive
power reference, which is followed almost instantaneously.
Phase currents in Fig. 7b show little distortion, especially for a variable frequency DPC algorithm, with a
THD of 2.49%.
5. Direct control MPPT
The main idea that drives Direct Power Control is to
directly control delivered active and reactive power by
means of simple, fast and robust relay controllers. These

J. Alonso-Martnez et al. / Solar Energy 84 (2010) 11751186

1180

(V)

750

20

dc

700

0.4

0.6

0.8

Vdc1, Vdc2 (V)

t, s

I (A)

10
650
0.2
400

10

350
300
0.2

0.4

0.6

20
0.1

0.8

0.2

0.3

t, s

pv1

2270
2260

0.3

0.35

0.4

0.45

0.998
0.2

0.5

0.4

0.6

0.8

pv2

2000

(W)

0.8

2200

pv2

0.6

t, s

t, s

1800
0.25

0.5

1.002

Ppv1(W)

2280

2250
0.25

0.4

t, s

0.3

0.35

0.4

0.45

0.99
0.2

0.5

0.4

t, s

t, s

40

6000

20

V (V), I (A)

2000
0

P (W), Q (VAr)

4000

2000

20
4000
6000
0.2

0.4

0.6

0.8

t, s

40
0.6

0.65

0.7

0.75

0.8

t, s

Fig. 7. Simulation results for conventional MPPT. Time step is 20 ls.

objectives are not fully met by the MPPT proposed in the


previous section, as there are still linear controllers that
need adjustment and reduce the system dynamics, such as
the vdc PI controller. There could be even more parameters
to adjust if a more complex MPPT method had been used.
5.1. MPPT integration in EDPC control
In this section, a novel MPPT strategy is presented that
tightly integrates with Extended Direct Power Control. The

idea is to implement an MPPT that directly outputs commands to the EDPC controller, avoiding the need for intermediate references or controllers that require parameter
tuning. Such an MPPT is shown in Figs. 8 and 9.
To understand how this MPPT works it is important to
clarify what is the eect of each of the commands that can
be given to the EDPC controller.
The rst command is the desired variation of the active
power delivered to the grid DP*. This power is drawn from
the dc link capacitors. If DP* = 1, the power drawn from

J. Alonso-Martnez et al. / Solar Energy 84 (2010) 11751186

1181

Fig. 8. Second approach of EDPC with MPPT.

Fig. 9. Direct Control MPPT.

both dc-link capacitors will increase, and if DP* = 0, it will


decrease.
The command to control of the middle point of the
inverter Dvm determines if the middle point current im, as
shown in Fig. 10, is going to be lower, equal or greater than
zero. This allows to control the dierence in voltage
between the two PV if it is possible to apply an inverter vector so that im  (iPV2  iPV1) can have an arbitrary sign. im
depends on the vector applied by the inverter as shown in
Table 1.
The command Dvm does not change the total active
power delivered to the grid, so it will not signicantly
change the total dc link voltage, assuming that vPV1 and
vPV2 are not very dierent and the PV power remains
constant.

Fig. 10. One branch of a three-level NPC voltage source inverter with
attached PV arrays.

J. Alonso-Martnez et al. / Solar Energy 84 (2010) 11751186

1182

Table 2
Edpc commands output by Direct Control MPPT.
DvPV1

DvPV2

DP*a

Dvm

1
1
1
1

1
1
1
1

1
toggle DP*
toggle DP*
0

0
1
1
0

a
These dP* commands will be overridden if jP  PPVj>limit: If
P  PPV > Plimit, dP* = 0. If P  PPV < Plimit, dP* = 1.

The rst stage of the MPPT is similar to the one in the


previous sections. The sign of the voltage variations that
maximize the power in each PV array is computed. Given
those signs and the previous considerations, the commands
given to the EDPC algorithm are shown in Table 2.
In DPC, the variation of active power delivered to the
grid is extremely fast. On the other hand, the variation of
the dc voltage is proportional to P  PPV. To achieve stable operation, it should be possible to change the sign of

20

700
10
650
0.2

0.4

0.6

0.8

Vdc1, Vdc2 (V)

400

10

350
300
0.2

0.4

0.6

20
0.1

0.8

t, s

0.3

pv1(W)

2260
0.3

0.35

0.4

0.45

0.998
0.2

0.5

0.4

(W)

0.35

0.8

0.6

0.8

pv2

(W)
P

2000

0.3

0.6

t, s

2200

pv2

0.5

t, s

1800
0.25

0.4

1.002

2270

2250
0.25

0.2

t, s

2280

Ppv1(W)

t, s

I (A)

dc

(V)

750

0.4

0.45

0.99
0.2

0.5

0.4

t, s

t, s

6000

40

20

V (V), I (A)

2000
0

P (W), Q (VAr)

4000

2000

20
4000
6000
0.2

0.4

0.6

t, s

0.8

40
0.6

0.65

0.7

t, s

Fig. 11. Simulation results for Direct Control MPPT. Time step is 20 ls.

0.75

0.8

J. Alonso-Martnez et al. / Solar Energy 84 (2010) 11751186

1183

the dc voltage variation fast enough to avoid oscillations


around the maximum power point. Therefore, it is needed
to limit the dc-link dynamics, so the commands for DP*
will be overridden in order to keep jP  PPVj below a certain limit as shown in Table 2.
5.2. Simulation results
Simulation results for Direct Control MPPT are shown
in Fig. 11.
All parameters and system characteristics are the same
as in the previous simulation results for the conventional
voltage reference based MPPT. As in the previous case,
an irradiance step is applied to array PV2 at t = 0.3 s,
changing from 1000 to 750 W/m2. For array PV1, the irradiance remains constant and equal to 1000 W/m2.
As can be seen from Fig. 11ae, this MPPT accurately
tracks the maximum power point, both in stationary conditions and after the irradiance step. It exhibits a much faster
response than the previous MPPT, even though the
jP  PPVj limit has been set to a rather conservative value
of 200 W.
Also, no cross eects in voltage and eciency can be
seen. In the previous case, the irradiance step in array
PV2 somewhat aected the voltage and the eciency of
array PV1 during a short period of time. With Direct Control MPPT, no perturbation whatsoever can be seen in
array PV1.
As in the previous case, a step in reactive power reference was introduced at t = 0.7 s. The system response can
be seen in Fig. 11e and f. The response is quite similar to
the previous case, as no change in active power is involved.
Phase currents in Fig. 11b show little distortion, especially for a variable frequency DPC algorithm, with a
THD of 2.49%.
6. Experimental results
The proposed control method has been tested in an
experimental rig (Fig. 12) with a ds1102 real-time control
board. A three-level NPC voltage source inverter has been
connected through an inductive lter of 20 mH to a 380 V
grid, in order to compare simulated and experimental
results. Due to real-time control board limitations, experimental test have been carried out at 200 ls with an inverter
switching frequency of 5 kHz. DC link capacitors are
2200 lF. Each PV array is made of 2 strings of 11 Isofoton
I110/24 modules connected in series. These modules have
the following characteristics: Voc = 43.2 V, Isc = 3.38 A,
Vmpp = 34.8 V, Impp = 3.16 A.
The Figs. 13 and 14 show the performance of the system
under a change in irradiance for PV2, with both MPPT
strategies. This change produced a power variation from
about 1900 W to about 1500 W in array PV2. Voltage
sweeps were performed in both arrays just before and after
the change in irradiance to obtain the maximum power
point voltage. Those voltages corresponded to the voltages

Fig. 12. Experimental rig.

shown in Fig. 13a and b, conrming that both MPPT modules were correctly tracking the optimum.
Only phase-a grid voltage and current and DC link voltages oscilloscope screenshots are shown. Voltage and current probes transformation ratio are, respectively, 60 and
2.5. Powers have been post processed and plotted in
MATLAB.
Phase currents in Fig. 13 show signicant distortion,
with a THD of 29.7% for conventional MPPT and a
THD of 19.8% for Direct Control MPPT. This was
expected because of the very high step time (200 ls), due
to real-time control board limitations. With this time step,
any variable switching frequency algorithm, as EDPC, will
always show such high THD values, even in simulation. It
is worth noting that Direct Control MPPT performs significantly better with high time steps, due to faster dynamics.
Besides, power factor control capability of the inverter
has also been tested. In Figs. 13 and 14, a step of 2500 VAr
is applied, yielding a capacitive power factor. Both MPPT
algorithms perform quite similarly in this case, as no active
power change is involved.
7. Conclusion
A new method for controlling a grid connected photovoltaic system has been presented. Two separate PV arrays
have been connected to the grid by means of a three-level
NPC voltage source inverter with an inductive lter. This
inverter has been controlled using Extended Direct Power
Control, a generic control method for multilevel inverters
which allows direct control of active and reactive powers
and middle point voltage control without parameter tuning, showing an excellent dynamic response. Besides, two

1184

J. Alonso-Martnez et al. / Solar Energy 84 (2010) 11751186

Fig. 13. Experimental results for conventional MPPT and Direct Control MPPT. Voltages and currents. Time step is 200 ls.

MPPT algorithms have been included to maximize power


generation from PV arrays. Both of them allow independent control of each PV array voltage, so that power can
be maximized in case of heterogeneous conditions across
PV arrays, such as partial shadows, dust accumulation,
module mismatches, etc. Moreover, the second MPPT
algorithm is a new MPP tracking method which directly
outputs power increments to EDPC, rather than voltage

references, making it ideal for integration with EDPC, as


it maintains the direct power control philosophy and
requires the tuning of fewer parameters. Simulation and
experimental results show a very good performance in both
cases, both for delivered power maximization and for
power factor control, even for such a low power test bench
with considerable low switching frequency for Direct
Control.

6000

6000

4000

4000

P (W), Q (VAr)

P (W), Q (VAr)

J. Alonso-Martnez et al. / Solar Energy 84 (2010) 11751186

2000
0
2000
4000
6000

2000
0
2000
4000

24.9

25

6000

25.1

79

79.1

t, s

(W)

1500

pv1

2000

Ppv1(W)

1500
24.9

24.95

25

25.05

1000

25.1

79.1

t, s
(W)

1500

pv2

2000

24.9

24.95

25

25.05

1000

25.1

79.3

79.4

79.1

79.2

79.3

79.4

t, s

6000

6000

4000

4000

P (W), Q (VAr)

P (W), Q (VAr)

t, s

2000
0
2000
4000
6000
53.6

79.2

2500

(W)

pv2

1000

79.4

t, s

2500

1500

79.3

2500

2000

2000

79.2

t, s

2500

1000

1185

2000
0
2000
4000

53.7

53.8

53.9

t, s

6000

111.8 111.85 111.9 111.95 112 112.05

t, s

Fig. 14. Experimental results for conventional MPPT and Direct Control MPPT. Active, reactive and PV power. Time step is 200 ls.

References
Barkati, S., Baghli, L., Berkouk, E.M., Boucherit, M., 2008. Harmonic
elimination in diode-clamped multilevel inverter using evolutionary
algorithms. Electric Power Systems Research 78 (10), 17361746.
Bouaa, A., Krim, F., Gaubert, J., 2009. Design and implementation of
high performance direct power control of three-phase PWM rectier,
via fuzzy and PI controller for output voltage regulation. Energy
Conversion and Management 50 (1), 613.
Bouhali, O., Francois, B., Berkouk, E., Saudemont, C., 2007. DC link
capacitor voltage balancing in a three-phase diode clamped inverter
controlled by a direct space vector of line-to-line voltages. Power
Electronics, IEEE Transactions on 22 (5), 16361648.

Busquets-Monge, S., Rocabert, J., Rodriguez, P., Alepuz, S., Bordonau,


J., 2008. Multilevel diode-clamped converter for photovoltaic generators with independent voltage control of each solar array. Industrial
Electronics, IEEE Transactions on 55 (7), 27132723.
Calais, M., Agelidis, V.G., Meinhardt, M., 1999. Multilevel converters for
single-phase grid connected photovoltaic systems: an overview. Solar
Energy 66 (5), 325335.
Calais, M., Hinz, H., 1998. A ripple-based maximum power point tracking
algorithm for a single-phase, grid-connected photovoltaic system.
Solar Energy 63 (5), 277282.
Chaoui, A., Krim, F., Gaubert, J., Rambault, L., 2008. DPC controlled
three-phase active lter for power quality improvement. International
Journal of Electrical Power & Energy Systems 30 (8), 476485.

1186

J. Alonso-Martnez et al. / Solar Energy 84 (2010) 11751186

Chu, C., Chen, C., 2009. Robust maximum power point tracking method
for photovoltaic cells: a sliding mode control approach. Solar Energy
83 (8), 13701378.
Depenbrock, M., 1988. Direct self-control (DSC) of inverter-fed induction
machine. Power Electronics, IEEE Transactions on 3 (4), 420429.
Eloy-Garcia, J., Alves, R., 2006. DSP-based direct power control of a VSC
with voltage angle estimation. In: Transmission & Distribution
Conference and Exposition, Latin America, 2006. TDC 06. IEEE/
PES. pp. 15.
Eloy-Garcia, J., Arnaltes, S., Rodriguez-Amenedo, J., 2007. Extended
direct power control for multilevel inverters including DC link middle
point voltage control. Electric Power Applications, IET 1 (4), 571580.
Esram, T., Chapman, P., 2007. Comparison of photovoltaic array
maximum power point tracking techniques. Energy Conversion, IEEE
Transactions on 22 (2), 439449.
Hansen, A., Srensen, P., Hansen, L., Binder, H., 2000. Models for a
Stand-alone PV System. Ris National Laboratory, Roskilde.
Hohm, D., Ropp, M., 2003. Comparative study of maximum power point
tracking algorithms. Progress in Photovoltaics 11 (1), 4762.
Kim, H., Akagi, H., 1999. The instantaneous power theory on the rotating
pqr reference frames. In: Power Electronics and Drive Systems,
1999. PEDS 99, Proceedings of the IEEE 1999 International Conference on, vol. 1. pp. 422427.
Kimball, J., Krein, P., 2008. Discrete-time ripple correlation control for
maximum power point tracking. Power Electronics, IEEE Transactions on 23 (5), 23532362.
Malinowski, M., Kazmierkowski, M., Hansen, S., Blaabjerg, F., Marques,
G., 2001. Virtual-ux-based direct power control of three-phase PWM
rectiers. Industry Applications, IEEE Transactions on 37 (4), 1019
1027.
Malinowski, M., Kazmierkowski, M.P., 2003. Control of three-phase
PWM rectiers. In: Kazmierkowski, M.P., Krishnan, R., Blaabjerg, F.

(Eds.), Control in Power Electronics. Academic Press, Burlington, pp.


419459.
Ozdemir, E., Tolbert, L., Ozpineci, B., 2008. Fundamental frequency
modulated multilevel inverter for three-phase stand-alone photovoltaic
application. In: Applied Power Electronics Conference and Exposition, 2008, APEC 2008, Twenty-Third Annual IEEE. pp. 148153.
Rahim, N., Selvaraj, J., Krismadinata, C., 2010. Five-level inverter with
dual reference modulation technique for grid-connected PV system.
Renewable Energy 35 (3), 712720.
Rodriguez, J., Lai, J., Peng, F.Z., 2002. Multilevel inverters: a survey of
topologies, controls, and applications. Industrial Electronics, IEEE
Transactions on 49 (4), 724738.
Shimizu, T., Hashimoto, O., Kimura, G., 2003. A novel high-performance
utility-interactive photovoltaic inverter system. Power Electronics,
IEEE Transactions on 18 (2), 704711.
soon Kang, F., Cho, S.E., Park, S., Kim, C., Ise, T., 2005. A new control
scheme of a cascaded transformer type multilevel PWM inverter for a
residential photovoltaic power conditioning system. Solar Energy 78
(6), 727738.
Takahashi, I., Noguchi, T., 1986. A new quick-response and higheciency control strategy of an induction motor. Industry Applications, IEEE Transactions on IA-22 (5), 820827.
Walker, G., Sernia, P., 2004. Cascaded DCDC converter connection of
photovoltaic modules. Power Electronics, IEEE Transactions on 19
(4), 11301139.
Yao, W., Hu, H., Lu, Z., 2008. Comparisons of spacevector modulation
and carrier-based modulation of multilevel inverter. Power Electronics,
IEEE Transactions on 23 (1), 4551.
Zhou, G., Wu, B., Xu, D., 2007. Direct power control of a multilevel
inverter based active power lter. Electric Power Systems Research 77
(3-4), 284294.

You might also like