Research Article| Submit Manuscript

Turkish Journal of Electromechanics & Energy

ISSN-2547-975X

Direct torque control for an autonomous

photovoltaic system with MPPT control
A. Mohammedi1, D. Rekioua1*

1
Laboratoire LTII, Faculté de Technologie, Université de Bejaia, 06000 Bejaia, Algérie

Received: 27 September 2020; Revised: 27 October 2020; Accepted: 27 November 2020; Published: 30 December 2020
Turk J Electrom Energ Vol.: 5 No: 2 Page: 49-56 (2020)
SLOI: http://www.sloi.org/
*Correspondance E-mail: [email protected]

ABSTRACT This work deals with the control of a photovoltaic pumping system. It is based on
an induction motor, and direct torque control (DTC) is applied. The method adopted is based on
the instantaneous space vector theory. By optimal selection of the space voltage vectors in each
sampling period, DTC archives effective control of the stator flux and torque . The most
important advantage of this method is its rotor position free control system. To improve
photovoltaic (PV) system performances, Perturb & Observe maximum power point tracking
(MPPT) algorithm has been used to achieve sufficient power to operate the water pumping system
under variable conditions. Through this work, it is aimed to obtain a simple and low-cost system
for the photovoltaic water pumping systems with rapid tracking of the maximum power. The
proposed system with DTC and MPPT strategy not only improves system performance but also
extends the working hours of solar photovoltaic systems. Simulation of the adopted method is also
presented.
Keywords: Photovoltaic, Water Pumping, Direct Torque Control, Maximum power point tracking
Cite this article: A. Mohammedi, D. Rekioua, Direct torque control for an autonomous photovoltaic
system with MPPT control, Turkish Journal of Electromechanics & Energy, 5(2), 49-56, (2020).

1. INTRODUCTION Presented perturb and observe (P&O) algorithm which is

The output power induced in the photovoltaic based on iterative algorithms to track continuously the
modules depends on climatic conditions [1-3]. To MPP through the current and voltage measurement of the
maximize power, several MPPT algorithms have been PV module. Most control schemes use the P&O
used [4-18]. The most used control consists to act on the technique because it is easy to implement, but the
duty cycle automatically to place the generator at its oscillation problem is unavoidable [10]. The conductance
optimal value whatever the variations of the metrological incremental algorithm named as “IncCond” presented by
conditions or sudden changes in loads. To track the Ghayed et al [11] requires a complex control circuit. The
maximum power point (MPP), a look-up table on a two last strategies have some disadvantages such as; high
microcomputer is employed by Ramli et al [4]. It is based cost, complexity, and instability. Also, intelligent
on the use of a database that includes parameters and data algorithms based on control by MPPT were introduced
such as typical curves of the PV generator for different [12-18]. The fuzzy logic controller (FLC) optimizes the
irradiances and temperatures. In the curve-fitting method, magnitude of the increment to obtain fast and fine
the nonlinear characteristic of the PV generator is tracking [10]. In sliding mode controller (SMC), the
modeled using mathematical equations or numerical control circuit adjusts the duty cycle of the switch control
approximations. These two algorithms have a waveform to obtain MPPT as a function of the evolution
disadvantage that they may require a large memory of the power input at the DC/DC converter. Neuro-fuzzy
capacity, for calculation of the mathematical systems are fuzzy systems based on the theory of neural
formulations and storage of the data. The open-circuit networks algorithm (ANN). The learning technique
voltage photovoltaic generator method is employed by operates according to the local information and produces
Karami et al [5]. It approximates linearly the voltage of only local changes in the fuzzy system origin. If the
PV generator at the MPP to its open-circuit voltage and a MPPT algorithm performances must be optimal, even for
linear dependency between the current at the MPP and a complex implementation, the choice will be relatively
the short-circuit current for the short circuit photovoltaic focused on advanced algorithms. If the implementation
generator method presented by Reisi et al [6]. These simplicity is preferred, the choice will be based on
methods are simple and economical, but they are not able classical algorithms [4-8].
to adapt to changeable environmental conditions [7-9].
Science Literature TM © All rights reserved.
A. Mohammedi, D. Rekioua, T. J. Electrom. Energ. (5)2 49-56 (2020)

The control of induction motors has gained a lot of The studied system is modeled and simulated under
attention both in academia and industry, recently. One of the MATLAB environment, and obtained results by
the most significant developments in this area has been simulation are presented.
the field-oriented control where partial feedback
linearization, together with a proportional-integral. This 2. MODELING OF THE PROPOSED SYSTEM
technique is very useful even though it is very sensitive In the case of a method speed drive system of an
to parameter variations such as rotor time constant and induction motor supplied by a PV generator, two types of
incorrect flux measurement or estimation at low speeds. stand-alone PV systems can be considered [2].
DTC for induction machines (IM) is a robustness control • A structure with two static converters.
method for motor drives. It provides a better solution to • A single static converter structure.
improve motor characteristics and the voltage inverter In this paper, the second structure, which needs only
source [19-20]. In principle, the DTC method is one converter, has been used. In this case, the proposed
performed with the instantaneous space vector theory. system is represented in Figure 1. It consists of an
Thus, the number of space voltage vectors and switching induction motor fed by a photovoltaic generator through
frequency directly influence the performance of the DTC a three-phase voltage inverter controlled by the DTC
control system. technique.
In this work, a DTC-IG system of a photovoltaic
system is proposed. An application has been performed 2.1. Photovoltaic Array Modelling
for water pumping. The P&O MPPT is applied to the 2.1.1. One diode model
studied system which can be considered as a low-cost The equivalent circuit consists of a single diode for
prototype used for photovoltaic water pumping system the phenomena of cell polarization and two resistors
and quick tracking of maximum power. (series and shunt) for the losses is presented in Figure 2.

PV panels Boost Converter DC/AC Inverter

Converter Converter

Sa Sb Sc
IM

Sa Sb Sc

D
Ipv
Sa Sb Sc
MPPT Controller
Teref
Vpv

Direct Torque Control

Vs Is

Fig.1. Proposed studied system

This model is given by the following Equation (1): This model is characterized by a very simple
resolution. It requires only four parameters namely Isc,
I pv  I ph  I d  I Rsh (1) Voc, Vmp, and Imp. The model is as follows [2]:
   V pv  
 exp   1  (2)
I pv  I sc 1  C1 
   

   C 2 .Voc  
 Vm 
 
V  1
C2  
oc 
I (3)
ln(1  m )
I sc
Im Vm
C1  (1  ) exp(  )
I sc C 2 .Voc
Fig. 2. Equivalent circuit (one diode model) [2]
50
A. Mohammedi, D. Rekioua, T. J. Electrom. Energ. (5)2 49-56 (2020)

2.1.2 Two diode model Results of power characteristics are obtained by

In this model, the two diodes represent the PN different mathematical models and those obtained by
junction polarization phenomena. These diodes represent experiment are presented in Figures 4-6. With: Es solar
the recombination of the minority carriers which are irradiation (W/m²) and Ta ambient temperature (°C). It is
located both at the surface of the material and within the noticed a good agreement between experimental and
volume of the material as in Figure 3. simulation characteristics especially in low and moderate
irradiations which validates the mathematical model. The
error on power is much greater for the two diode model.
The one diode model results are closest to the
experimental ones, therefore this model is adopted in
current study.

Fig. 3. The equivalent circuit for two diode model

The following equation is then obtained:

I pv  I ph - I d2  I d2   I Rsh
(4)

With Iph and IRsh maintaining the same expressions as

above. For the recombination currents, we have: (a)
2diodes
  q.(V pv  R .I pv ) 
 I d1  I 01 .exp( s )  1 2,275%

 

A.N s .k.T j 


 
 s pv )  1
q.(V pv R .I ) explicit
 I d2  I 02 .exp(  0,758%

 

2.A.N s .k.T j 
 1diode
(5) 0,108%

The saturation currents are written as follows;

(b)
   Eg 
 I 01  P4 .T j 3 .exp  Fig. 4. Power characteristic at Es=244 W/m2, Ta=24°C,
 
 k.T j  (a)Photovoltaic power, (b) Power error calculation

   Eg 
3  
 I 02  P5 .T j .exp 2.k.T 
  j 
(6)

Where Ns is the number of cells in branched series, Eg

represents the gap energy. The final equation of the
model is written as:

     V pv RRs .I pv  - P04 .T j 3 .exp k.TEg . (7)

 
I pv  P1 .Es . 1  P2 . E s - E s ref  P3 . T j - Tref 
sh  j 
  V pv  Rs .I pv     Eg    V pv  Rs .I pv  
x exp q.   1  P14 .T j 3 .exp .exp q.   1
  A.Ns .k.Tj    2.k.Tj

   2.A.Ns .k.Tj  
   
(a)
Validation has been made using PV panel Siemens
SM110-24 as seen in Table1. 2diodes
explici
Table 1. Parameters of the PV panel [2] 4,3%
3,44%
Parameters Values
Pmpp 110W 1diode
Impp 3.15A 1,29%
Vmpp 35V
Isc 3.45A
Voc 43.5V
αsc 1.4mA/°C (b)
βoc -152mV/°C Fig. 5. Power characteristic at Es=460 W/m2, Ta=29°C,
(a)Photovoltaic power, (b) Power error calculation
51
A. Mohammedi, D. Rekioua, T. J. Electrom. Energ. (5)2 49-56 (2020)

2.1. Machine Modelling

An induction motor is modelled using voltage and flux
equations, which are referred to as a general frame and
are shown as follows:
Stator voltage equations:

 d S
VS  RS I S  dt
(8)
 d S
VS  RS I S 
 dt

(a)
where: I, Isare forstator currents respectively.
2diodes s are for stator flux respectively, Rs is the
5,44%
stator resistance.

 d R d
0  VR  RR I R    R (9)
 dt dt
explicit
1,848%  d R d
1diode
1,32% 0  VR  RR I R    R

 dt dt

where: IR, IR are for  rotor current respectively,

RR are for  rotor flux respectively, Rr is the
(b) rotor resistance.
Fig. 6. Power characteristic at Es=856 W/m2, Ta=32°C,
(a)Photovoltaic power, (b) Power error calculation  S  LS I S  Lm I R
  LS I S  Lm I R
 S (10)
2. 2. MPPT Algorithm 
 R  LR I R  Lm I S
Many methods have been developed to determine the  R  LR I R  Lm I S

MPPT control. In this paper, the P&O method is used due
to its implementation simplicity in photovoltaic water
pumping system. The flowchart of the P&O algorithm is Mechanical equations:
given in Figure 7.
d r
Te  TL  J m .  
dt
Te  P  (  s  I s   s  I s )

where: Te is the electromagnetic torque, TL the load

torque, J m the inertia, P is the pole number and  r
the rotor angular speed.

2. 3. Inverter Modelling
For each possible switching configuration, the output
voltages can be represented in terms of space vector,
according to the following equation:

2  2   4 
VS  VS  jVS  Va  Vb exp  j   Vc exp  j 
3   3   3  (12)

where: Va, Vb, and Vc are phase voltages.

2.4.Pumping System Modelling

The model of the power input (P) is given as [14]:

Fig. 7. Flowchart of the P&O MPPT algorithm
P( h ,Q )  a( h ).Q 3  b( h ).Q 2  c( h ).Q  d ( h ) (13)
The panel voltage is deliberately perturbed, then the
power is compared to the power obtained before to
disturbance. Specifically, if the power panel is increased where: a(h), b(h), c(h), d(h) are the different coefficients:
due to the disturbance, the following disturbance will be
made in the same direction. If the power decreases, the a(h)  a0  a1h1  a2 h2  a3 h3 (14)
1 2 3
new perturbation is made in the opposite direction. b(h)  b0  b1h  b2 h  b3 h (15)

52
A. Mohammedi, D. Rekioua, T. J. Electrom. Energ. (5)2 49-56 (2020)

c(h)  c0  c1h1  c2 h2  c3 h3 (16)

d(h)  d0  d1h  d2 h  d3 h
1 2 3
(17)

Photovoltaic power (W)

where: ai, bi, ci, and di constants.
For P > 0, we have:

F ( Qk 1 ) (18)
Qk  Qk 1 _
F' ( Qk 1 )
where:
3 2
F(Qk 1 )  a Q k 1 bQ k 1  c Qk 1  d  Pa (19) Time (s)
(a)
3. STUDIED SYSTEM CONTROL
Six non-zero voltage vectors and two zero-voltage

Photovoltaic power (W)

vectors can be applied to the machine terminals. The
stator flux can be estimated as [15]:

t
s ( t )   ( Vs  Rsis )dt (20)
0

Since stator resistance Rs is relatively small, Vs ›› Rs is ,

Time (s)
(b)
s ( t )  Vs .T  s ( 0 ) (21)
Fig. 8. Solar irradiance, (a) one day profile, (b) sudden
variation
where  s ( 0 ) is the initial value of stator flux, T the
sampling period The stator voltage vector Vs is selected The stator flux in Figure 9 keeps its reference
using Table 2, where signs errors ET and E are (sref=0.8 Wb) during the two different irradiation
determined with a zero hysteresis band. profiles.
1

E T  Te ref  Te (22) 0.8

E   s ref  s (23) 0.6

Beta-stator flux (Wb)

0.4
 s  (  s )  (  s  )
2 2
(24) 0.2
0
Table 2 provides the associated inverter switching -0.2
states. -0.4
-0.6
Table 2. Switching table for DTC
-0.8
ET E N
-1
1 2 3 4 5 6 -1 -0.5 0 0.5 1
Alfa-stator flux (Wb)
1 V2 V3 V4 V5 V6 V1
Fig. 9. Stator flux trajectory
1 0 V6 V1 V2 V3 V4 V5
It is noticed that the continuous power obtained at the
1 V3 V4 V5 V6 V1 V2
output of the photovoltaic generator in Figure 10 (a) is in
0 0 V5 V6 V1 V2 V3 V4 accordance with the solar irradiation profile. It reaches a
maximum value at the maximum irradiation level. It can
4. SIMULATION RESULTS also be observed that the MPPT algorithm perfectly
A simulation of the proposed system has been carried tracks and locates the point of maximum power for a
out by means of MATLAB/Simulink. Two different solar given irradiance. This implies that the photovoltaic
irradiance variations in Figure 8 have been used for one- generator is operating at its optimal efficiency. The
day profile and sudden variation. The sampling period of electromagnetic torque, which presents significant
the system is about T=100μs. The obtained results are ripples in Figure 10 (b), follows the reference and takes
represented in Figures 9-11. the shape of the solar irradiance. This means that for high
values of irradiance, the machine accelerates by
increasing the water pumped volume.
The stator current in Figure 10 (c) follows the
irradiance profile. It is noticed that the voltage DC bus
remains constant and follows its reference within a small
53
A. Mohammedi, D. Rekioua, T. J. Electrom. Energ. (5)2 49-56 (2020)

range of variation without being affected by the change addition, with the introduction of MPPT algorithm, the
of solar irradiance in Figure10 (d). The water flow rate in mechanical speed Figure 11 (a) increases with the
Figure 10 (e) changes with the intensity of the solar increase of the water flow in Figure 11 (b) during the
irradiation since the motor pump uses a continuously observed period. The same remark can be made for the
variable power, and therefore a variable speed. In sudden variation profiles in Figures 11(b) and 11(c).
Photovoltaic power (W)

Photovoltaic power (W)

Time (s) Time (s)
(a)

Electromagnetic torque
Electromagnetic torque

(N.m)
(N.m)

Time (s) Time (s)

(b)
Stator current (A)
Stator current (A)

8
8
6
6

Courant statorique (A)

4
Courant statorique (A)

4
2
2
0
0
-2
-2
-4
-4
-6
-6
-8
-8 11.94 11.95 11.96 11.97 11.98 11.99 12
11.94 11.95 11.96 11.97 11.98 11.99 12 Temps (s)
Temps (s)

Time (s) Time (s)

(c)
DC voltage (V)
DC voltage (V)

Time (s) Time (s)

(d)
Water flow (m3/s)
Water flow (m3/s)

Time (s) Time (s)

(e)
Fig. 10. System performances under two different solar irradiance (Left: under solar irradiation profile, Right under
sudden variation profile) a) Photovoltaic power, b) Electromagnetic torque, c) Stator curent, d) DC voltage variation with
time, (e) Water flow
54
A. Mohammedi, D. Rekioua, T. J. Electrom. Energ. (5)2 49-56 (2020)

800 6. CONCLUSION
without MPPT
Mechanical power (W)

with MPPT Direct torque control of an induction generator

600 supplied by a photovoltaic system has been proposed in
this paper. The results have shown that the performances
400 obtained are very attractive for the application of this
control in photovoltaic system, especially in water
200 pumping installations. Simulation results showed that the
PV pumping system offers maximum pumping flow rate
0
due to the MPPT (P&O method). The application of DTC
0 2 4 6 8 10 12 14 16 18 20 22 24
Time (s)
in a photovoltaic pumping system maximizes power,
which improves the total efficiency of the system. It was
(a)
x 10
-3 noticed that there is a lot of ripples in the torque; and it
3.5
with MPPT would be interesting to use a modulated hysteresis
3 without MPPT controller, which leads to an important reduction of the
Flow rate (m3/s)

2.5 torque ripples. Besides, with the introduction of power

2 optimization, the algorithm gives better tracking and
allows us to improve the performance of the proposed
1.5
system with a very high intake of water flow.
1

0.5 References
0 [1] S. O. Amrouche, D. Rekioua, T. Rekioua, Overview
0 2 4 6 8 10 12 14 16 18 20 22 24
Time (s) of energy storage in renewable energy systems,
(b) Proceedings of 2015 IEEE International Renewable
700 and Sustainable Energy Conference, IRSEC 2015,
art. no. 7454988, (2016).
Mechanical power (W)

600
[2] D. Rekioua, E. Matagne, Optimization of
500
Photovoltaic Power Systems, Modelization,
400 Simulation and Control, Edition Springer, Green
300
Energy and Technology, 102, (2012).
[3] M. S. Ebrahim, A.M. Sharaf, A.M. Atallah, and A.S.
200 Emarah, An Efficient Controller for Standalone
100 without MPPT Hybrid-PV Powered System, Turkish Journal of
with MPPT
0
Electromechanics & Energy 2(1), 9-15, (2017).
0 2 4 6 8 10 12 14 16 18 20 22 24 [4] M. A. Ramli, S. Twaha, K. Ishaque, A. YAl-Turki,
Time (s)
A review on maximum power point tracking for
(c) photovoltaic systems with and without shading
-3
x 10
3.5 conditions’, Renewable and Sustainable Energy
3
Reviews, 67, 144–159, (2017).
[5] N. Karami, N. Moubayed, R. Outbib, General
Flow rate (m3/s)

2.5 review, and classification of different MPPT

2 Techniques, Renewable and Sustainable Energy
Reviews, 68(1), 1-18, (2017).
1.5
[6] A. R. Reisi, M. H. Moradi, S. Jamasb, Classification
1 and comparison of maximum power point tracking
0.5 with MPPT
techniques for photovoltaic system: a review,
without MPPT Renew. Sustainable Energy Rev. 19 (C), 433–443,
0 (2013).
0 2 4 6 8 10 12 14 16 18 20 22 24
Time (s) [7] D. P. Hohm and M. E. Ropp, Comparative study of
(d) maximum power point tracking algorithms, Progress
Fig.11. System performanceswith and without MPPT, in Photovoltaics: Research and Applications, 11, 47-
(a) Mechanical power during a profile of one day, (b) 62, (2003).
Flow rate during a profile of one day, (c) Mechanical [8] V. Salas, E. Olias, A. Barrado, A. Lazaro, Review of
power during irradiance variation, (d) Flow rate during The Maximum Power Point Tracking Algorithms for
irradiance variation. Stand-Alone Photovoltaic Systems, Solar Energy
Materials & Sol. Cells, 90 (11), 1555–1578 (2006).
It is noticed that the mechanical power increases with [9] S. Necaibia, M. S. Kelaiaia, H. Labar, A. Necaibia,
the introduction of the MPPT, and thus the water flow Implementation of an Improved Incremental
increases. In addition, with the variation of irradiance, Conductance MPPT Control Based Boost Converter
mechanical power and flow rate follow the variations in Photovoltaic Applications, International Journal
with a marked improvement with the introduction of the of Emerging Electric Power Systems, 18 (4), 1-19,
MPPT. (2017).

55
A. Mohammedi, D. Rekioua, T. J. Electrom. Energ. (5)2 49-56 (2020)

[10] R. Cakmak, I. H. Altas, A. M. Sharaf, Modeling of Nomenclature

FLC-Incremental based MPPT using DC-DC boost ANN Neural networks algorithm
converter for a standalone PV system, International DTC Direct Torque Control
Symposium on Innovations in Intelligent Systems FLC Fuzzy logic controller
and Applications, Trabzon, 1-5, (2012) IG Induction generator
[11] M. S. Ghayad, N. M. Badra, A. Y. Abdelaziz, A. M. IncCond Conductance incremental algorithm
Sharaf, M. A. Attia, A New efficient self-adjusting MPP Maximum power point tracking
PSO algorithm to enhance reactive power response P&O Perturb & Observe
of VSC-HVDC system, Turkish Journal of SMC Sliding mode controller
Electromechanics & Energy, 5(1), 28-36, (2020). Es Solar irradiation (W/m²)
[12] S. Bensmail, D. Rekioua, H. Azzi, Study of hybrid Is,Is stator current (A),
photovoltaic/fuel cell system for stand-alone Jm Mechanical inertia (kg.m²),
applications, International Journal of Hydrogen LR, LS Stator and rotor inductance (H),
Energy, 40(39), 13820-13826, (2015). Lm Magnetizing inductance (H),
[13] M. R. Douiri, Particle swarm optimized neuro-fuzzy Ns Number of cells in branched series,
system for photovoltaic power forecasting model, P Pole number
Solar Energy, 91-104, (2019). Rs Stator resistance ()
[14] H. Hamdi, C.B. Regaya, A. Zaafouri, A sliding- Ta Ambient temperature (°C),
neural network control of induction-motor-pump Te Electromagnetic torque (N.m)
supplied by a photovoltaic generator, Protection and TL Load torque (N.m)
Control of Modern Power Systems, 5 (1), art. no. 1, Vs Voltage vector (V),
(2020).   Concordia transformation components
[15] D. Rekioua, F. Zaouche, H. Hassani, T. Rekioua, S. ss stator flux respectively,
Bacha, Modeling and fuzzy logic control of a stand- r Rotor angular speed (rad/s)
alone photovoltaic system with battery storage,
Turkish Journal of Electromechanics & Energy, 4 Biographies
(1), 11-17, (2019). Ahmed Mohammedi received his
[16] S. Abouda, F. Nolet, A. Chaari, N. Essoumbouli, Y. Master and PHD diploma from the
Koubaa, Direct Torque Control DTC of Induction University of Bejaia (Algeria). He is
Motor Used for Piloting a Centrifugal Pump associate Professor at the Electrical
Supplied by Photovoltaic Generator, International Engineering Department-University
Journal of Electrical, Computer, and of Bouira (Algeria) since 2015. His
Communication Engineering, 7(8), 1110-1115, research activities have been
(2013). devoted to several topics:
[17] T. Rekioua, D. Rekioua, Direct torque control photovoltaic systems, modelling, batteries and control in
strategy of permanent magnet synchronous AC machines.
machines 2003 IEEE Bologna PowerTech - E-mail: [email protected]
Conference Proceedings, 2, art. no. 130466 , 861-
866, (2003). Djamila Rekioua received the
[18] P. K. Behera, M. K. Behera, A. K. Sahoo, engineer degree in electrical
Comparative Analysis of Scalar & Vector Control of engineering in 1987, the master’s
Induction Motor Through Modeling & Simulation, degree, in 1993, both from the Ecole
International Journal of Innovative Research, Nationale Polytechnique, Algiers,
Electronics, Instrumentation, and Control Algeria, and the Ph.D. degree in
Engineering, 2(4), 1340-1344, (2014). 2002. Her current research interests
[19] Z. Mokrani, D. Rekioua, T. Rekioua, Control of Fuel include control of AC machines and
Cells-Electric Vehicle Based on Direct Torque electric drives, and renewable energies. She is Professor
Control, Turkish Journal of Electromechanics & at the University of Bejaia (Algeria), head of the Team
Energy, 3(2), 8-14, (2018). Renewable Energy in LTII Laboratory of the University
[20] S. J. Salehi, J. Siahbalaee, Loss minimization of Bejaia.
strategy under direct torque control method using E-mail address: [email protected]
matrix converter fed IPMS Motor, Journal of
Electrical Engineering, 16 (2),135-143, (2016).

56