Analysis and control of induction generator supplying stand-alone AC loads employing a Matrix Converter

Engineering Science and Technology, an International Journal xxx (2017) xxx–xxx

Contents lists available at ScienceDirect

Engineering Science and Technology, an International Journal journal homepage: www.elsevier.com/locate/jestch

Full Length Article

Analysis and control of induction generator supplying stand-alone AC loads employing a Matrix Converter Sumedha Mahajan a, SenthilKumar Subramaniam a,⇑, Kumaresan Natarajan a, Ammasai Gounden Nanjappa Gounder a, Devendra Varma Borru b a b

National Institute of Technology, Tiruchirappalli, Tamilnadu 620 015, India Vignan’s University, Guntur, Andhra Pradesh, India

a r t i c l e

i n f o

Article history: Received 17 June 2016 Revised 17 January 2017 Accepted 23 February 2017 Available online xxxx Keywords: Capacitor-Excited Induction Generator (CEIG) Matrix Converter (MC) Particle Swarm Optimization (PSO) Space Vector Modulation (SVM) Steady-state analysis Wind energy

a b s t r a c t This paper proposes a Capacitor Excited Induction Generator (CEIG)-Matrix Converter (MC) system for feeding stand-alone AC loads. The variable output voltage magnitude and frequency from CEIG is converted into a constant voltage magnitude and frequency at the load terminals by controlling MC using Space Vector Modulation (SVM) technique. This single-stage MC is turned up as a good alternative for the proposed system against commonly used AC/DC/AC two stage power converters. The configuration and implementation of the closed-loop control scheme employing dSPACE 1103 real time controller have been fully described in the paper. The proposed closed-loop controller regulates the AC load voltage irrespective of changes in the prime mover speed and load. A method for predetermining the steady-state performance of the proposed system has been developed and described with relevant analytical expressions. The effectiveness of the proposed system is exemplified through simulation results for various operating conditions. The proposed control technique is further validated using an experimental setup developed in the laboratory. Ó 2017 Karabuk University. Publishing services by Elsevier B.V. This is an open access article under the CC BY-NC-ND license (http://creativecommons.org/licenses/by-nc-nd/4.0/).

1. Introduction The stand-alone power generation system is being considered as a feasible alternative to grid supply to power remote areas. This has become essential in growing economies where installing power lines to such locations has become uneconomical. Exploitation of locally available renewable or alternate energy resources such as wind, small hydro and biomass is an added attraction to establish such stand-alone systems. Among these renewable energy sources, wind energy will contribute to the considerable amount of electric power generation if it is installed in moderate/high wind potential areas. Squirrel-cage induction machines are extensively employed for power generation from wind energy due to simple and rugged rotor construction, low cost, almost nil maintenance and generator operation without the need of DC supply [1–2]. To operate squirrel-cage IM as a generator, a suitable value of excitation ⇑ Corresponding author at: Department of Electrical and Electronics Engineering, National Institute of Technology, Tiruchirappalli, Tamilnadu 620 015, India. E-mail addresses: [email protected] (S. Mahajan), [email protected] (S. Subramaniam), [email protected] (K. Natarajan), [email protected] (A.G. Nanjappa Gounder), [email protected] (D.V. Borru). Peer review under responsibility of Karabuk University.

capacitor has to be connected at the stator terminals [3–4]. This configuration of operating the IM as a generator is termed as Capacitor Excited Induction Generator (CEIG). For a given excitation capacitor bank, the output voltage magnitude and frequency of the CEIG vary depending upon the rotor speed and load at the stator terminals. Certain frequency insensitive loads which work safely with a voltage of variable magnitude can use this AC power as such. For applications requiring a controllable DC supply, in the beginning stages of development, conventional three-phase controlled rectifiers were used [5]. For applications requiring a controllable AC supply, many configurations are proposed in the literatures [6–14]. In this context, Ahmed et al. have developed dead beat current controller for a voltage source converter and vector controlled variable speed three-phase squirrel cage induction generator with phase locked loop for regulating AC and DC voltages. In this scheme, a DSP is used for implementation of the close loop control to operate the converter for maintaining constant AC as well as DC voltages [9]. Senthilkumar et al. have developed close loop control scheme for wind-driven induction generator system employing diode bridge rectifier-inverter for supplying stand-alone AC loads. This scheme requires the battery bank to be connected between three-phase Diode Bridge Rectifier (DBR) and the inverter for maintaining

http://dx.doi.org/10.1016/j.jestch.2017.02.006 2215-0986/Ó 2017 Karabuk University. Publishing services by Elsevier B.V. This is an open access article under the CC BY-NC-ND license (http://creativecommons.org/licenses/by-nc-nd/4.0/).

Please cite this article in press as: S. Mahajan et al., Analysis and control of induction generator supplying stand-alone AC loads employing a Matrix Converter, Eng. Sci. Tech., Int. J. (2017), http://dx.doi.org/10.1016/j.jestch.2017.02.006

2

S. Mahajan et al. / Engineering Science and Technology, an International Journal xxx (2017) xxx–xxx

Nomenclature a b C q E

per unit frequency = fg/fr per unit speed = N/Ns per phase excitation capacitance, F voltage transfer ratio per phase air-gap voltage, V vAB, vBC, vCA Instantaneous line voltages at the input terminals of the MC, V vab, vbc , vca Instantaneous line voltages at the output terminals of the MC, V VML,VMP rms line and phase voltage at the matrix converter output terminals, respectively, V Is line current of stator, A Pg generator output power, W PL AC load power, W

power balance [10]. Jayaramaiah and Fernandes have proposed voltage and frequency controller for SEIGs with one hysteresis current controller and two PI controllers. Current controlled voltage source inverter is employed in this scheme and this scheme requires a start-up battery [11]. In all these schemes for supplying stand-alone AC loads from CEIG, two/three power converters stages are employed along with energy storage elements such as inductor and electrolytic capacitor. With the advent of power electronic technology and fast acting digital controllers, researchers have shown tremendous interest in developing new power electronic topologies for improved performance [15–18]. In this context, MC is proposed in energy converting systems due to compact design and elimination of bulky capacitor and inductor. Considering these advantages, an attempt has been made in this paper by employing MC to supply standalone AC loads from a variable voltage source i.e. output of CEIG. MC is turned up as a good alternative for the proposed system against commonly used AC/DC/AC two-stage power converters. Independent control of the output voltage magnitude, frequency and phase angle as well as the input power factor is the added advantage of employing the MC for the proposed wind energy conversion system. The closed-loop control scheme for maintaining the required load voltage magnitude and frequency has been developed for the proposed system and implemented using a dSPACE 1103 real time controller board. The space vector pulse width modulation technique is employed to generate modulating functions as well as gating signals to the power switches in the MC. The other objective of the proposed controller is to maintain unity displacement power factor at the MC input terminals. This reduces the burden on the excitation capacitor by supplying only the reactive power needed for the CEIG. The procedure for the predetermination of the performance of the proposed CEIG-MC system has been developed employing Particle Swarm Optimization (PSO) technique. The simulation results using Matlab/Simulink model of the CEIG-MC system along with the experimental ones under various operating conditions are given in the paper to support the usefulness of the present proposal. A detailed configuration of the proposed system, its analysis, derivation of equivalent resistance for representing the MC supplying AC loads at the CEIG terminals are presented in the succeeding sub sections. 2. Proposed CEIG-MC system The complete structure of the proposed system is given in Fig. 1. It consists of a prime mover which drives the CEIG, a MC and direct Space Vector Modulation (SVM) based control scheme. Here, the

R1, X1 R2, X2 Xm Re VgL, VgP Z fg fr RL

per phase resistance and per phase leakage reactance of the stator, respectively, X rotor per phase resistance and per phase leakage reactance (as referred to stator), respectively, X per phase magnetizing reactance, X generator terminal per phase equivalent resistance corresponds to AC loading, X line and phase voltage at the generator terminals, respectively, V per phase equivalent circuit loop impedance, X generator frequency, Hz rated frequency, Hz load resistance per phase, X

output of CEIG is connected to the MC input terminals and the AC load is connected across the MC output terminals. The line voltages and line currents of the stand-alone AC load and voltages of the CEIG are fed to the SVM based closed-loop controller through suitable sensors. The modulation index of the MC is adjusted in a closed-loop to maintain the desired voltage magnitude at the load terminals. The load voltage is properly scaled down and filtered for giving feedback signals to the controller. The entire closed-loop control strategy has been developed by using dSPACE 1103 based real time digital controller board. The steps involved in the calculation of rms value of voltage are given in reference [10]. The error i.e. output of the summer is given to the digital PI controller which gives the modulation index (q), through which voltage at the MC output terminals is maintained constant. The virtually generated unit peak sinusoidal signal (50 Hz) is multiplied by the control signal (i.e., modulation index) to generate the modulating signal. This modulating signal is given to the SVM to generate gating pulses for the IGBTs in MC. The PWM signals from dSPACE 1103 board are amplified by ULN 2003 IC and given to the MC gate driver circuit. 3. Control strategy for matrix converter Matrix converter (MC) consists of m
n bidirectional switches connecting directly the ‘m’ phase power supply to the ‘n’ phase load and providing single stage AC/AC conversion [19–20]. For the proposed system, MC consists of 3
3 bidirectional switches (BDS). Connection of any of the input phases (A, B and C) to any of the output phases (a, b and c) can be made with the appropriate control of these switches shown as in Fig. 1. With the nine BDS, 512 (29) different switching states are possible for the MC. These switching states are reduced to 27 (33) by considering constraints, namely, (i) the input phases should never be shorted and (ii) the output phases should never be opened. Various control strategies are available in the literature for operating MC fed with constant AC voltage source [21–23]. Among various control techniques, SVM algorithm possesses the following advantages as compared to the other modulation techniques: (i) it provides a maximum voltage ratio of 0.866, (ii) it reduces the number of switch commutations in each cycle period, (iii) it is easy to implement, and (iv) it is easy to operate under unbalanced conditions. Further, SVM algorithm has the capability to control input current vectors and output voltage vectors independently [23]. Hence SVM technique has been chosen in the work for controlling MC. In SVM technique, selection of valid MC switching states and calculation of corresponding on-time durations are needed to con-

Please cite this article in press as: S. Mahajan et al., Analysis and control of induction generator supplying stand-alone AC loads employing a Matrix Converter, Eng. Sci. Tech., Int. J. (2017), http://dx.doi.org/10.1016/j.jestch.2017.02.006

3

S. Mahajan et al. / Engineering Science and Technology, an International Journal xxx (2017) xxx–xxx

Matrix converter S11 S12 S13

Capacitor- excited induction generator

A IG B C

PM Prime Mover

Mechanical Coupling

230V RMS value constant f=50Hz

S21 S22

3-phase AC Load

S23

C

i iC iB A

C

S31

vbc

S32

vAB

C

Voltage sensor

S33

Input current sensors

DSPACE based closed-loop controller

Generator frequency estimation

φi = 0

SKY PER 32 PRO R Gate driver circuits

Phase voltage estimation

SVM

Input current vector angle and sector estimation

vab Voltage sensors

ADC

Generation of matrix converter control signal

Digital LPF

Active states order determination

Vrms calculation Voact (RMS)

–+

4 Active and 1 zero switching state selection

PI controller Voltage Controller

Timing calculation Triangular wave

Duty cycle Calculation for active states

Voref 230V

Output voltage vector angle and sector estimation

q

vab vbc foref =50Hz

dSPACE 1103

Fig. 1. Schematic diagram of the proposed stand-alone CEIG-MC system.

struct the required three-phase output voltage. For this, following instantaneous space-vector representations have been used [23]:

 2
2 v AB þ ej2p=3 v BC þ ðej2p=3 Þ v CA ¼ jv i jejai 3

ð1Þ

 2
2 v ab þ ej2p=3 v bc þ ðej2p=3 Þ v ca ¼ jv o jejao Vo ¼ 3

ð2Þ

Vi ¼

where ai, ao are required input and output voltage angles. Using (1) and (2), the valid 27 switching states are classified into three groups. First group forms the six rotating vectors with fixed magnitude and variable angle. These six switching states are not used in implementing SVM technique. Second group forms eighteen rotating vectors with fixed angle and variable magnitude. Magnitude of these switching states depends on the instantaneous values of output line currents and input line voltages. The third group consists of three zero output vectors which determine the zero output voltage and input current vectors. In summary, the 21 active switching states are used out of 27 valid switching states for implementing SVM technique. The working of SVM algorithm is based on four active state selection and suitable time duration for which they are applied

within each switching period Tp. The switching period is completed by applying the zero states to the rest of the duration. The required modulation duty cycles for switching states are obtained using [24] as follows:

dI ¼ ð1Þ

kv þki þ1

dII ¼ ð1Þ

kv þki


   ~i  p ~ o  p3 cos b 3 2 cos a pffiffiffi q cos ui 3


   ~i þ p ~ o  p3 cos b 3 2 cos a pffiffiffi q cos ui 3


   ~i  p ~ o þ p3 cos b cos a 3 2 dIII ¼ ð1Þkv þki pffiffiffi q cos ui 3 kv þki þ1

dIV ¼ ð1Þ


   ~i þ p ~ o þ p3 cos b 3 2 cos a pffiffiffi q cos ui 3

d0 ¼ 1  ðdI þ dII þ dIII þ dIV Þ

ð3Þ

ð4Þ

ð5Þ

ð6Þ ð7Þ

Please cite this article in press as: S. Mahajan et al., Analysis and control of induction generator supplying stand-alone AC loads employing a Matrix Converter, Eng. Sci. Tech., Int. J. (2017), http://dx.doi.org/10.1016/j.jestch.2017.02.006

4

S. Mahajan et al. / Engineering Science and Technology, an International Journal xxx (2017) xxx–xxx

The output voltage sectors are represented by kv = 1,2,. . .,6 and ~i ~ 0 and b the input current sectors are represented by ki = 1,2,. . .,6. a are the output voltage and input current phase angles measured with respect to the bisecting line of the corresponding sector and they differ froma0 and bi according to the output voltage and input current sectors. In these equations, the following angle limits apply:



p 6

~0 6 þ 6a

p 6

and 

p 6

~i 6 þ 6b

p 6

At any sampling instant, the reference quantities, i.e. the input current displacement angle (ui ) and output voltage vector are to be defined. In the proposed control scheme, input current displacement angle is made to zero (ui = 0) so that the unity displacement power factor is maintained at the MC input terminals. 4. Steady-state analysis of the proposed system

f ða; X m Þ ¼ absðZÞ

In the stand-alone operation of CEIG, stator voltage magnitude and frequency will vary with prime mover speed, capacitor value and load parameter. Hence, the steady-state equivalent circuit of three phase induction machine has been suitably modified to include this variable nature of frequency and the resultant equivalent circuit is shown in Fig. 2 [25–26]. In this circuit all the parameters are divided by a so that the magnetisation characteristic (E/a vs Xm), and reactance value of the machine, capacitance and load obtained at rated frequency can be used for the steady-state analysis. For the circuit shown in Fig. 2, the loop impedance, Z can be written as

Z¼

   

 

R el jX C R1 R2 el ll þ þ jX 1 þ jXmll þ jX 2 2 a a ab a

jX1

ð9Þ

where Z is given in (8) The sequence of steps involved in determining the a and Xm using the objective function given in (9) is given in the flowchart of Fig. 3. It is to be noted that the final value of the objective function will be made to zero through PSO technique. Hence, the values, namely, a and Xm obtained will be unique for any operating conditions using this technique. After obtaining the values of a and Xm for a given value of b by PSO technique, the induced emf can be determined from the E/a Vs Xm characteristics of CEIG. Then, using the equivalent circuit shown in Fig. 2, the load voltage and power output from CEIG can be calculated as [26]

ð8Þ

Since no voltage is applied to the stator, only rotor residual magnetism causes emf to build in the CEIG. So, referring to Fig. 2, for any operating point, there will be a load voltage and also voltage across individual circuit components, but the net loop voltage will be zero. Hence, if Z is net complex loop impendence as given in (8) and I is loop current shown in Fig. 2, then the product of I and Z can be equated to zero i.e., IZ = 0. Since the current, I has to be finite value at any steady-state operating point, only the value of Z has to be zero. This is possible because Z encompasses of inductive and capacitive reactances and positive and equivalent negative value {i.e., R2/(a-b)} of resistances. For the circuit shown in Fig. 2, the values of a, b and Xm are unknown and the remaining parameters are known and assumed to be constant for carrying out the steady-state analysis of CEIG. So, the first step in steady-state analysis of CEIG is to determine these unknown parameters for a given operating condition,

jX2

namely, for a given value of b, excitation capacitor and load parameters along with the machine parameters. To determine these unknown variables, the real and reactive parts of (8) are separately equated to zero and then going through a lengthy derivation, a polynomial in p.u. frequency is obtained when p.u. speed (b) is taken as the known parameter [25–26]. Then, this polynomial has to be solved by adopting some numerical methods to obtain a and Xm. Recently, much simpler methods have been proposed by a few authors for the analysis of CEIG by making use of optimization techniques [26–29]. The advantage of such methods is that the loop impedance given in (8) can be taken as it is and solved straightaway without any lengthy derivation or simplifications [26–29]. In this paper PSO technique is employed to arrive at the values of a and Xm for a given value of b, machine, excitation and load parameters by taking loop impedance as the objective function.

( VgP ¼

)1=2

R2L þ X2L

IP ¼ VgP =Re

ð10Þ

E

fðR1 =aÞ þ RL g2 þ ðX1  XL Þ2 and Pe ¼ 3V gP IP

ð11Þ

where

RL ¼

Re X2C a½a2 R2e

þ

X2C 

and XL ¼

R2e XC ½a2 R2 þ X2C 

It can be noted from (8), (10) and (11), the effective value of resistance, Re at the generator terminals is required for carrying out the steady-state analysis. So, the power supplied to the stand-alone AC load through MC should be appropriately reflected at the output terminals of the generator and the procedure for obtaining this equivalence is described in the next sub-section.

R1/a

3

2

1 R2 a−b

jXm

E/a

I

− jX C a2

Re/a

V gP

a

Fig. 2. Equivalent circuit of CEIG with load. All parameters are referred to stator side and reactances at rated frequency I: Loop current, ① Induction generator, ② Excitation capacitance and ③ Equivalent load resistance.

Please cite this article in press as: S. Mahajan et al., Analysis and control of induction generator supplying stand-alone AC loads employing a Matrix Converter, Eng. Sci. Tech., Int. J. (2017), http://dx.doi.org/10.1016/j.jestch.2017.02.006

S. Mahajan et al. / Engineering Science and Technology, an International Journal xxx (2017) xxx–xxx

5

Fig. 3. Flowchart for computing the performance of the proposed system for maintaining a constant AC load voltage at the MC output terminals. VMr: required line voltage at MC output terminals, Pbest: Individual best location.

5. Equivalence of AC load at the CEIG terminals

Re ¼

In the proposed CEIG-MC system, a unity displacement power factor has been maintained at MC input terminals by a closedloop controller. Therefore, the equivalence of AC load at the CEIG terminals can be represented with an equivalent resistance, Re and procedure for this estimation is given below: Let the voltage and the fundamental in-phase component of current at the input terminals of the MC are

v gP

V gP Ip1

ð14Þ

The power delivered by the generator to the MC is

Pg ¼ 3V gP Ip1 ¼ 3

V 2gP Re

ð15Þ

The per phase load voltage at the MC output is

V MP ¼ qV gP

ð16Þ

pffiffiffi ¼ 2V gP sin xt

ð12Þ

pffiffiffi 2Ip1 sin xt

where q is voltage transfer ratio Assuming 10% losses in the MC, the power at the generator terminal can be written as

ð13Þ

Pg ¼ 1:1PL

and

ip1 ¼

Then, the equivalent resistance per phase at the generator terminal is

where load power

ð17Þ PL ¼ 3V MP Io1 cos uL

Please cite this article in press as: S. Mahajan et al., Analysis and control of induction generator supplying stand-alone AC loads employing a Matrix Converter, Eng. Sci. Tech., Int. J. (2017), http://dx.doi.org/10.1016/j.jestch.2017.02.006

6

S. Mahajan et al. / Engineering Science and Technology, an International Journal xxx (2017) xxx–xxx

P L ¼ 3V MP Io1 ¼

For resistive load;

3V 2MP RL

ð18Þ

Substituting Pg from (15) and PL from (18) in (17) the equivalent resistance at the CEIG output can be given as

Re ¼

RL 1:1 q2

ð19Þ

It can be seen from (19) that the voltage transfer ratio, q that would give the required constant load voltage (VMr), would be an unknown quantity at the start of predetermination process. Hence, an iterative procedure has been adopted, till the predetermined load voltage VML becomes equal to the required voltage VMr as given in Fig. 3. In this, the initial value for q is assumed for calculating Re. Then PSO method is used for calculating a, Xm and then p VgP. Then, VML(= 3VMP)is determined and verified whether it is equal to VMr. If it is not equal, then in an outer loop, the voltage transfer ratio is appropriately incremented or decremented and the inner loop calculations are repeated. Thus, in a nested loop of calculations when |VMr  VML| converges to 0.1, the system performance is computed for the given speed and load of the CEIG-MC system. 6. Experimental investigations To demonstrate the working of the proposed system, a 3-phase, 4-pole, 230 V, 50 Hz (1 p.u. frequency), 3.7 kW, delta-connected squirrel-cage induction machine with an excitation capacitor of 100mF per phase has been considered. The parameters of this machine at rated frequency of 50 Hz are R1 = 1.30 X, R2 = 1.75 X, X1 = X2 = 2.6 X. To obtain the magnetization characteristics of the CEIG, it was driven at a constant speed of 1500 rpm (b = 1 p.u. for 4-pole machine at this speed) and the machine was fed from a variable voltage source at a rated frequency of 50 Hz (a = 1 p.u.) [28]. The input current was measured for each input voltage. Under this condition, the equivalent circuit of the induction machine will have only the stator winding components (i.e. R1 + jX1) in series with Xm. Then, the induced emf, E and Xm are calculated as:

E ¼ Vin  Iin ðR1 þ jX1 Þ

Then using the curve fitting, the relationship between (E/a) versus Xm characteristic for the experimental machine has been expressed as the following polynomial:

E=a ¼ 296:35
1010 X6m þ 759:97
108 X5m  784:84
106 X4m þ40:75
103 X3m  111:07
102 X2m þ 12:94Xm þ 245:92 ð22Þ 6.1. Self-excitation process of CEIG It is known that the generator will stay in self-excitation for a given load and speed only within a given range of terminal capacitance i.e., between the minimum and maximum (Cmin and Cmax) values [30–32]. At these two boundary values, the magnetizing reactance of the generator will reach the highest value called critical magnetizing reactance (Xmc). For a given machine, the two values of XC corresponding to Cmin and Cmax can be obtained for a given per unit frequency, a. Then, using the base frequency (i.e., 50 Hz in the present case), the value of capacitances, Cmin and Cmax can be calculated. For the chosen 230 V, 3.7 kW, 4-pole, 50 Hz squirrel-cage induction machine, Cmin and Cmax were calculated as 94 lF and 555 lF respectively. The maximum value of excitation capacitance, Cmax would result in higher level of saturation, which results in overloading the generator due to the excitation capacitor

PL = 1 kW PL= 1.5kW PL= 2kW PL= 2.5kW

Calculated PL= 1kW PL= 1.5kW PL= 2kW PL= 2.5kW

Stator line voltage, V Matrix converter input current, A

ð20Þ

where Vin – per phase applied voltage at the stator terminal and Iin – per phase stator current.

Xm ¼ E=Iin

ð21Þ

Fig. 5. Performance of the CEIG-MC system with variable speed operation. C = 100 mF, VML = 230 V, 50 Hz.

300 Stator Line Voltage, V

150

Stator Line Voltage, V

Voltage axis: 350 V/div

Stator Line Current, A

Current axis: 50 A/div

0 -150 -300 0 20 10 0 -10 -20 0

1

2

3

4

5

6

7

8

9

10

Stator Line Current, A

Time axis: 1sec/div

1

2

3

4

5

6

7

Time, sec

Simulated results

8

9

10

Experimental results

Fig. 4. Voltage built-up process of CEIG. N = 1500 rpm, C = 100 lF per phase.

Please cite this article in press as: S. Mahajan et al., Analysis and control of induction generator supplying stand-alone AC loads employing a Matrix Converter, Eng. Sci. Tech., Int. J. (2017), http://dx.doi.org/10.1016/j.jestch.2017.02.006

S. Mahajan et al. / Engineering Science and Technology, an International Journal xxx (2017) xxx–xxx

current alone. Hence, the excitation capacitance will be chosen close to the lower boundary i.e., Cmin. In the present case, the value of excitation capacitance is chosen as 100 mF per phase so that the machine delivers the rated power at rated speed, which also guarantees the voltage stability of generator. A DC motor was used as a prime-mover to set any required speed of operation. To show the successful voltage build-up of CEIG, with the 100 mF capacitor, generator rotated at 1500 rpm using DC motor and then the excitation capacitor bank was closed. The instantaneous values of the stator voltage and current were recorded using DSO and this experimen-

7

tal result is given in Fig. 4 along with the MATLAB simulated results. 6.2. Development of experimental set-up A prototype model has been built in the laboratory to illustrate the closed-loop operation of the proposed system shown in Fig. 1. The MC has been built using the SEMIKRON make IGBT modules (MD B6CI 800/415-30F) along with the gate driver circuits. A DS1103 PowerPC processor board has been used for developing

Calculated Output power, W Stator line voltage, V Equivalent resistance, Ω

Experimental Output power, W Stator line voltage, V

a

Calculated Frequency, Hz Magnetising reactance, Ω Stator line current, A Experimental Frequency, Hz Stator line current, A

b Fig. 6. Performance of the CEIG-MC system with a constant speed operation. N = 1500 rpm, C = 100 mF, VML = 230 V at 50 Hz. a) Variation of output power, equivalent resistance and stator voltage against matrix converter input current. b) Variation of frequency, magnetizing reactance and stator current against matrix converter input current.

Please cite this article in press as: S. Mahajan et al., Analysis and control of induction generator supplying stand-alone AC loads employing a Matrix Converter, Eng. Sci. Tech., Int. J. (2017), http://dx.doi.org/10.1016/j.jestch.2017.02.006

8

S. Mahajan et al. / Engineering Science and Technology, an International Journal xxx (2017) xxx–xxx

Table 1 Constant speed operation (N = 1500 rpm, C = 100mF). AC load power (PL), W

VML, V

1000

250 230 210 250 230 210

2500

VgP, V

Is, A

Re , X

fg, Hz

Cal.

Exp.

Cal.

Exp.

Cal.

Exp.

Cal.

274 274 274 253 253 253

272 272 272 251 251 251

15 15 15 14 14 14

14 14 14 13 13 13

49.38 49.38 49.38 48.49 48.49 48.49

48.39 48.39 48.39 47.52 47.52 47.52

208 208 208 70 70 70

Cal. – Calculated, Exp. – Experimental.

Stator line voltage, V

Matrix converter input current, A

500V/div

5A/div

Load voltage and its fundamental, V 500V/div

Load current, A

Load voltage (rms), V

5A/div

150V/div

Time axis: 40msec/div

tlc

a

Simulated m late mu

tlc

Experimental

Load current, A 5A/div

Load voltage (L-L), V 500V/div

Time axis 4.6msec//div

Simulation

b

Experimental

Fig. 7. Dynamic response to step change in load C = 100 mF, N = 1500 rpm. a) tlc: instant at which load change initiated from 0.88 kW to 1.3 kW. b) Zoomed waveform for load current and load voltage for PL = 0.88 kW

Please cite this article in press as: S. Mahajan et al., Analysis and control of induction generator supplying stand-alone AC loads employing a Matrix Converter, Eng. Sci. Tech., Int. J. (2017), http://dx.doi.org/10.1016/j.jestch.2017.02.006

S. Mahajan et al. / Engineering Science and Technology, an International Journal xxx (2017) xxx–xxx

the closed-loop controller. The advantage of DS1103 board is to have the real control of the power converters using Matlab/Simulink toolbox. Further, Control Desk software from dSPACE provides instrumentation, parameterization, measurements, and close loop control [33]. The control algorithm given in Fig. 1 has been modelled first in Matlab/Simulink toolbox. Then, this has been compiled using the real time workshop (RTW) available in the Matlab for generating appropriate code suitable for operating the dSPACE 1103 board. LEM make voltage and current transducers are employed for sensing the voltages and currents at various points shown in Fig. 1. These signals are appropriately scaled down before giving it to the dSPACE based closed-loop controller. The output of the controller is of SVPWM pulses and these pulses are amplified by ULN 2003 IC and given to the gate driver circuits of the MC. 6.3. Results and discussions To show the efficacy of the developed controller, experiments have been conducted for various operating conditions and these results are discussed along with predetermined values in the following sub-sections. 6.3.1. Steady-state operation Experiments have been conducted on the proposed CEIG-MC system supplying constant load power by operating the generator with different rotor speeds and the results are given in Fig. 5. The load voltage has been maintained at 230 V ± 2% by dSPACE based closed-loop controller. The calculated values using the predetermination procedure described in Section 4 are also given in the figure. It can be seen from Fig. 5 that the stator voltage reduces with speed for a given load and the operating speed range decreases with increase in load power. To keep that load constant, the input current to the MC increases with the reduction of speed. Generator comes out of excitation, if speed is reduced below certain value and hence constant power cannot be supplied below some value

9

of speed with fixed excitation capacitor. Experiments have also been conducted on the proposed system for supplying variable load power by operating the generator with a constant speed of 1500 rpm. In this experiment also the load voltage is maintained at 230 V and results are furnished in Fig. 6 along with the predicted values. It can be seen from this figure that the stator voltage reduces with increase in load. Despite constant rotor speed, frequency decreases with increase in output power. This shows that the generator frequency of CEIG not only depends on the rotor speed but also depends on the load, excitation capacitor and machine parameters. This figure also gives the variation of load power, equivalent resistance (Re), magnetizing reactance and stator line current with matrix converter input current. The performance of the proposed system has been examined for supplying the constant load power with different set values of AC load voltages for a constant speed operation. Table 1 shows the performance of the system for three different load voltages, namely, 250 V, 230 V and 210 V for two different load settings at 1500 rpm. Interestingly system performance is same for given speed and load power with different load voltages. This is due to the fact that the equivalent load resistance, Re for these settings turns out to be same. For example, for PL = 1 kW, predetermined value of Re is 208 X for any load voltage settings. 6.3.2. Dynamic performance To validate the successful working of the proposed system and closed-loop controller, experiments and simulations have been carried out for certain step change in load, speed and reference output voltages. In each case experimentally recorded waveforms of stator line voltage, MC input current, load voltage (and its fundamental), load current and rms value of load voltage along with the simulated waveforms are presented. (i) Performance without filter Fig. 7 shows the experimental and simulated waveforms for step change in load from 0.88 kW to 1.3 kW. For this experiment,

Fig. 8. Dynamic response to step change in speed from 1300 rpm to 1500 rpm, C = 100 mF, PL = 1.2 kW, tsc: instant at which speed change is initiated.

Please cite this article in press as: S. Mahajan et al., Analysis and control of induction generator supplying stand-alone AC loads employing a Matrix Converter, Eng. Sci. Tech., Int. J. (2017), http://dx.doi.org/10.1016/j.jestch.2017.02.006

10

S. Mahajan et al. / Engineering Science and Technology, an International Journal xxx (2017) xxx–xxx

Stator line voltage, V 500V/div

Matrix converter input current, A 5A/div

Load voltage and its (fundamental), V 500V/div

Load current, A

Load voltage (rms), V

5A/div

150V/div

Time axis: 40msec/div

tcr

tcr

Simulated m late mu

erimen Experimental

Fig. 9. Dynamic response to step change in reference voltage (VML) from 230 V to 200 V, C = 100 mF, N = 1500 rpm, PL = 1.2 kW, tcr: instant at which change in reference voltage initiated.

Input filter

PM MC MC- Mechanical Coupling PM-Prime Mover

A I B IG C

Output filter Matrix converter

C

3-phase AC load

C Controller

Fig. 10. Schematic of the proposed stand-alone CEIG-MC system with filters.

generator was run at 1500 rpm with per phase excitation capacitance of 100 mF, reference voltage and frequency were set at 230 V, 50 Hz respectively. It can be observed from this figure that the stator terminal voltage decreases from 234 V to 229 V with increase in MC input current from 2.4 A to 3.6 A for step change in load. Further, MC supplies 2.2 A at 0.88 kW and it increases to 3.4 A when the load was increased to 1.3 kW and closed-loop controller maintains 230 V, 50 Hz at the load terminals. Fig 8b shows the enlarged view of load voltage and load current waveforms for a load of 0.88 kW. Experiments have been conducted for step change in speed from 1300 to 1500 rpm for a constant load of 1.2 kW and the results are given in Fig. 8 along with the corresponding simulated results. In experimental set up, speed of prime mover i.e. separately excited DC motor was changed by suddenly decreasing its field current. From this figure it can be observed that stator terminal voltage increases from 215 V to 234 V and frequency increases from 42.08 Hz to 49.23 Hz with increase in speed but controller maintains load voltage and frequency constant at set values i.e., 230 V, 50 Hz. MC input current decreases from 3.5 A to 3 A to supply this constant power i.e., 1.2 kW.

To check the effectiveness and robustness of the controller, experiments have been conducted for step change in reference load voltage (rms). Initially load voltage reference was set at 230 V and this value was suddenly changed to 200 V by keeping the rotational speed at 1500 rpm with the load frequency of 50 Hz. From Fig. 9, it can be noted that the closed-loop controller operates successfully for tracking these set values of load voltages with in a cycle and maintains the frequency at 50 Hz for both of the settings. This shows the independent control of voltage magnitude and frequency of the proposed CEIG-MC system. The simulated results for this operation are also given in Fig. 9. (ii) Performance with filter The performance of the system has also been studied with filters as shown in Fig. 10. For simulation, L = 2mH and C = 2 mF have been used for input filter and these values are 10mH and 5 mF for output filter. Simulation has been carried out for different operating conditions and for the sake of brevity, results are given in Fig. 11 for step change in load. It can be noted from this figure that the voltage and currents at various points are sinusoidal as com-

Please cite this article in press as: S. Mahajan et al., Analysis and control of induction generator supplying stand-alone AC loads employing a Matrix Converter, Eng. Sci. Tech., Int. J. (2017), http://dx.doi.org/10.1016/j.jestch.2017.02.006

S. Mahajan et al. / Engineering Science and Technology, an International Journal xxx (2017) xxx–xxx

a

tlc

b

11

tlc

Fig. 11. Dynamic response to step change in load with filters. C = 100 mF, N = 1500 rpm, tlc: instant at which load change is initiated from (a) 0.88 kW to 1.3 kW and (b) 1.3 kW to 0.88 kW.

a

b

Fig. 12. Load current waveform and its Harmonic spectrum (a) without filters and (b) with filters. N = 1500 rpm, C = 100 mF, PL = 1.2 kW and VML = 230 V, 50 Hz.

Please cite this article in press as: S. Mahajan et al., Analysis and control of induction generator supplying stand-alone AC loads employing a Matrix Converter, Eng. Sci. Tech., Int. J. (2017), http://dx.doi.org/10.1016/j.jestch.2017.02.006

12

S. Mahajan et al. / Engineering Science and Technology, an International Journal xxx (2017) xxx–xxx

pared to without filter given in Fig. 7. Fig. 12 shows the load current wave form and corresponding THD with and without filter. So, by using a very small value of filter components, the THD can be reduced to very low value and this value is within the acceptable limit as per the harmonic standards.

7. Conclusions The working of a stand-alone system of CEIG supplying AC loads through Matrix Converter (MC) operated with Space Vector Modulation (SVM) control scheme has been studied. The self-excitation and reactive power support for the IG of the proposed system are obtained by connecting AC capacitors at the stator terminals of the induction machine. A closed-loop controller has been developed employing SVM technique for maintaining the set value of voltage and frequency at the stand-alone load terminals irrespective of rotational speed of induction generator and the value of load. It is simple to set the input power factor of the MC to any desired value with SVM technique. Hence, to have the reduced burden on the excitation capacitor banks, the input displacement power factor of the MC is maintained at unity by this technique. This makes the effective load (i.e., MC along with the load) at the generator terminal as the only resistance (Re). The procedure for the calculation of this equivalence has been described in the paper. The SVM based closed-loop controller has been implemented using dSPACE 1103 real time controller board. A systematic approach has been developed for the analysis of CEIG-MC system supplying stand-alone AC loads. This approach involves steady-state equivalent circuit of CEIG and PSO technique. The proposed system was also simulated using MATLAB/Simulink toolbox for the transient analysis. Simulated results on the various performance quantities along with the predetermined performance characteristics confirm the successful working of the developed controller and usefulness of the proposed system. It is to be noted that the present stand-alone CEIG-MC system will have the voltage harmonics at the AC load terminals and current harmonics at the generator terminals. Hence an appropriate filter circuits are essential for improving the quality of the waveform. These aspects are also discussed in the paper. To show the efficacy of the proposed control strategy and successful working of the proposed system, complete experimental setup was assembled in the laboratory. The prototype system has been tested in various steady-state and transient operating conditions. The results obtained from experimental set-up with a 3phase delta connected, 230 V, 50 Hz, 3.7 kW IG show the effectiveness of the proposed system to regulate the load voltage at the set value irrespective of changes in prime mover speed, and load. A close agreement between the experimental and predetermined characteristics demonstrates the successful working of the system and validates the method developed for the performance predetermination. Acknowledgment Authors wish to thank the authorities of the National Institute of Technology, Tiruchirappalli, India for all the facilities provided for carrying out the experimental and simulation work for the preparation of this paper. References [1] R.C. Bansal, Three-phase self-excited induction generators: an overview, IEEE Trans. Energy Convers. 20 (2) (2005) 292–299. [2] G.K. Singh, Self-excited induction generator research – a survey, Electr. Power Syst. Res. 69 (2) (2004) 107–114.

[3] V. Nayanar, N. Kumaresan, N. Ammasai Gounden, A single sensor based MPPT controller for wind-driven induction generators supplying DC microgrid, IEEE Trans. Power Electron. 31 (2) (2016) 1161–1172. [4] V. Naynar, N. Kumaresan, N. Ammasai Gounden, Wind-driven SEIG supplying DC micro grid through a single-stage power converter, Int. J. Eng. Sci. Tecnol. 19 (2016) 1600–1607. [5] N. Ammasaigounden, M. Subbiah, Microprocessor based voltage controller for wind-driven induction generators, IEEE Transac. Indus. Electron. 37 (6) (1990) 531–537. [6] W.L. Chen, Y.H. Lin, H.S. Gau, C.H. Yu, STATCOM controls for a self-excited induction generator feeding random loads, IEEE Trans. Power Del. 23 (4) (2008) 2207–2215. [7] B. Singh, S.S. Murthy, S. Gupta, Analysis and design of electronic load controller for self-excited induction generators, IEEE Trans. Energy Convers. 21 (1) (2006) 285–293. [8] G.V. Jayaramaiah, B.G. Fernandes, Analysis of voltage and frequency control for grid connected 3Ø self excited induction generator using current controlled voltage source inverter, IEEE Conf. TENCON. C (3) (2004) 468.S–471.S. [9] T. Ahmed, K. Nishida, M. Nakaoka, Advanced control for PWM converter and variable speed induction generator, IET Electr. Power Appl. 1 (2) (2007) 239– 247. [10] S. Senthilkumar, N. Kumaresan, N. Rakesh, K. Vijayakumar, M. Subbiah, WindDriven SEIGs for supplying stand-alone loads employing DSP based power electronic controllers, Wind Eng. 36 (6) (2012) 739–758. [11] G.V. Jayaramaiah, B.G. Fernandes, Novel voltage controller for stand-alone induction generator using PWM-VSI, IEEE Indus. Appl. Conf. 1 (2006) 204–208. [12] Mhamdi. Taoufik, Sbita. Lassad, Experimental stand-alone self-excited induction generator driven by a diesel motor, J. Electr. Syst. Inf. Technol. (2016). doi.org/10.1016/j.jesit.2016.08.005. [13] Mohamed Barara, Chimezie Adiuku, Abdul Rhaiman Beig, Khalifa Hasan Alhosani, Naji Al Sayari, Mohamed Akherraz, Implementation of a DSPACEbased standalone renewable energy supply feeding an isolated load, Int. J. Energy Environ. Eng. 7 (2) (2016) 125–135. [14] K. Idjdarene, D. Rekioua, T. Rekioua, A. Tounzi, Vector control of autonomous induction generator taking saturation effect into account, Energy Convers. Manage. 49 (10) (2008) 2609–2617. [15] A. Iqbal, Sk.M. Ahmed, Haitham Abu-Rub, Space vector PWM technique for a three-to-five-phase matrix converter, IEEE Trans. Indus. Applicat. 48 (2) (2012) 697–707. [16] J. Ebrahimi, E. Babaei, G.B. Gharehpetian, New multilevel converter topology with reduced number of power electronic components, IEEE Trans. Indus. Electron 59 (2) (2012) 655–667. [17] S. Padmanaban, M. Pecht, An isolated/non-isolated novel multilevel inverter configuration for dual three-phase symmetrical/asymmetrical star-winding converter, Eng. Sci. Tech. Int. J. (2016) 1–8, http://dx.doi.org/10.1016/ j.jestch.2016.08.006. [18] L.I. Jose, S. Kouro, F.G. Leopoldo, R. Jose, B. Wu, The essential role and the continuous evolution of modulation techniques for voltage-source inverters in the past, present, and future power electronics, IEEE Trans. Indus. Electron. 63 (5) (2016) 2688–2701. [19] F. Gruson, P. LeMoigne, P. Delarue, A. Videt, X. Cimetiére, M. Arpillière, A simple carrier-based modulation for the SVM of the matrix converter, IEEE Trans. Indus. Informat. 9 (2) (2013) 947–956. [20] J. Rodriguez, E. Silva, F. Blaabjerg, P. Wheeler, J. Clare, J. Pontt, Matrix converter controlled with the direct transfer function approach: analysis, modelling and simulation, Int. J. Electron. 92 (2) (2005) 63–85. [21] L. Helle, B.L. Kim, A.H. Jorgensen, S. Munk-Nielsen, F. Blaabjerg, Evaluation of modulation schemes for three-phase to three-phase matrix converters, IEEE Trans. Indus. Electron. 51 (1) (2004) 158–171. [22] P.W. Wheeler, J.C. Clare, J. Rodriguez, A. Weinstein, L. Empringham, Matrix converters: a technology review, IEEE Trans. Indus. Electron. 49 (2) (2002) 276–288. [23] S.M. Dabour, E.M. Rashad, Analysis and implementation of space-vectormodulated three-phase matrix converter, IET Power Electron. 5 (8) (2014) 1374–1378. [24] D. Casadei, G. Serra, A. Tani, L. Zarri, Matrix converter modulation strategies: a new general approach based on space-vector representation of the switch state, IEEE Transac. Indus. Electron. 49 (2) (2002) 370–381. [25] N. Ammasaigounden, M. Subbiah, M.R. Krishnamurthy, Wind-driven selfexcited pole-changing induction generators, IEE Proc. B. 133 (5) (1986) 315– 332. [26] S. Subramaniam, N. Kumaresan, M. Subbiah, Modelling, analysis and control of stand-alone self excited induction generator-pulse width modulation rectifier systems feeding constant DC voltage applications, IET Gener. Transm. Distrib. 8 (6) (2014) 1140–1155. [27] N. Kumaresan, Analysis and control of three-phase self-excited induction generators supplying single-phase AC and DC loads, IEE Proc. Electr. Power Appl. 152 (3) (2005) 739–747. [28] R. Karthigavial, N. Kumaresan, M. Subbiah, Analysis and control of self-excited induction generator-converter systems for battery charging applications, IET Electr. Power Appl. 5 (2) (2011) 247–257. [29] A.L. Alolah, M.A. Alkanthal, Optimization based steady state analysis of threephase self-excited induction generator, IEEE Trans. Energy Convers. 15 (1) (2000) 61–65.

Please cite this article in press as: S. Mahajan et al., Analysis and control of induction generator supplying stand-alone AC loads employing a Matrix Converter, Eng. Sci. Tech., Int. J. (2017), http://dx.doi.org/10.1016/j.jestch.2017.02.006

S. Mahajan et al. / Engineering Science and Technology, an International Journal xxx (2017) xxx–xxx [30] M. Senthil Kumar, N. Kumaresan, R. Karthigaivel, M. Subbiah, Determination of boundary values of excitation capacitance and minimum load impedance for wind-driven SEIGs, IEEE Conf. ICIAS (2010) 1–6. [31] Li Wang, Jian-Yi Su, Determination of minimum and maximum capacitances of an isolated SEIG using eigenvalue sensitivity approach, Proc. of the International Conference on Power System Technology (POWERCON ’98), 1, 1998, pp. 610–614.

13

[32] N. Kumaresan, Design optimisation and speed extension of wind driven selfexcited induction generators – a new approach, Electric Power Compon. Syst. 32 (2004) 215–228. [33] DS1103 PPC Controller Board Hardware Installation and Configuration, Germany: dSPACE GmbH, 2007, Release 6.0.

Please cite this article in press as: S. Mahajan et al., Analysis and control of induction generator supplying stand-alone AC loads employing a Matrix Converter, Eng. Sci. Tech., Int. J. (2017), http://dx.doi.org/10.1016/j.jestch.2017.02.006