See discussions, stats, and author profiles for this publication at: https://www.researchgate.

net/publication/325978011

Maximum Power Point Controller for Large-Scale Photovoltaic Power Plants

Using Central Inverters Under Partial Shading Conditions

Article  in  IEEE Transactions on Power Electronics · June 2018

DOI: 10.1109/TPEL.2018.2850374

CITATIONS READS

21 625

3 authors:

Baburaj Karanayil Salvador Ceballos

UNSW Sydney Tecnalia Corporación Tecnológica
32 PUBLICATIONS   734 CITATIONS    138 PUBLICATIONS   4,393 CITATIONS   

SEE PROFILE SEE PROFILE

J. Pou
Nanyang Technological University
318 PUBLICATIONS   11,620 CITATIONS   

SEE PROFILE

Some of the authors of this publication are also working on these related projects:

High-voltage dc grids for flexible and efficient electricity transmission View project

BEST PATHS View project

All content following this page was uploaded by Baburaj Karanayil on 03 July 2018.

The user has requested enhancement of the downloaded file.

 1

Maximum Power Point Controller for Large Scale

Photovoltaic Power Plants Using Central Inverters
under Partial Shading Conditions
Baburaj Karanayil, Senior Member, IEEE, Salvador Ceballos and Josep Pou, Fellow, IEEE

Abstract—The central inverter topology based on three-level radiation and low temperatures (-20 ◦ C). There are currently
converters is widely used in large scale photovoltaic (PV) power developments of 2.3 MW at 1500 V, PV inverters reported [4],
plants because of its simple and reliable structure and high [5]. These inverters at 1000 V -1500 V dc have only a single
efficiency. However, when two PV arrays are connected in series
for higher dc-link voltage, single maximum power point tracking MPPT operation.
(MPPT) operation cannot function properly when there is partial Central inverters are usually connected to several PV arrays,
shading between the arrays. This paper presents a configuration each made up of many PV panels connected to the dc-link
where the two PV arrays are connected in series together with of the inverter [6]. This solution offers the cheapest price of
an auxiliary power converter feeding a neutral-point-clamped the inverter in $ /kW. Central inverters thus provide a simple
central inverter. This configuration ensures maximum power pro-
duction from both PV arrays during partial shading conditions, structure, reliable and efficient converter, making it one of the
achieving the MPPT operation for both PV arrays. The proposed most common solutions for large-scale PV plants. Because the
converter together with their controls are verified by simulation whole array is connected to a single inverter, this configuration
and experiment. can only provide a single MPPT operation and hence it is not
Index Terms—Photovoltaic systems, grid-connected inverters, a good solution when operating under partial shading, where
maximum power point tracking. part of the PV array will bring down the generated power
output from the array significantly.
I. I NTRODUCTION Improvement of the MPPT control without dc-dc converters
for three-phase inverter have been reported in [7], [8]. In [7],
The global solar photovoltaic (PV) energy conversion sys- the two capacitor voltages are controlled by adding or subtract-
tems grew significantly, to at least 75.4 GW in 2016, which ing the minimum turn-on time to the three-phase turn-on times.
represents a 50% growth year-on-year [1]. The total global The output distortions are then eliminated by modifying the
installed capacity at the end of 2016 amounted to at least 303 reference voltages. In [8], a zero-sequence voltage was used
GW. Grid-connected PV systems account for more than 99% for the asymmetric control of dc-link voltages. This method
of PV installed capacity compared to the stand-alone systems does not need additional hardware investment. However, the
[2]. effectiveness of the MPPT algorithm is limited, especially
Large scale PV power systems up to 1 MW are made of PV when the main converter is working at high modulation
arrays usually connected to 1000V dc central inverters. As the indexes and the power distribution between the PV sources is
technology becomes mature, there has been continuous effort very unbalanced. Under these conditions, the MPPT algorithm
in improving the system efficiency and in reducing the cost. is not able to work at the maximum power point.
The effective way to achieve these goals is to increase the A dc-bus voltage balancing algorithm for three-level neutral-
system voltage. Also, the voltage constraint imposes a limit point-clamped (NPC) traction inverter drive is proposed in
on the maximum number of modules that can be connected. [9] using modified virtual space vector PWM (VSVPWM)
Over time, PV inverter technology has settled for installations strategy, which uses both the positive and negative redundant
under 1000 V (open-circuit), and a typical MPPT bus range voltage vectors alternatively as a virtual vector.
covers from 575 V (lower dc-bus limit) to 850 V (upper dc-bus A single power stage topology was proposed for grid-
limit). At certain extreme temperatures (-20 ◦ C and 70 ◦ C) will connected inverter using three-level central NPC inverter [10]
position the MPP and power curves well outside the inverter by adding an auxiliary converter in parallel to the dc-link. This
dc-bus voltage operating area, compromising the inverter abil- solution follows a similar approach to the one proposed in this
ity to shield maximum power extraction [3]. Lower dc-bus paper. However, it requires additional inductive and capacitive
limit extension allows energy capture under low radiation and elements that make it a bit more complex and potentially
high PV temperatures (70 ◦ C) and the extension of the upper expensive.
dc-bus voltage allows for greater energy extraction under high In [11], the main focus is to develop a control strategy for
B. Karanayil is with the School of Electrical Engineering and Telecommu- the three-level NPC inverter when the different zones receive
nications, UNSW Sydney, NSW 2052, Australia (e-mail: [email protected]). different solar irradiance levels. Such a strategy would increase
S. Ceballos is with Tecnalia Research and Innovation, Derio, Spain (e-mail: power production under partial shading without additional
[email protected]).
J. Pou is with the School of Electrical and Electronic Engineering, Nanyang dc/dc converters and without communication between individ-
Technology University, 639798 Singapore (e-mail: [email protected]). ual inverter controllers. A real MPPT method for mismatching
 2

compensation in PV array under partial shading conditions operate at higher efficiency. Besides, the power rating of this
is presented in [12]. In this case, the full PV array power auxiliary unit is considerably low which will make the system
flows through a dc-dc converter to the grid-connected inverter. economically feasible and operate with reduced switching and
The distributed MPPTs enables better power generation during conduction losses. The generated power from the PV array
partial shading at the expense of higher switching losses in is maximized to achieve the best system efficiency. All these
the additional power converters during less shading conditions features make the proposed solution particularly feasible for
and vice versa for the central MPPTs [13]. The annual energy large power plants with central inverters, where the system
yield with the help of distribution of dc-dc converters and efficiency is very critical.
MPPT controllers in a large scale PV power plant, where This paper is organized as follows. Section II presents the
the dc-dc converters are connected between the PV arrays conventional and the proposed MPPT schemes for central NPC
and central inverter [14]. In common large-scale installations, inverter topology. The operation of the auxiliary converter
hundreds or even thousands of PV panels are arranged in for the proposed MPPT is described in Section III. The
arrays and connected to a single inverter. Even though central elimination of low-frequency ripples is explained in Section
converter uses a simple topology, has high reliability and good IV. Simulation results for both MPPT schemes and their
conversion efficiency, it has poor MPPT capabilities, which comparison are presented in Section V. Experimental results
impacts the overall system performance [15]. The effect of are discussed in Section VI. A comparitive analysis of MPPTs
partial shading in a large scale PV power plant is investigated is provided in Section VII. Finally, Section VIII summarizes
in [16], but here, microinverters are connected to the PV the main conclusions of this work.
panels. In the power plant proposed in [17], the PV module
output feeds the inverter through a current fed dual-active
bridge dc-dc converter with high-frequency isolation. A new
II. PV TO G RID I NTERFACE T OPOLOGIES
architecture for PV grid-connected converters based on a
string mini boost stage was introduced in [18]. In [19], an
A typical large scale PV inverter operates up to 1000 V dc
energy storage system is added at inverter level; enabling the
and has a single PV array connected to a the dc-bus. When
PV system to store energy during standard operation, and
the dc-link voltage is increased to 1500 V dc, it becomes
releasing stored energy when performing PV curve scan, hence
necessary to connect more panels in series or two PV arrays
achieving global MPPT in central inverter PV power plants.
in series as in Fig. 1, similar to a configuration described
The methods previously proposed in [8]–[18] can be cate-
in [20]. The single MPPT controller uses the total PV array
gorized in two groups. The first category is where the MPPT
voltage vP V and current iP V to estimate the MPP voltage
control is implemented without the need of an additional
vP∗ V Conv which sets the dc-link voltage reference vdc ∗
to
power converter. The implementation of these methods does
the inverter. The disdavantage is that when there is partial
not need additional investment in power converters which
shading on one of the arrays, the MPP operation fails and
make them attractive. In particular, the method proposed in
the net power extracted from the whole PV array is reduced.
[8] uses an NPC converter as the main converter, similar to
The optimal PV terminal voltage was calculated using an MPP
the solution proposed in this paper. However, as discussed
algorithm dP/dV control [21]. Few typical modelling results
above, the effectiveness of the MPPT in [8] is reduced at
for this scenario are described in Section V.
high modulation indexes when the power production of the
PV arrays is highly unbalanced. The method proposed in In order to extract the maximum power from the PV arrays
this paper overcomes this issue and guarantees the maximum during partial shading, the conventional MPPT of the central
power extraction over the whole range of working conditions. inverter is modified as shown in Fig. 2. Here, an auxiliary
The second category is where additional power converters converter unit is inserted in parallel to the dc-link. This system
are required as in [11]–[18]. Most of them use a dc-dc operates in two different modes based on the solar radiation
converter in series with the dc-link. This causes the entire levels on the two arrays. The ratio (pP V1 -pP V2 ) / Pmax of
dc power to go through this converter irrespective of the fact the measured powers of PV arrays P V1 and P V2 is used
that there is partial shading, which reduces the overall system to switch between conventional and proposed MPPT modes.
efficiency. In addition, whenever a dc-dc converter is added When the ratio (pP V1 -pP V2 ) / Pmax is < 20%, it operates in
in series between the PV array and the inverter, not only conventional MPPT mode. On the other hand, when (pP V1 -
the power loss of that converter reduces the overall system pP V2 ) / Pmax is ≥ 20%, it operates on the proposed MPPT
efficiency, but also the power rating of the dc-dc converter mode.
will be the same as that of the inverter, which makes this The grid-connected inverter control used in Figs. 1 and 2 are
option expensive. In large central inverters, the decision to voltage oriented control (VOC) implemented in the rotational
invest in extra power converters to gain power generation dq reference frame.
during partial shading is overshadowed by the additional cost As it is observed, it is based on the use of three proportional-
of this extra converter. These issues are overcome with the integral (PI) regulators to control the dc voltage and the dq
proposed solution. It consists of a partially-rated modular components of the current. This control scheme is well known
auxiliary power converter leg that is inserted in parallel to and there are multiple publications dealing with it. In order to
the dc-link. The auxiliary converter comes into operation only tune the PI regulators, the guidelines given in the Appendix
during excessive partial shading, which indirectly makes it of [22] can be followed.
 3
1

iP V
vP V

vC2 Lg
∗
vdc = vP∗ V−Conv ia ea
Lg ib eb
vdc i0 ic
Lg ec

vC1 eC eB eA

id iq ed eq

Fig. 1. Large scale grid-connected PV energy conversion system: Conventional MPPT for central NPC inverter topology (two arrays in series forming the
dc-link).

iP V2 iP V
T4

vP V2
T3 vC2
vP V
Lg ia ea
iL
vdc Lg ib eb
iP V1 T2 i0 Lg ic ec
T1
vP V1 vC1
eC eB eA

id iq ed eq
T1
T2 vP∗ V−Conv
T3
T4
vP∗ V−P rop

vP∗ V−Conv ∗
vP∗ V−P rop vdc
 
 pP V1 −pP V2 
 Pmax 

Fig. 2. Large scale grid-connected PV energy conversion system: Proposed MPPT for central NPC inverter topology.

A. Conventional MPPT Mode then the MPPT tracks the total power of the two arrays as if
it was only one. The calculated total dc-link voltage reference
In this case, the two PV arrays P V1 and P V2 are connected vP∗ V−Conv is provided to the NPC inverter control in Fig. 1.
to the grid only with the three-level NPC inverter as in
Fig. 1. During this conventional MPPT mode, the dc voltage
reference for the maximum power point is calculated by a B. Proposed MPPT Mode
MPPT controller based on the total dc voltage of two arrays
in series (vP V ) and the PV array current (iP V ) to the inverter. In this case, the auxiliary converter is added in parallel to
If the solar radiation level in the two arrays is similar which the dc-link and its output is connected to the midpoint of the
satisfies: two PV arrays P V1 and P V2 . In this mode, there are two
independent MPPT controllers for both PV arrays P V1 and
 
 pP V1 − pP V2  P V2 . The sum of the two voltage references calculated by
  < 20% (1)
 Pmax  the MPPT controllers (vP∗ V1 + vP∗ V2 = vP∗ V−P rop ) becomes the
 4

vP∗ V1
iP V2
T4 vP∗ V1 T1
∗
vP V1 T2
DC vP V2 C2 v2 iP V1 T3
T4
T3 vP V1
L iL vP V1 +vP V2
vP∗ V2
iP V1 vL vP V2 vP∗ V−P rop
iP V2
T2
∗
DC vP V1 C1 v1
Fig. 4. Control scheme for proposed auxiliary converter unit.
T1

vP∗ V1
Fig. 3. Proposed auxiliary converter. (V) vP V2
vP V1

dc-link voltage reference to the NPC inverter control in Fig. (a)

2.
When the solar radiation in one array is 20% different than
in the other one,
  (A) iP V2
 pP V1 − pP V2  iP V1
  ≥ 20% (2)
 Pmax 
then the two MPPT controllers operate independently. The (b)
reference voltage of the lower PV array (vP∗ V1 ) is introduced
iL
in the control of the auxiliary leg so that the current injected
into the NP of the PV arrays leads each individual voltage to (A)
the required value to achieve MPP.

III. AUXILIARY C ONVERTER AND C ONTROL

MPP operation of the two PV arrays is achieved with an (c)
auxiliary converter unit connected between the PV arrays and Time (s)
the NPC inverter, as shown in Fig. 2. This unit will be of
Fig. 5. Averaged waveforms of the auxiliary converter when there is partial
a lower power rating compared to that of the NPC inverter shading of P V1 .
as it only processes a portion of the dc current through the
inverter leg. The auxiliary converter has four semiconductor
switches T1 , T2 , T3 and T4 . However, only two-level voltages To illustrate the operation of this controller, an average
are applied to supply the difference in currents of the two PV model of the circuit shown in Fig. 3 is used. Fig. 5 shows
arrays, when there is partial shading in one of the arrays. the results obtained when the radiation of Array 1 (P V1 )
The operation of the proposed MPPT can be explained with drops. When the radiation drops, the MPPT algorithms of the
the help of Figs. 2 and 3. When the solar radiation on both PV PVs generate the optimum voltage references vP∗ V1 and vP∗ V2 to
arrays is alike, the voltage references vP∗ V1 and vP∗ V2 will be maintain the PV arrays working at their optimum power points.
similar and the auxiliary converter unit will be off. When there As the radiation of P V2 does not change, the voltage of P V2
is partial shading and the difference in solar radiation of the (vP V2 ) remains constant (Fig. 5 (a)). However, the voltage
two PV sources reaches a threshold, the voltage references reference of P V1 ( vP∗ V1 ) changes to maintain the maximum
vP∗ V1 and vP∗ V2 will differ and the auxiliary converter will power extraction of P V1 . The controller of the auxiliary
start operation. Under these circumstances, the sum of the converter makes the voltage of P V1 (vP V1 ) track the voltage
two voltage references ( vP∗ V1 +vP∗ V2 ) is provided as the input reference vP∗ V1 . Fig. 5(a) shows a good matching between
for the VOC of the NPC converter that regulates the total the voltage reference vP∗ V1 and the real voltage vP V1 at the
dc voltage. In addition, the auxiliary converter is controlled terminals of P V1 . In addition, as the radiation of P V1 drops,
in such a way that the voltage of the neutral point (NP) the current iP V1 also drops while iP V2 remains unaltered
is regulated to track the optimum voltage reference of P V1 (Fig. 5(b)). The current difference is supplied by the auxiliary
(vP∗ V1 ). The algorithm used to control the auxiliary converter converter through the inductor iL (Fig. 5(c)). A symmetric
is shown in Fig. 4. It consists on a PI controller to regulate behavior would be obtained if the partial shading affected P V2
the NP voltage together with a feedforward term to improve instead of P V1 . In this way, the auxiliary converter makes
its dynamic response. possible the operation of both panels at the optimum point
 5

and improves the efficiency of the whole system.

1
IV. E LIMINATION OF L OW F REQUENCY R IPPLES IN Irradiance
-Array PV1
DC-L INK 0.5
(kW/m2)
A drawback of the NPC converter is that a low frequency
voltage oscillation appears in the NP under certain operating 0

conditions. Therefore, the dc-link capacitors and the power

devices have to be oversized to withstand higher voltages 1
caused by the oscillation. This is totally eliminated using the Irradiance
-Array PV2
method proposed in [23]. However, this method requires an (kW/m2) 0.5
extra control loop for voltage balancing. A voltage balance
compensator for the NPC converter is proposed in [24]. 0
Nevertheless, in the proposed implementation, the inverter
does not need to take care of regulating capacitor voltages, 3000
since the auxiliary leg will perform this task. Grid Proposed MPPT
2000
In order to eliminate low-frequency capacitor voltage rip- power
ples, the locally averaged value of NP current must be zero. (W) Conventional
1000 MPPT
This is achieved by generating two reference signals per phase,
which are obtained as follows [23]: 0
7 AM 9 AM 11 AM 1 PM 3 PM 5 PM 7 PM
vi − min(va , vb , vc )
vip = (3) Time (Hr)
2
vi − max(va , vb , vc ) Fig. 6. Simulation results for comparison of generated grid power in
vin = (4) conventional and proposed MPPTs, from top to bottom: Irradiance of PV
2 Array P V1 , irradiance of PV Array P V2 , generated grid power for 14th
for i ∈ {a, b, c}, where vi is the normalized reference signal May, 2016.
for phase i, and vip and vin are the new reference signals for
that phase ranging in the intervals vip =[0,1] and vin =[0,-1] TABLE I
under linear operation mode. G AIN IN THE P OWER G ENERATION D UE TO THE P ROPOSED MPPT-
The three phase sinusoidal modulation signals of the inverter S IMULATION
are modified according to (3) and (4) so that locally-averaged Day Conventional MPPT Proposed MPPT Gain
NP current i0 is always zero. 4 May 2016 12.95 kWHr 14.32 kWHr 10.57 %
This modulation strategy does not introduce low frequency 14 May 2016 10.35 kWHr 12.76 kWHr 23.28 %
distortion in the output voltages even when the two dc-link 18 May 2016 12.45 kWHr 13.71 kWHr 10.15 %
capacitor voltages are unbalanced. This is demonstrated in this 31 May 2016 11.28 kWHr 13.06 kWHr 15.78 %
paper for the first time, as follows. 1 June 2016 4.48 kWHr 8.84 kWHr 97.29%
The locally-averaged voltage generated at phase i of the
9 June 2016 9.69 kWHr 12.63 kWHr 30.3%
NPC converter is:
vi0 = vin vC1 + vip vC2 (5)
V. S IMULATION R ESULTS
Substituting (3) and (4) in (5): The grid-connected NPC inverter was simulated to study the
 
1 effects of partial shading on PV array, both conventional and
vi0 = (vi vdc − max(va , vb , vc )vC1 − min(va , vb , vc )vC2 the proposed MPPT controllers are tested. The solar irradiance
2
(6) data on different days were used where the depth of partial
where vdc = vC1 + vC2 . shading were different. The gain in the power generation due
Considering the line-to-line voltage vij = vi0 −vj0 for i, j ∈ to the proposed MPPT was compared for different days in a
{a, b, c}, and i 6= j : season for a solar power plant with the specifications listed
vdc in Table II. The solar radiation data was taken from a solar
vij = vij (7)
2 power plant in operation in Brisbane, Australia.
where vij = vi − vj is the normalized reference for the corre- The power delivered to the grid was found when the two
sponding line-to-line voltage. As it can be observed in (7), the PV arrays, P V1 and P V2 , P V1 without shading and P V2 with
locally-averaged line-to-line voltage does not depend on each partial shading, for the solar irradiance data of 14th May,
capacitor voltage, but on the sum of the two capacitor voltages, 2016 and the results are shown in Fig. 6. The simulation was
i.e. the total dc-link voltage vdc . Therefore, no low-frequency repeated with different solar radiation data of 1st June, 2016
distortion will be produced in the line-to-line voltages due and the results are shown in Fig. 7.
to capacitor voltage imbalances when using this modulation The modelling was repeated for six different days of the
technique. This is important in this application because the season and the cumulative gains in power generation in the
dc-link capacitors will be unbalanced during partial shading day due to the proposed MPPT technique are listed in Table
of the PV arrays. I.
 6

1
Irradiance
-Array PV1 0.5
(kW/m2) 1
0

2
1
Irradiance
-Array PV2
(kW/m2) 0.5
3
0

3000 4
Grid Proposed MPPT Conventional MPPT
2000
power 5
(W)
1000
6
0
7 AM 9 AM 11 AM 1 PM 3 PM 5 PM 7 PM
Time (Hr) 7

Fig. 7. Simulation results for comparison of generated grid power in

conventional and proposed MPPTs, from top to bottom: Irradiance of PV
Array P V1 , irradiance of PV Array P V2 , generated grid power for 1st June,
2016.
8

VI. E XPERIMENTAL R ESULTS

9
The standard MPPT controller of a central NPC inverter,
where two equal PV arrays connected in series and the
proposed MPPT controller with an auxiliary power converter
10
connected in parallel to the dc-link, were investigated in detail
with a three-phase grid-connected NPC inverter set up in
11
the laboratory. Two 1000 V/10 A Regatron PV simulators
(TC16.1000.20 Q1-LIN) were used as two PV arrays P V1 ,
P V2 shown in Figs. 1 and 2. The specifications of the 12
experimental setup are listed in Table II. The grid-connected
inverter and the auxiliary converter in Fig. 2 were controlled 13
with a dSPACE 1006 system with two DS5203 FPGA boards.
A photograph of the experimental setup is shown in Fig. 8.
Experiments were conducted at around 2.0 kW power level.
Fig. 8. Experimental setup of the proposed MPPT for three-phase grid-
The main parts of the prototype are: connected NPC inverter.
1) Personal computer
2) Oscilloscope
3) Yokogawa WT1800 power meter A. Partial Shading on P V2 with Conventional MPPT
4) Sensor panel To investigate the effect of partial shading of one PV array,
5) DYN11 isolation transformer the PV array was split into two equal arrays P V1 and P V2
6) Yokogawa current sensor connected in series. At first, both Regatron PV simulators
7) Yokogawa voltage sensors were programmed to produce 1 kW output power each at
8) dSPACE DS1106 system irradiance of 1000 W/m2 . The steady state MPP curve was
9) NPC inverter recorded from the Regatron simulator. The MPPT curve of
10) Regatron photovoltaic simulator P V1 is shown in Fig. 9. Similarly, the MPPT curve of P V2
11) Regatron photovoltaic simulator is shown in Fig. 10. Both PV sources produced 1 kW output
12) Inductor at their MPP with a PV voltage vM P P = 125 V. Then, the
13) Grid filter inductors solar irradiance of P V2 was changed to 100 W/m2 over one
The three-level NPC inverter was controlled to connect to second time period. The array voltages vP V1 and vP V2 were
the 400-V ac grid through a step-up isolation transformer. The measured with differential voltage probes. The array currents
two Regatron PV simulators were used to function the two PV iP V1 and iP V2 were measured with differential current probes.
arrays P V1 , P V2 whose solar irradiance can be programmed The array powers were computed using the math function in
to change from 1000 W/m2 to 100 W/m2 . the oscilloscope. The steady state MPP curve was recorded
 7

TABLE II
S PECIFICATIONS OF THE NPC T HREE -P HASE PV-I NVERTER IN THE
E XPERIMENTAL S ETUP

Parameter Value
Maximum power of the PV array at Vmp , Pmax 1.0 kW
Open circuit voltage of the PV array, Voc 173.5 V
Voltage at maximum power, Pmax 125.0 V
Rated power of Regatron ACS grid simulator 50 kVA
Line-to-line voltage of Regatron ACS grid simulator, ELL 400 V
Grid-side line-line voltage, DYN11 isolation transformer 400 V
Inverter-side line-line voltage, DYN11 isolation transformer 200 V
Dc-link capacitance, C1 = C2 = C 4950 µF
External line inductor, 3 x Lg 10 mH
Auxiliary converter inductor, L 20 mH
Inverter switching frequency, fs 10 kHz
Auxiliary converter switching frequency, fc 10 kHz
MPPT controller frequency 100 Hz

Fig. 10. Experimental results: Power of PV Array P V2 with full solar

irradiation on PV arrays P V1 and P V2 with conventional MPPT.

Fig. 9. Experimental results: Power of PV Array P V1 with full solar

irradiation on PV arrays P V1 and P V2 with conventional MPPT.

from the Regatron simulator. The power of P V1 dropped to Fig. 11. Experimental results: Power of PV Array P V1 with partial shading
190 W at 168V. The MPPT curve of P V1 is shown in Fig. 11, on PV Array P V2 with conventional MPPT.
and the power of P V2 dropped to 90 W at 105 V, as indicated
by the MPPT curve in Fig. 12. The total output power from
1000 W/m2 . The steady state grid currents are recorded and
the two PV arrays dropped to 280 W. The variations in the
they are shown in Fig. 15. The THD of the grid voltage Uthd
array voltages, currents and powers were recorded in the
is 1.13 % and that of the grid current Ithd is 2.49 %, as shown
oscilloscope and are shown in Fig. 13. The harmonic spectrum
in Fig. 16. The THD values are calculated by averaging the
of grid voltage and current were measured with the power
individual THD values of the three phases. Then the same test
analyzer WT1800. The total harmonic distortion (THD) of the
as in Section VI-A was repeated. In this case, the power of
grid voltage Uthd is 1.552 % and that of the grid current Ithd is
P V2 was found to drop to 92 W at 123 V as indicated from
3.828 %, as shown in Fig. 14. The THD values are calculated
the MPP curve in Fig. 17, recorded from the first Regatron
by averaging the individual THD values of the three phases.
simulator. However, the power of P V1 was maintained at 1000
W as shown in the MPP curve in Fig. 18 recorded from the
B. Partial Shading on P V2 with Proposed MPPT Regatron simulator. The total power output from the two PV
In this test, the auxiliary converter was added, as shown arrays dropped to 1092 W. The variations in the array voltages,
in Fig. 2. As before, both Regatron PV simulators were currents and powers were recorded in the oscilloscope and is
programmed to produce 1 kW output power at irradiance of shown in Fig. 19. Here the current in the inductor L of the
 8

Fig. 14. Experimental results with Conventional MPPT: THD values of grid
currents and voltages without partial shading.

ia ib ic

Fig. 12. Experimental results: Power of PV Array P V2 with partial shading

on PV Array P V2 with conventional MPPT.

vP V
Time (10 ms/div)
vP V1
Fig. 15. Experimental results with proposed MPPT: Grid currents ia , ib , ic
(5A/div) without partial shading.

vP V2
iP V1

iP V2

pP V1
Fig. 16. Experimental results with proposed MPPT: THD values of grid
currents and voltages without partial shading.
pP V2

TABLE III
C OMPARISON OF THD VALUES
Time (500 ms/div)
Grid variables Conventional MPPT Proposed MPPT
Fig. 13. Experimental results for partial shading on PV Array P V2 with voltage ea 1.559 % 1.118 %
conventional MPPT, from top to bottom: PV array voltages vP V2 and vP V1 voltage eb 1.535 % 1.199 %
(100V/div), vP V (250V/div), PV array currents iP V2 and iP V1 , (10A/div),
PV array powers pP V2 and pP V1 (1000W/div). voltage ec 1.562 % 1.081 %
current ia 3.814 % 2.432 %
current ib 3.676 % 2.386 %
current ic 3.994 % 2.679 %
auxilary power converter, rises to 7.25 A as shown in Fig.
19. The difference between the two PV array currents iP V2 -
iP V1 because of the partial shading flows through the inductor
(L) as iL . The voltages vP V1 and vP V2 are not balanced now, Fig. 20. It was found that the THD of the grid voltage Uthd is
still there is no low frequency distortion in the grid currents 1.13 % and that of the grid current Ithd is 3.25 %, as shown
because of the modified PWM strategy described in Section in Fig. 21. A comparison of THD values of grid voltages and
IV. currents with the conventional and the proposed MPPTs is
The steady state grid currents are recorded and shown in shown in Table III.
 9

vP V

vP V1

vP V2
iP V1

iP V2

iL

Fig. 17. Experimental results: Power of PV Array P V2 with partial shading

on PV Array P V2 with proposed MPPT. pP V1
pP V2

Time (200 ms/div)

Fig. 19. Experimental results for partial shading on PV Array P V2 with

proposed MPPT, from top to bottom: PV array voltages vP V2 and vP V1
(100V/div), vP V (250V/div), PV array currents iP V2 and iP V1 (10A/div),
current iL (10A/div), PV array powers pP V2 and pP V1 (1000W/div).

ia ib ic

Fig. 18. Experimental results: Power of PV Array P V1 with partial shading

on PV array P V2 with proposed MPPT.
Time (10 ms/div)

Fig. 20. Experimental results with proposed MPPT: Grid currents ia , ib , ic

C. Partial Shading on P V1 with Conventional MPPT (5A/div) with partial shading on PV Array P V2 .
The test carried in Section VI-A was repeated here for
the solar irradiance change from 1000 W/m2 to 100 W/m2 in
P V2 to confirm the full operation of the auxiliary converter.
The power of P V1 dropped to 90 W. And the power of P V2
dropped to 190 W. The total power output from the two PV
arrays dropped to 280 W. The variations in the array voltages,
currents and powers were recorded in the oscilloscope and is
shown in Fig. 22.
Fig. 21. Experimental results with proposed MPPT: THD values of grid
D. Partial Shading on P V1 with Proposed MPPT currents and voltages with partial shading on PV Array P V2 .

The test performed in Section VI-B with the auxiliary con-

verter was repeated here for solar irradiance of P V2 changing
 10

vP V
vP V

vP V2 vP V2

vP V1
iP V2 vP V1
iP V2
iP V1

iP V1

pP V2 iL
pP V2
pP V1

Time (500 ms/div)

pP V1
Fig. 22. Experimental results for partial shading on PV Array P V1 with
conventional MPPT, from top to bottom: PV array voltages vP V2 and vP V1
(100V/div) , vP V (250V/div), PV array currents iP V2 and iP V1 (10A/div),
PV array powers pP V2 and pP V1 (1000W/div). Time (200 ms/div)

Fig. 23. Experimental results for partial shading on PV Array P V1 with

from 1000 W/m2 to 100 W/m2 . The power of P V1 dropped to proposed MPPT, from top to bottom: PV array voltages vP V2 and vP V1
(100V/div), vP V (250V/div), PV array currents iP V2 and iP V1 (10A/div),
92 W, and the power of P V2 remained at 1000 W. The total current iL (10A/div), PV array powers pP V2 and pP V1 (1000W/div).
power output from the two PV arrays dropped to 1092 W.
The variations in the array voltages, currents and powers were TABLE IV
recorded in the oscilloscope and are shown in Fig. 23. It can be G AIN IN THE P OWER G ENERATION D UE TO THE P ROPOSED MPPT-
noted that the voltages vP V1 and vP V2 are not balanced during E XPERIMENT
the shading, nevertheless because of the modulation strategy Mode Conventional MPPT Proposed MPPT Gain
employed, the grid currents could be controlled without low (pP V1 + pP V2 ) (pP V1 + pP V2 )
frequency distortion. Shading on P V1 280 W 1092 W 290 %
Shading on P V2 280 W 1092 W 290 %
VII. C OMPARISON OF MPPT S
It was found in Section VI.A, that when the conventional
MPPT controller is in operation, a partial shading of PV currents, therefore the cost is higher. Nevertheless, the benefits
array P V2 from 1000 W/m2 to 100 W/m2 , the combined of producing more energy justifies by far the extra cost of
output of two PV arrays ( pP V1 + pP V2 ) dropped to 280 W. the auxiliary leg, control and sensors.. A comparative analysis
Whereas, after changing to the proportional MPPT controller of proposed method with the previously proposed methods is
as discussed in Section VI.B, the total power output of the shown in Table V.
two PV arrays P V1 and P V2 increased to 1092 W. This is a
290% gain as indicated in Table IV. VIII. C ONCLUSION
Similarly, it was mentioned in Section VI.C, that when the The auxiliary power converter proposed in this paper en-
conventional MPPT controller operates, a partial shading on ables the maximum power production from the PV arrays used
P V1 from 1000 W/m2 to 100 W/m2 , the total PV array output in central inverters for large scale PV power plants during
dropped to 280 W. After replacing the conventional MPPT partial shading in one of the PV arrays. This modular unit
controller with a proportional MPPT controller, the total power adds only a small additional cost to the total system as the
( pP V1 + pP V2 ) increased to 1092 W as discussed in Section power ratings are only a fraction of that of the inverter. As
VI.D. This is also 290% gain as shown in Table IV. As this unit is switched off and goes to standby when there is no
explained in Section II.A, the conventional MPPT controller partial shading, there are no power losses with this MPPT unit
uses only two sensors, one for voltage and another one for and thus maximum conversion efficiency is achieved.
current. However, the proposed MPPT controller described in This configuration together with central inverters combines
Section II.B, needs four sensors for two voltages and two the best features of central inverters together with the complete
 11

TABLE V
C OMPARITIVE A NALYSIS OF MPPT S

Real MPPT [12] Distributed MPPT [13] Conventional MPPT [15] Proposed MPPT
Reliability High Low High High
Number of sensors High High Low High
Cost Low Highest Low High
Conversion efficiency High Low High High
Only one High number of Only one Additional converter
Number of converters power power converters in power of lower power rating
converter addition to main one converter
Same as that Same as that Same as that Can be selected depending
Rated power of MPPT
of the inverter of the inverter of the inverter on the partial shading
MPPT algorithm Complex Simple Complex Simple
All the If one or more All the All the power is lost if the
power dc-dc power main converter fails and no
is lost converter is lost power loss if the auxiliary
Redundancy if the fails only if the converter fails and there
dc-dc part of the main is no partial shading.
converter available power converter Otherwise a fraction of the
fails is lost fails available power is lost.

maximum power extraction of the PV arrays, besides being [9] A. Choudhury, P. Pillay, and S.S. Williamson, “DC-Bus voltage balancing
economically feasible. Since the modulation strategy applied algorithm for three-level Neutral-Point-Clamped (NPC) traction inverter
drive with modified virtual space vector,” IEEE Trans. Ind. Appl., vol.
to the NPC inverter does not produce any low frequency 52, pp. 3958-3967, Sep./Oct. 2016
oscillations in the NP voltage, operation of the MPPT units [10] T. Naik, R.G. Wandhare, and V. Agarwal. “Three-level NPC inverter
is not affected by such oscillations. This proposed converter with novel voltage equalization for PV grid interface suitable for partially
shaded conditions,” in Proc. Power and Energy Conf. PECI 2013, pp.
can also replace a standard series MPPT module in multi- 186-193. Feb. 2013.
string inverters to reduce system cost and improve efficiency. [11] S. Essakiappan , H.S. Krishnamoorthy, P. Enjeti, R.S. Balog, and S.
Because of modularity of the proposed auxiliary converter unit, Ahmed, “A new control strategy for megawatt scale multilevel photo-
it can also be added to the NPC inverters of any existing solar voltaic inverters under partial shading,” in Proc. 2nd Int. Symp. on Power
Elect. for Dist. Gen. Systems, Jun. 2012.
PV power plant. [12] J. Young-Hyok, J. Doo-Yong, K. Jun-Gu, K. Jae-Hyung, L. Tae-Won,
and W. Chung-Yuen, “A real maximum power point tracking method
for mismatching compensation in PV array under partially shaded condi-
tions,” IEEE Trans. Power Electron., vol. 26, no. 4, pp. 1001-1009, Apr.
R EFERENCES 2011.
[13] J.D Bastidas-Rodriguez, E. Franco, G. Petrone, C.A Ramos-Paja and
[1] International Energy Agency, “Photovoltaic power systems programme G. Spagnuolo, “Maximum power point tracking architectures for photo-
report: Snapshot of global photovoltaic markets 2016,” Rep. IEA PVPS voltaic systems in mismatching conditions: a review,” IET Pow. Electron.,
T1-31:2017. vol. 7, issue. 6, pp. 1396-1413, Aug. 2014.
[2] S. Kouro, J. I. Leon, D. Vinnikov, and L. G. Franquelo, “Grid-connected [14] A. Elasser, M. Agamy, J. Sabate, R. Steigerwald, R. Fisher, and M.
photovoltaic systems: An overview of recent research and emerging PV Harfman-Todorovic “A comparative study of central and distributed
converter technology,” IEEE Ind. Electron. Mag., vol. 9, no. 1, pp. 47-61, MPPT architectures for megawatt utility and large scale commercial
Mar. 2015. photovoltaic plants,” in Proc. of Ind. Electron. Conf. IECON, pp. 2753-
[3] E. Serban, M. Ordonez, and C. Pondiche, “DC-bus voltage range exten- 2758, Nov. 2010.
sion in 1500V photovoltaic inverters,” IEEE Jour. Emerg. Sel. Topics in [15] N. Foureaux, A. Machado, E. Silva, I. Pires, J. Brito, and F. B. Car-
Power Electr., vol.3 no.4, pp. 901-917, Dec. 2015. doso,“Central inverter topology issues in large-scale photovoltaic power
[4] Xin Hao, Kwok-wai Ma, Jia Zhao, and Xin-Yu Sun, “Design of NPC1 plants: Shading and system losses,” in Proc. IEEE Photovolt. Spec. Conf.,
power stack beyond megawatt for 1500V solar inverter application,” in pp. 4–9, Jun. 2015.
Proc. 3r d Int. Future Energy Electron. and ECCE Asia 2017, pp. 1155- [16] C. Rahmann, V. Vittal, J. Ascui, and J. Haas, “Mitigation control against
1161, Jun. 2017. partial shading effects in large-scale PV power plants,” IEEE Trans.
[5] R. Inzunza, R. Okuyama, T. Tanaka, and M. Kinoshita, “Development of Sustain. Energy, vol. 7, no. 1, pp. 173-180, Jan. 2016.
a 1500VDC photovoltaic inverter for utility-scale PV power plants,” in [17] Y. Shi, R. Li, Y. Xue, and H. Li, “High-frequency-link-based grid-tied
Proc. IEEE 2n d Int. Future Energy Electron. Conf., pp. 1–4, Nov. 2015. PV system with small DC-link capacitor and low-frequency ripple-free
[6] S. Kouro, K. Asfaw, R. Goldman, R. Snow, B. Wu, and J. Rodriguez, maximum power point tracking,” IEEE Trans. Power Electron., vol. 31,
“NPC multilevel multistring topology for large scale grid connected no. 1, pp. 328–339, Jan. 2016.
photovoltaic systems”, in Proc. 2nd IEEE Int. Symp. Power Electr. for [18] E. Serban, F. Paz, and M. Ordonez, “Improved PV inverter operating
Dist. Gen. Systems (PEDG), pp. 400-405, Jun. 2010. range using a miniboost,” IEEE Trans. Power Electron., vol. 32, no. 11,
[7] U.M Choi, F. Blaabjerg, and B.K. Lee, “Control strategy of two capacitor pp. 8470–8485, Nov. 2017.
voltages for separate MPPTs in photovoltaic systems using Neutral-Point- [19] N. Müller; H. Renaudineau; F. Flores-Bahamonde, S. Kouro, and P.
Clamped inverters,” IEEE Trans. Ind. Appl., vol. 51, no. 4, pp. 3295-3303, Wheeler, “Ultracapacitor storage enabled global MPPT for photovoltaic
Jul./Aug. 2015. central inverters,” in Proc. IEEE Int. Symp. on Ind. Electron. (ISIE), pp.
[8] Y. Park, S. K. Sul, C. H. Lin, W. C. Kim, and S. H. Lee, “Asymmetric 1046-1051, Jun. 2017.
control of dc-link voltage for separate MPPTs in three-level inverters,” [20] T. Kerekes, R. Teodorescue, M. Liserre, C. Klumpner, and M. Sumner,
IEEE Trans. Power Electron., vol. 28, no. 6, pp. 2760–2769, Jun. 2013. “Evaluation of three-phase transformerless photovoltaic inverter topolo-
 12

gies,” IEEE Trans. Power Electron., vol. 24, no. 9, pp. 2202–2211, Sep. Salvador Ceballos received the M.S. degree in
2009 physics from the University of Cantabria, Santander,
[21] T. Esram, and P.L. Chapman, “Comparison of photovoltaic array max- Spain, in 2001, and the M.S. and Ph.D. degrees in
imum power point tracking techniques,” IEEE Trans. on Energy Conv., electronic engineering from the University of the
vol. 22, no. 2, pp. 439-449, June 2007. Basque Country, Bilbao, Spain, in 2002 and 2008,
[22] S. Ceballos, J. Rea, I. López, J. Pou, E. Robles, and D. O’Sullivan, respectively. Since 2002 he has been with Tecnalia
“Efficiency optimization in low inertia wells turbine-oscillating water Research and Innovation, Derio, Spain, where he
column devices,” IEEE Trans. on Energy Conv., vol. 8, no. 3, pp. 553-564, is currently a Researcher in the Energy and Envi-
Sep. 2013. ronment Division. From May 2008 to May 2009
[23] J. Pou, J. Zaragoza, P. Rodriguez, S. Ceballos, V.M. Sala, R.P. Burgos he was a Visiting Researcher at the Hydraulic and
and D. Boroyevich, “Fast-processing modulation strategy for the neutral- Maritime Research Centre, University College Cork
point-clamped converter with total elimination of low-frequency voltage (UCC), Cork, Ireland. From November 2014 to May 2015 he was a Visiting
oscillations in the neutral point,” IEEE Trans. Ind. Electron., vol. 54, no. Researcher at the Australian Energy Research Institute, UNSW, Sydney,
4, pp. 2288–2294, Aug. 2007. Australia. He has authored more than 120 published technical papers. His
[24] J. Jaragoza, J. Pou, S. Ceballos, E. Robles, C. Jaen, and M. Corbalan, research interests include multilevel converters, fault-tolerant power electronic
“Voltage-balance compensator for a carrier-based modulation in the topologies, and renewable energy systems.
neutral-point-clamped converter,” IEEE Trans. Ind. Electron., vol. 56, no.
2, pp. 305–314, Feb. 2009.

Josep Pou (S’97–M’03–SM’13–F’17) received the

B.S., M.S., and Ph.D. degrees in electrical engi-
neering from the Technical University of Catalonia
(UPC), in 1989, 1996, and 2002, respectively.
In 1990, he joined the faculty of UPC as an
Assistant Professor, where he became an Associate
Baburaj Karanayil (M’86-SM’09) was born in Professor in 1993. From February 2013 to August
Kerala, India. He received the B.Tech. (Hons) degree 2016, he was a Full Professor with the University of
from the University of Calicut, Kerala, India, the New South Wales (UNSW), Sydney, Australia. He is
M.Tech. degree from Indian Institute of Technol- currently an Associate Professor with the Nanyang
ogy, Bombay, India, and the Ph.D. degree from the Technological University (NTU), Singapore, where
University of New South Wales (UNSW Sydney), he is Program Director of Power Electronics at the Energy Research Institute
Australia, in 1984, 1986, and 2005, respectively, all at NTU (ERI@N) and co-Director of the Electrical Power Systems Integration
in electrical engineering. Lab at NTU (EPSIL@N). From February 2001 to January 2002, and February
From 1986 to 1994, he was a Senior Engineer in 2005 to January 2006, he was a Researcher at the Center for Power Electronics
the power electronics and electrical drives industry Systems, Virginia Tech, Blacksburg. From January 2012 to January 2013,
in India. From 1995 to 2011, he was a Professional he was a Visiting Professor at the Australian Energy Research Institute,
Officer in power electronics and electric drives research in the School of Elec- UNSW, Sydney. He has authored more than 270 published technical papers
trical Engineering and Telecommunications, UNSW Sydney. From February and has been involved in several industrial projects and educational programs
2012 to June 2016 he was a Research Fellow with the Australian Energy in the fields of power electronics and systems. His research interests include
Research Institute, UNSW Sydney. From July 2016 he is a Professional modulation and control of power converters, multilevel converters, renewable
Officer in power electronics and power systems in the School of Electrical energy, energy storage, power quality, HVDC transmission systems, and more-
Engineering and Telecommunications, UNSW Sydney, Australia. electrical aircraft and vessels.
He is Associate Editor of IEEE Transactions on Industrial Electronics and
IEEE Journal of Emerging and Selected Topics in Power Electronics.

View publication stats