IET Power Electronics

Research Article

Solar powered sensorless induction motor ISSN 1755-4535

Received on 23rd June 2017
Revised 15th September 2017
drive with improved efficiency for water Accepted on 8th October 2017
E-First on 22nd November 2017
pumping doi: 10.1049/iet-pel.2017.0452
www.ietdl.org

Saurabh Shukla1 , Bhim Singh1

1Department of Electrical Engineering, Indian Institute of Technology Delhi, New Delhi 110016, India
E-mail: saurabh.shukla.ee@gmail.com

Abstract: This study deals with a speed sensorless induction motor drive (IMD) with efficiency optimisation. This drive is driven
by a solar photovoltaic array for water pumping. The elimination of speed sensor increases the robustness and decreases cost
of the IMD. The speed estimation is achieved by rotor flux-oriented control in stationary reference frame. Moreover, the
parameters, namely stator resistance adaptation, make the system parameters insensitive. The efficiency of the drive system is
enhanced by optimising the excitation current by using a particle swarm optimisation technique. The smooth operation of overall
system during starting and steady-state condition is simulated in MATLAB/Simulink environment and validated on a prototype
developed in the laboratory. The efficiency of the system with loss minimisation technique is compared with the conventional
IMD. Simulated results comply with the test results and a comprehensive comparison is made to validate the suitability of
proposed system.

1 Introduction (SVPWM), which has the merit of 15% greater DC-link voltage
utilisation over sine pulse width modulation technique [16, 17].
An induction motor drive (IMD) is widely used in industries as it Moreover, an attempt is made for even-order harmonics
has been serving various purposes since inception. The properties elimination already reported in [18] from line voltage of SVPWM
of an induction motor being rugged and cost effective, make it inverter as most of the standards, namely IEEE 519-1992, have
reliable to be utilised in industries. The field-oriented control very strict guidelines on even-order harmonics than on odd
(FOC) of an induction motor is the famous technique, which is harmonics [18, 19].
widely used for high-performance application [1]. However, this It is worth mentioning that FOC is not, in itself, sufficient to
technique requires voltage, current and speed sensors. These maximise the efficiency of the motor and hence an optimisation
sensors have to be eliminated to make it robust and less costly. algorithm is required with the control loop. This is known that an
Speed sensorless drive has been widely accepted in industries as IMD has maximum efficiency at full load and it gradually
discussed before because of its robustness and low cost. Various deteriorates as the load is reduced. This is because that the
techniques have been reported for speed estimation [2–5]. This excitation current component (ids*) is kept constant and hence the
paper incorporates the stationary rotor flux component for
core loss (PFe) or no load loss is not minimised. There are mainly
estimation of rotor speed.
Although a number of publications have been reported two techniques reported in the literature for loss minimisation,
describing various techniques of speed sensorless operation of an namely flux optimisation method and input loss minimisation
IMD, very few publications are available on reduced current sensor method [20–23]. The first method depends upon the loss
technique. Verma et.al [6] have described the single current sensor minimisation through modelling of machine, whereas the latter one
scheme for lower speed applications. However, the reduced current depends upon the minimisation of power input for a given speed/
sensor technique on the motor side has not been successfully torque. In this paper, the loss minimisation technique is adopted
implemented in water pumping as the speed requirement for this and for this purpose, the optimisation technique used is based on
purpose, is quite high in order to overcome the static friction to particle swarm optimisation (PSO) [24, 25], which optimises the
start the pump [7]. It is evident that as the irradiance is varied, the magnetising current component to minimise the core loss. In this
solar PV power (Ppv) and output power (Pm) also vary as per the paper, an attempt is made to correlate the motor rated power (Pm)
irradiance. Hence, the PV array requires certain control technique with the solar PV power to use it for MPPT and thus, to develop a
to ensure peak power point tracking. The efficiency of the PV array technique for operating the speed sensorless drive with reduced
depends on the series/parallel configuration for the impedance current sensor on solar PV side [26].
matching [8–11]. This paper is arranged such that subsequent sections deal with
The motor drive is fed from a solar photovoltaic (SPV) array as overall control schematic of an IMD system. It comprises of design
the SPV array is extensively used in irrigation sector in arid or of SPV array feeding speed sensorless drive. The IMD is fed from
semi-arid region of the world. The control of SPV array comprises solar PV array and MPPT is achieved by a three-phase VSI. The
of two-stage and single-stage topologies with and without DC–DC switching of the inverter is controlled by SVPWM technique. The
converter, respectively [12]. Here, a single-stage solar PV array three-phase VSI is also responsible for DC bus voltage regulation.
topology is used for feeding an induction motor. As it is known that The proposed system is designed in order to optimise the
solar Ppv−Vpv characteristics have zero slope at MPP, negative on performance of the drive feeding a water pump in terms of
improved efficiency. The results and discussion shown in the last
right side and positive at left of MPP. The P&O algorithm keeps
section verify the suitability of the drive for the particular
the operating point near MPP to harness maximum energy from a
application.
PV array [13, 14]. For this purpose, perturb and observe maximum
power point tracking (MPPT) algorithm is used, which is most
simple and effective [15]. 2 System configuration
The switching signals of three-phase voltage source inverter Fig. 1a represents the overall schematic of the speed sensorless
(VSI) are generated by space vector pulse width modulation drive, which comprises of single-stage solar powered drive with a

IET Power Electron. 1

© The Institution of Engineering and Technology 2017
Fig. 1  Block diagram of proposed system
(a) Solar powered fed FOC of IMD,
(b) Switching signal generated through SVPWM technique,
(c) SVM technique for even-order harmonics reduction,
(d) Even-order harmonics reduction with and without proposed SVM technique

maximum power point tracking being controlled by three-phase and three-phase VSI control are discussed in next section. The
VSI. The speed is estimated by rotor flux-oriented field-oriented detailed data of the system are given in Appendices 1 and 2.
control (FOC). The implemented SVPWM (as shown in Fig. 1b) is
used for VSI control, which has reduced even-order harmonics in 3.1 SPV array design
line voltage. The sector selection of the implemented space vector
modulation (SVM) technique is represented in Fig. 1c and The detailed design equations of single-stage SPV array given in
explained in later section. The output harmonic spectra for [9] are used and the design specifications of solar PV panel and PV
comparative study with and without implemented SVPWM are array are given in Appendices 1 and 2.
obtained as Fig. 1d. The PSO is used for loss minimisation of the
drive under partial loading conditions. 3.2 DC bus voltage calculation
The DC bus voltage should be greater than the amplitude of the
3 System design phase–phase voltage for proper control of output current and it is
Fig. 1a exhibits a block diagram required for a three-phase given as
induction motor of 2.2 kW (3 HP), 230 V for the pump operation
powered by a 2.5 kW PV array. The design equations of PV array, V dc = 2 × V L = 2 × 230 = 325 V (1)
modelling of system, control technique used for both motor control
Hence, DC bus voltage is selected at 400 V.

2 IET Power Electron.

© The Institution of Engineering and Technology 2017
Fig. 2  MPPT control
(a) Flowchart of perturb and observe algorithm,
(b) Outer control loop for the generation of reference speed

4 Control of the system 4.2 Speed estimation of IMD

The control of the system comprises of MPPT control and DC bus The rotor speed of the motor is estimated by the rotor flux, which
voltage control by a three-phase VSI and motor control by is calculated using DC bus voltage and phase current. These
sensorless FOC with modified SVM switching control. The components are further used for space vector modulation for
detailed descriptions of them are given here. generating switching pulses for an inverter. The step-by-step
equations are as follows.
4.1 Perturb and observe algorithm Two components of VSI voltages (vα, vβ) are calculated from
the DC bus voltage (Vdc) as per expression as
This is the most famous technique, where the PV power and PV
voltage are compared at each iteration from their previous values V dc V dc
and accordingly the voltage is changed to track the maximum vα = × 2Sa − Sb − Sc , vβ = × 3 Sb − Sc (5)
power. The flowchart of this algorithm to estimate the reference 3 3
voltage Vref is given in Fig. 2a. This uses the non-linear property of
where Sa, Sb and Sc are switching logic (set either one or zero) of
the solar PV power and voltage. As soon as the peak power is
tracked, the switching of the VSI is adjusted so that it remains there the three legs of VSI, respectively.
at maximum power point. That is why this tracking algorithm is The three-phase motor currents are converted into two axes
also called hill climbing algorithm. currents (iα, iβ) by the expressions given here
The PV power is directly proportional to the output power
rating of the motor-drive system and since the flux is regulated as 1
iα = ia, iβ = ib − ic (6)
per the insolation variation, and the variation in motor flux can 3
comply with the PV power (Ppv). Therefore, motor rated power
(Pm) can be treated as one of operating variable for MPPT. The stationery components of flux are expressed as
Although the performance is a bit sluggish as the motor operation
is not so prompt; however, there can be some trade-off since it
completely eliminates one current sensor.
ψβ = ∫ vβ − Rs × iβ dt, ψα = ∫ vα − Rs × iα dt (7)
As per Fig. 2b, the reference voltage, Vref, is the output of the
control algorithm and it is compared with the PV voltage, Vpv, to ψ s = ψ α2 + ψ β2 (8)
generate the reference speed (ωref). The mathematical expression
of it is given as follows. Lr′
ψ αr = ψ α − iα × Lls −i L , ψ βr = ψ β − iβ × Lls
The DC bus voltage error is obtained as Lm α lr
(9)
Lr′
V dcl(k) = V ref − V dc(k) (2) − iβLlr
Lm
This error signal Vdcl(k) is the input of the DC bus voltage 2 2
ψ r = ψ αr + ψ βr (10)
proportional-integral (PI) controller and the output of which is
speed signal at the kth instant at sampling rate and is given as
follows: The components of speed signals are obtained to estimate the
motor speed (ωm)
ω1(k) = ω1(k − 1) + Kpdc V dcl(k) − V dcl(k − 1) + KidcV dcl(k) (3)
ψ αriβ − ψ βriα Lm
ωsl = × (11)
Hence, the motor reference speed is estimated as ψr Tr

ωref = ω1 (4) ψ αr × (d/dt) ψ βr − ψ βr × (d/dt) ψ αr

ωsyn = (12)
ψ r2
This reference speed is used for speed control of three-phase IMD.
where ωsl is the slip speed, ωsyn is the synchronous speed,
T r = Lr′ /Rr′, Lr′ is the rotor inductance, Lm is the magnetising

IET Power Electron. 3

© The Institution of Engineering and Technology 2017
inductance, Llr, Lls is the rotor and stator leakage inductances, becomes null as time, t approaches infinity. Hence, the final
respectively. Rr′ is the stator referred rotor resistance and Rs is the observer equation becomes
stator resistance.
d
The estimated motor speed is given as X = AX + BV s + K iss − iss (20)
dt
ωm = ωsyn − ωsl (13)
The positive definite Lyapunov's function is defined as
4.3 Loss minimisation technique of IMD 2
Rs − Rs
The details of the relation between the no load losses (PFe) and the V = eTe + (21)
δ
electromagnetic torque (Te) and motor speed (ωm) are discussed in
[22] and the direct relation is written as The derivative of the above function is written as

T e2 T
PFe = Rd Ids∗ 2 + Rq (14) V̇ = eT A + KC + A + KC e
K ′2Ids∗ 2 2 (22)
Rs − Rs d
+2 R − W Te − eTW
where Rd = Rs + Lm 2
/ Rg + Rr′ ωm
2
, Rq = Rs + RgRr′/ Rg + Rr′ and δ dt s
K′ = (3 × P × Lm)/4. The stator resistance (Rs) is estimated by
ΔRs ΔRs
adaptation mechanism to make the system more robust and W Te = eTW = ie − ie
parameters insensitive. Hence, by minimising (14), the no-load loss ρLs α iα ρLs β iβ
′ ′ (23)
can be minimised. This minimisation is achieved by PSO as Lr Lr
− ΔR i e − ΔR i e
described here. Lm s α ψαr Lm s β ψ βr

4.4 Stator resistance (Rs) adaptation where

The machine equations involved for the stator resistance
eis iss − iss T T
estimation, already reported in the literature [27], are explained = , iss = iα iβ and ψ rs = ψ αr ψ βr
here. These are required to construct the state variables matrices eψr ψ rs − ψ rs

d Lm2
Rr′ + RsLr′2 Lmωm Substituting (23) into (22), the final expression considering only
iα = − iα + × ψ βr the stator current error becomes
dt ρLsLr′2
ρLsLr′
(15)
LmRr′ 1
+ × ψ αr + V T
ρLsLr′2 ρLs α V̇ = eT A + KC + A + KC e
(24)
Rs − Rs d 2ΔRs s
d LmRr′
2
+ RsLr′2 Lmωm +2 R − i e − is e
i = − iβ − × ψ αr δ dt s ρLs ds iα qs iβ
dt β ρLsLr′2 ρLsLr′
(16)
LmRr′ 1 If the gain matrix, K, is chosen such that the whole term [(A + 
+ × ψ βr + V KC)T + (A + KC)] is negative semi-definite, the observer system is
ρLsLr′2 ρLs β
asymptotically stable provided the summation of second and third
terms on the RHS of (24) is equal to zero
d Lr′ Lr′
ψ αr = Vα − R + ρLsS iα (17)
dt Lm Lm s d δ
R = i e −i e (25)
dt s μ α iα β iβ
d Lr′ Lr′
ψ = V − R + ρLsS iβ (18) Considering the change in speed to be fast, a PI controller with
dt βr Lm β Lm s
proper selection of Kp and KI gains, is used for speed adaptation.
By taking system state matrix [A], input matrix [B] and output The final expression is assessed as
matrix [C] from the above non-linear state-space equations and
following similar procedure as above, the error matrix w.r.t. change
in stator resistance (ΔRs), i.e. [W] is given as
Rs = Kp iαeiα − iβeiβ + KI ∫ iαeiα − iβeiβ dt (26)

−ΔRs 4.5 Expression for core-loss resistance (Rg)

0 0 0
ρLs
A no-load test is performed with the rotor rotated at the
−ΔRs iα synchronous speed with a DC machine and the flux is varied by
0 0 0
ρLs iβ applying variable three phase voltages to the induction motor at
ΔAX = W = (19) rated frequency. This test gives accurate results, since the slip is
Lr′ΔRs ψ αr now exactly equal to zero (i.e. rotating at the synchronous speed).
− 0 0 0
Lm ψ βr The curve is plotted and by curve-fitting method, the relationship
Lr′ΔRs between core loss resistance (Rg) and core loss (PFe) is obtained as
0 − 0 0
Lm
Rg = 3 × 10−5PFe
3 2
− 0.02PFe + 5.35PFe (27)
The stability of obtained function is shown with Lyapunov's
stability criterion. The model is based on stator current model in 4.6 FOC of IMD
which, stator d–q axes currents (iα and iβ) are estimated and
The FOC of an induction motor means the control of both the
compared with sensed motor currents. The error signal, ei thus phase and magnitude of phase currents and voltage applied to the
obtained, introduces a corrective signal Kei, so that ei matrix motor. It mainly comprises of three-step control.
The excitation current (Idse*) is obtained as

4 IET Power Electron.

© The Institution of Engineering and Technology 2017
 e∗ ∗ d ∗ 4.8 Particle swarm optimisation
Ids = Ids + Tr × I (28)
dt ds
PSO is used for optimising the excitation current (Ids*). The
During steady-state condition, the second term in the right of the algorithm for the implementation of PSO is given as
above equation becomes null.
The quadrature-axis current (Iqs*) is calculated as follows. Step 1: Start with the objective equation, PFe = RdIds*2 + Rq × 
The error signal of estimated motor speed (ωm) and the Te2/K′2Ids*2.
reference speed (ωref) is the input of speed PI controller and the Step 2: Initialise the random position and velocity for each particle
output is the motor torque (Te(k)*) Iidsk.
Step 3: Calculate the fitness for each of ‘k’ particle positions and
ωerror = ωref − ωm (29) velocities of Idsk.

T e∗(k) = T e∗(k − 1) + K pω ωerror(k) − ωerror(k − 1) + Kiωωerror(k) (30) V Iimr = V Iimr1 V Iimr2 V Iimr3 V Iimr4 … V Iimrk
T

Thus, the quadrature-axis component is calculated as Step 4: Find the best fitness for each particle IdsPbestk and overall
best IdsGbest.
∗ T e∗ Step 5:
Iqs = (31)
K × Ids∗ V Iidsk
+1
= V Iidsk + c1rand1 IdsPbest1 − Idsk + c2rand2 IdsGbest − Idsk .
where k = 1, 2, 3, 4, ..., k.
where K = 3PLm /4Lr′, P is the pole number. Step 6: Update Ids as Ids i+1 i i+1
1 = Ids1 + V Ids1 and update IdsPbestk and
The reference slip speed (ωsl*) is calculated as IdsGbest.
Step 7: Go to Step 2 until the error satisfies the convergence
∗
Iqs criteria.
ωsl∗ = (32)
∗
T r × Ids End.

This speed, along with the motor speed (ωm), is used to estimate 5 Simulated performance
the reference synchronous speed (ωe*) The recommended single-stage solar PV array fed FOC-based IMD
∗ ∗
is modelled and simulated in MATLAB/Simulink environment for
ωe = ωm + ωsl (33) starting, steady-state and dynamic conditions. Figs. 3a–c
demonstrate the detailed Simulink diagrams of proposed system.
The synchronous speed is used to determine the flux-angle (θe) at The proposed system constitutes following blocks:
the kth instant of sampling period as
i. Reference speed estimation block: The reference speed is
θe k = θe k − 1 + ωe∗ × T (34) estimated from DC-link voltage controller using (4).
ii. Iabc/Vabc->α–β block: This block contains the transformation of
where T is the sampling period of the signal. phase currents (isa, isb) and DC-link voltage (Vdc) to currents
The values of Ids and Iqs are obtained by following relations and voltages in stationary (αβ0) domain using (6) and (5),
using sensed currents as respectively.
iii. Stationary flux estimation block: This block constitutes the
Iqs = iβcos θe − iαsin θe (35) calculation of stationary fluxes in stator as well as rotor
reference frame using (7)–(10).
Ids = iβsin θe + iαcos θe (36) iv. Ids/Iqs block transforms the current in stationary frame of
reference to rotational reference frame (dq0 reference frame)
The controller outputs of compared values of d-axis and q-axis using the flux-angle θe obtained from ‘Theta calculation block’
currents, estimated above from reference values are d-axis and q- using (35) and (36).
axis voltages Vds and Vqs, respectively, which are further used for v. Resistance adaptation block: Resistance adaptation is achieved
generating switching signals. using (26).
vi. Slip speed (ωsl), synchronous speed (ωsyn) and finally motor
4.7 SVM technique speed (ωm) are obtained using (11)–(13).
Fig. 1c demonstrates the alternative switching pattern for even- vii Finally, ωref and ωm are used for FOC of the IMD using
order harmonics reduction. These harmonics play significant role . modified SVM switching technique, explained in previous
in a two-level VSI. Its line current THD should act in accordance section.
with the IEEE 519-1992 standard. Hence, an attempt is made to
reduce even-order harmonics by using a new SVM technique. The proposed system is operated under various conditions of
The root cause of even-order harmonics generation is that the solar insolation change. The performance of the drive under
waveform generated by SVM-PWM scheme is not half-wave various conditions are explained here.
symmetrical. As it is known fact that there are mainly two
switching sequences one starting with the [0 0 0] and other with [1 5.1 Performance of PV array during starting and steady-state
1 1]. It has been observed that if the switching pattern is adjusted in conditions
such a way that both the sequences are used alternatively, the even-
order harmonics can be reduced. For this purpose, the assigned Fig. 4a shows the smooth starting and MPPT tracking of array. It is
switching pattern for 1.b–2.a, 3.b–4a and 5.b–6.a start with [1 1 1] observed that PV voltage (Vpv), PV current (Ipv), PV power (Ppv)
and rest of the sections start with [0 0 0]. Fig. 1d shows the and DC bus voltage (Vdc) achieve the desired value, which is equal
harmonic spectra of the motor-drive system operated by to the maximum rating of the solar PV array. The reference speed
conventional SVM technique and improved SVM technique, (ωref) achieves its desired state at 0.1 s. The perturb and observe
respectively, and it is observed that the even-order harmonics are tracking algorithm is used, which is simple and effective tracking
greatly reduced in latter case (although the overall THD remains algorithm as described in earlier section. Fig. 4b represents the
the same). intermediate signals, namely rotor fluxes (ψαr and ψβr) and

IET Power Electron. 5

© The Institution of Engineering and Technology 2017
Fig. 3  Detailed Simulink block diagram
(a) MATLAB model of proposed system,
(b) MATLAB model of FOC with space vector modulation,
(c) MATLAB model of P&O MPPT technique for ωref generation

resultant fluxes (ψr), which are used to estimate motor speed (ωm). and it achieves its steady-state condition instantly. Fig. 5c shows
As depicted from the diagram, the smooth starting of the drive is the behaviour of core losses and change in d-axis and q-axis
achieved. From Fig. 4c, it can be observed that the drive achieves currents. It is clear from the given figure that there is appreciable
its steady-state condition within no time as soon as the MPP is decrease in the core losses (PFe) as the flux is optimised by a well-
tracked. known PSO algorithm. Hence, the overall efficiency of the drive is
increased when the optimisation algorithm is used than the
5.2 Dynamic performance of proposed system conventional FOC IMD.

The dynamic performance of the drive includes the performance of 5.2.2 Performance at increase in irradiance and resistance
the system at variable atmospheric conditions realised here by adaptation: Figs. 6a–c present the desired performances of the
changing the irradiance. The observations at various conditions are
system when irradiance is increased from 500 to 1000 W/m2. The
discussed here.
speed of the drive smoothly increases from 100 to 150 rad/s.
Fig. 6d shows the stator resistance (Rs) adaptation when the
5.2.1 Performance at decrease in irradiance: Fig. 5a
reference resistance is altered to many folds of its rated value. This
represents the dynamic performance of the solar PV array when
condition is taken to show the robustness and adaptability of the
there is sudden decrement in the solar insolation depicted here as
drive to its parameters variation.
the step decrement is in insolation value from 1000 to 500 W/m2. It
is observed in Fig. 5b that the motor drive follows the same pattern
6 IET Power Electron.
© The Institution of Engineering and Technology 2017
Fig. 4  System performance at 1000 W/m2
(a) Solar PV array,
(b) Intermediate signals for speed estimation,
(c) Induction motor

Fig. 5  Performance of proposed system during dynamic change of irradiance decrement from 1000 to 500 W/m2
(a) Solar PV panel,
(b) Induction motor-pump,
(c) Decrease in core losses during insolation decrement

6 Experimental results motor. The irradiance is changed from 1000 to 500 W/m2 to

examine the system under observation. The recorded curves exhibit
The experiment is performed on a prototype developed in the
successful MPPT with the tracking efficiency always greater than
laboratory to validate the suitability of the drive. The specification
98.5%.
of induction motor is given in Appendices 1 and 2. The motor-
drive system consists of a solar PV simulator (AMETEK make
ETS600 × 17DPVF TerraSAS), a VSI (SEMIKRON MD B6CI 6.2 Starting and steady-state performance of the drive
600/415–35F), a voltage sensor (LV-25P) and two current sensors Fig. 8a exhibits the starting and steady-state performances of the
(LA-55P) for DC bus voltage and motor phase currents,
drive at 1000 W/m2. As soon as the PV indices reaches MPP the
respectively, real-time DSP controller (dSPACE 1104) to perform
drive reaches the steady-state condition. The motor reaches 150 
the motor speed control. A four-channel DSO (Digital Storage
rad/s and Vpv−Ipv reach at the MPP soon after the starting. Fig. 8b
Oscilloscope of Agilent Make) is used to record the waveforms.
The transistor (2N2222) circuitry is used to increase the PWM shows experimental results when similar test is performed for
voltage to 15 V required for the gate driver circuit of a three-phase lower irradiance (500 W/m2) and it is observed that the indices
inverter. Opto-couplers 6N136 are used for providing isolation possess steady-state values once the MPP is reached. The motor
between the DSP and gate driver circuit. The intermediate speed or achieves its speed of 100 rad/s with rated motor currents flowing in
flux signals are retrieved by digital-to--analogue (DAC) cable from the respective three windings.
DAC port of DSP-dSPACE 1104. The characteristic of water pump Figs. 9a and b represent the intermediate speed signals, namely
is comprehended with the help of separately excited DC generator slip speed (ωsl), synchronous speed (ωsyn), actual motor speed
with a resistive load, coupled to an induction motor. (ωm) and the reference speed (ωref) obtained from the DC-link
voltage controller. The flux equations, which are used to estimate
6.1 Experimental verification of P&O MPPT these speeds, are given in previous section.
Figs. 7a and b demonstrate the Ppv−Vpv and Ipv−Vpv curves at 1000
and 500 W/m2, respectively. The design value of solar PV array is
(7 A × 328 V = 2300 W) used to feed a 3 hp (2.2 kW) induction

IET Power Electron. 7

© The Institution of Engineering and Technology 2017
Fig. 6  Performance of proposed system during dynamic change of irradiance increment from 500 to 1000 W/m2
(a) Solar PV panel,
(b)Induction motor-pump,
(c)Increase in core losses during insolation increment from 500 to 1000 W/m2,
(d) Stator resistance adaptation as per reference stator resistance (Rs)

Fig. 7  Performance of solar PV array

(a) 1000 W/m2,
(b) 500 W/m2

6.3 Performance during dynamic condition of irradiance Fig. 10b shows the similar test results when the irradiance is
variation increased from 500 to 1000 W/m2. It extends the appropriateness
of the drive.
In order to manifest the suitability of the drive under varying
atmospheric conditions being realised as the change in insolation,
the following observations are drawn. 7 Comparative analysis of conventional and
Fig. 10a demonstrates the performance of the drive when the proposed system
solar insolation is decreased from 1000 to 500 W/m2. As insolation Figs. 11a–d represent the comparative analysis of the motor-drive
decreases, the solar MPP settles at lower operating point. This driven by proposed improved technique and conventional
leads to decreased speed of the drive (ωm). Consequently, the technique. It is observed that the efficiency is improved during
corresponding load applied to an induction motor is decreased partial loading with the improved loss-minimisation technique than
resulting in reduced motor phase current (isa). the drive driven by conventional technique. The irradiance is
varied from 500 to 1000 W/m2 and the core loss and ids are
observed. Contrary to the conventional method where d-axis

8 IET Power Electron.

© The Institution of Engineering and Technology 2017
Fig. 8  System performance during starting
(a) 1000 W/m2,
(b) 500 W/m2

Fig. 9  Intermediate speed signals estimation

(a) 1000 W/m2,
(b) 500 W/m2

Fig. 10  Drive performance under dynamic conditions

(a) 1000–500 W/m2,
(b) 500–1000 W/m2

current (ids) is kept constant irrespective of the loading, in the new core loss and hence reduced efficiency in conventional system than
control technique, when irradiance is not at rated value, the motor the proposed system.
is partially loaded and hence the flux-producing component of
current (ids) is regulated. The corresponding core loss is also 8 Conclusion
regulated and is at lower value during reduced insolation. Although
The proposed motor-drive system used especially for improving
the core loss is the function of system frequency which is changing
efficiency has been modelled and its performance has been
in both the cases as the irradiance is varied, still there is greater
simulated and the suitability is verified through the experimental
results. The single-stage solar PV system is used to feed the motor-
IET Power Electron. 9
© The Institution of Engineering and Technology 2017
Fig. 11  Comparison of proposed and conventional model
(a) Intermediate signals in terms of core loss with proposed system,
(b) Intermediate signals in terms of core loss with conventional system,
(c) Efficiency comparison,
(d) Core loss comparison at different irradiance

driven pump, which curtails the cost of the system and reduces the [3] Bhattacharya, T., Umanand, L.: ‘Improved flux estimation and stator-
resistance adaptation scheme for sensorless control of induction motor’, IET
losses in the system. The SVM inverter is used, which has added Electr. Power Appl., 2006, 153, (6), pp. 911–920
advantage of 15% more DC-link voltage utilisation. The [4] Soto, G.G., Mendes, E., Razek, A.: ‘Reduced-order observers for rotor flux,
implemented SVM technique reduces the even-order harmonics in rotor resistance and speed estimation for vector controlled induction motor
the line voltage of SVM inverter. The parameter adaptability drives using the extended Kalman filter technique’, IET Electr. Power Appl.,
1999, 146, (3), pp. 282–288
increases robustness of the system. The loss minimisation [5] Al-Tayie, J.K., Acarnley, P.P.: ‘Estimation of speed, stator temperature and
technique in the FOC has served the system with improved rotor temperature in cage induction motor drive using the extended Kalman
efficiency. The rated power of the motor-drive is used in MPPT filter algorithm’, IET Electr. Power Appl., 1997, 144, (5), pp. 301–309
instead of the PV power, which optimises the overall cost of the [6] Verma, V., Chakraborty, C., Maiti, S., et al.: ‘Speed sensorless vector
controlled induction motor drive using single current sensor’, IEEE Trans.
system by eliminating one current sensor. It is concluded from the Energy Convers., 2013, 28, (4), pp. 938–950
obtained performances that the SPV array feeding the drive system [7] Kumar, R., Singh, B.: ‘Solar PV powered BLDC motor drive for water
has smooth starting and instant steady-state achieving capability. It pumping using Cuk converter’, IET Electr. Power Appl., 2017, 11, (2), pp.
has wide range of insolation variation and very good efficiency 222–232
[8] Stokes, K., Bigger, J.: ‘Reliability, cost, and performance of PV-powered
because of loss minimisation by PSO technique. The detailed water pumping systems: a survey for electric utilities’, IEEE Trans. Energy
comparative analysis of the proposed system and the conventional Convers., 1993, 8, (3), pp. 506–512
system has evinced the superiority of the proposed system. [9] Singh, B., Kumar, R.: ‘Simple brushless DC motor drive for solar
photovoltaic array fed water pumping system’, IET Power Electron., 2016, 9,
(7), pp. 1487–1495
9 Acknowledgment [10] Jain, S., Thopukara, A.K., Karampuri, R., et al.: ‘A single-stage photovoltaic
system for a dual-inverter-fed open-end winding induction motor drive for
The authors acknowledge to Department of Science and pumping applications’, IEEE Trans. Power Electron., 2015, 30, (9), pp. 4809–
Technology (DST), Government of India for financing this project 4818
under grant no. RP02926. [11] Vitorino, M.A., de Rossiter Correa, M.B., Jacobina, C.B., et al.: ‘An effective
induction motor control for photovoltaic pumping’, IEEE Trans. Ind.
Electron., 2011, 58, (4), pp. 1162–1170
10 References [12] Singh, S., Singh, B.: ‘Solar PV water pumping system with DC link voltage
regulation’, Int. J. Power Electron., 2015, 7, (1/2), pp. 72–85
[1] Sahoo, S.K., Bhattacharya, T.: ‘Field weakening strategy for a vector- [13] Yahyaoui, I., Nafaa, J., Charfi, S., et al.: ‘MPPT techniques for a photovoltaic
controlled induction motor drive near the six-step mode of operation’, IEEE pumping system’. 6th Int. Renewable Energy Congress (IREC), 2015, pp. 1–6
Trans. Power Electron., 2016, 31, (4), pp. 3043–3051 [14] Zhang, S., Xu, Z., Li, Y., et al.: ‘Optimization of MPPT step size in stand-
[2] Smith, A.N., Gadoue, S.M., Finch, J.W.: ‘Improved rotor flux estimation at alone solar pumping systems’. IEEE Power Energy Society General Meeting,
low speeds for torque MRAS-based sensorless induction motor drives’, IEEE June 2006, pp. 1–6
Trans. Energy Convers., 2016, 31, (1), pp. 270–282 [15] Piegari, L., Rizzo, R.: ‘Adaptive perturb and observe algorithm for
photovoltaic maximum power point tracking’, IET Renew. Power Gen., 2010,
4, (4), pp. 317–328

10 IET Power Electron.

© The Institution of Engineering and Technology 2017
[16] Attaianese, C., Perfetto, A., Tomasso, G.: ‘A space vector modulation [26] Metry, M., Shadmand, M.B., Balog, R.S., et al.: ‘MPPT of photovoltaic
algorithm for torque control of inverter fed induction motor drive’, IEEE systems using sensorless current-based model predictive control’, IEEE
Power Eng. Rev., 2002, 22, (5), pp. 65–65 Trans. Ind. Appl., 2017, 53, (2), pp. 1157–1167
[17] Youm, J.H., Kwon, B.H.: ‘An effective software implementation of the space- [27] Orlowska-Kowalska, T., Dybkowski, M.: ‘Stator-current-based MRAS
vector modulation’, IEEE Trans. Ind. Electron., 1999, 46, (4), pp. 866–868 estimator for a wide range speed-sensorless induction-motor drive’, IEEE
[18] Wu, B.: ‘High power converters and AC drives’ (Wiley, New Jersey, 2006) Trans. Ind. Electron., 2010, 57, (4), pp. 1296–1308
[19] Feng, D., Wu, B., Liu, C., et al.: ‘Space vector modulation for high-power
three-level NPC rectifiers without even order harmonics’. IEEE Power
Electronics, Specialists Conf., 2005, pp. 1986–1991 11 Appendix 1. Solar PV array
[20] Findlay, R.D., Stranges, N., MacKay, D.K.: ‘Losses due to rotational flux in
three phase induction motors’, IEEE Trans. Energy Convers., 1994, 9, (3), pp.  
543–549
[21] Dominguez, J.R., Mora-Soto, C., Ortega-Cisneros, S., et al.: ‘Copper and core Voc = 497 V, Vmp = 400 V, Isc = 9.3 A, Impp = 8.7 A, fs = 10 kHz,
loss minimization for induction motors using high-order sliding-mode DC-link capacitor (Cdc) = 600 µF.
control’, IEEE Trans. Ind. Electron., 2012, 59, (7), pp. 2877–2889
[22] Uddin, M.N., Nam, S.W.: ‘New online loss-minimization-based control of an
induction motor drive’, IEEE Trans. Power Electron., 2008, 23, (2), pp. 926– 12 Appendix 2. Specification of induction motor
933
[23] Sridharan, S., Krein, P.T.: ‘Minimization of system-level losses in VSI-based for experimental validation
induction motor drives: offline strategies’, IEEE Trans. Ind. Appl., 2017, 53,
(2), pp. 1096–1105  
[24] Rajendran, S., Srinivasan, H.: ‘Simplified accelerated particle swarm
optimisation algorithm for efficient maximum power point tracking in
2200 W, three phase, 230 V, four poles, Rs = 0.603 Ω, Lls = 
partially shaded photovoltaic systems’, IET Renew. Power Gener., 2016, 10, 0.00293 H, Rr = 0.7 Ω, Llr = 0.00293 H, Lm = 0.07503 H, J = 0.011 
(9), pp. 1340–1347
[25] Saha, S.K., Kar, R., Mandai, T.I., et al.: ‘Optimal linear phase fir high pass
kg m2, voltage controller gains Kpdc = 1.5, Kidc = 0.65.
filter design using PSOCFIWA-WM’. Information and Communication
Technologies (WICT), World Congress, 2012, pp. 768–773

IET Power Electron. 11

© The Institution of Engineering and Technology 2017