Chapter 5

Modelling and Control Strategy for Single Phase Grid Connected PV

System

5.1 INTRODUCTION

In this chapter modelling and control strategy of single phase grid connected PV system
with efficient MPPT algorithm is proposed. The main focus of this section is to control
the current ripple in the DC link and reduce the capacitor size. Nevertheless, it is
mandatory to illustrate the control operation of PV inverter. Since, it is significant in grid
connected PV systems. Hence, an investigation on current control of single phase PV
inverters and different synchronization techniques is elaborated in this chapter. In order
to avoid the glitches in inverter and to reduce the harmonics, different filters have been
used. Design of LLCL filter with passive damping is done.

5.2 MODEL OF SINGLE PHASE GRID CONNECTED PV SYSTEM

With the help of Matlab/Simulink/PLECS Single phase PV system connected to the grid
is modelled. Simulink model of grid connected PV system is shown in Fig. 5.1. To track
the maximum power from the photovoltaic array, an MPPT algorithm known as perturb
and observe (P&O) is used. In this case, a full bridge inverter is connected to the grid
through the LCL filter as configuration shown in Fig. 5.2. There are several pulse width
modulation techniques that can be applied in full bridge inverter. Nevertheless, a bipolar
modulation technique is chosen due to its low leakage current.

Fig. 5.1 Simulink model of grid connected PV system

70
 Fig. 5.2 Model of full bridge inverter topology

5.3 GRID SYNCHRONIZATION TECHNIQUES

The phase angle of the utility voltage vector is important for emerging distributed
generation systems. Grid synchronization is a significant issue in the connection of power
converters to the grid.

Amplitude and phase angle of fundamental frequency of voltage component are the main
attributes in interconnecting renewable energies to the grid. Many research works depicts
various algorithm for detecting grid voltage phase angle such as zero crossing detection,
arctan function, phase locked loop under ideal grid voltage conditions. However, in the
presence of notches, harmonics and unbalanced conditions, it is necessary to concentrate
on highly robust grid synchronization techniques.

(Kaura and Blasko et al., 1997), (Chung 2000) proposed synchronous rotating frame PLL
(SRF PLL) for unbalanced grid voltage conditions. Later, (Ghartemani and Iravani 2004)
proposed a robust technique of second order generalized integrator (SOGI) to extract the
voltage positive sequence component from αβ reference frame.

71
5.3.1 DIFFERENT CURRENT CONTROL METHOD

When photovoltaic systems are connected to the existing grid using voltage source
converter at the front end, the grid current is normally controlled in order to control active
and reactive power exchange, as discussed by (Blaabjerg et al., 2006). The structure of
current controller plays a significant issue in design of grid connected power converters.
The injected current into the grid should have total harmonic distortion (THD) less than
5%. Under grid faults, the injected current varies from regular operating conditions. In
single phase systems, fault in the grid gives rise to variations in the amplitude and phase
angle of the single phase voltage at the point of connection of the power converter. A
classification of current control methods are depicted in Fig. 5.3. Such as proportional
integral (PI) controllers with complex computational requirements and existence of
steady state error as drawbacks (Zmood et al., 2001). Later, modified PI controller have
come up with an addition of grid voltage feed forward path and multiple-state feedback
and increase of proportional gain as given in equation (6) and (7). Inclusion of these
parameters have made difficult in implementing it in digital signal processor (DSP). In
the next decades, most widely used proportional resonant (PR) controller eliminates
steady state error with an infinite gain at selected resonant frequency (Sato et al., 1998).
Hysteresis controllers (Zhang et al., 2016), repetitive controllers (Zhang et al., 2003,
2016), sliding-mode controllers (Kolar et al., 1991), have come up with new features. The
deadbeat controller is from the hierarchy of predictive controllers (Wu et al., 1991), the
controller is developed on the basis of the model of the filter and grid, which predicts the
dynamics of the system. The basic proportional and integral equations are given below.

PI controller transfer function is described as,

ki
GPI  k p  (5.1)
s

72
PR controller transfer function is described as,
s
GPR  k p  ki 2 (5.2)
s  2

Fig. 5.3 Classification of current control methods

5.3.2 REPETITIVE CONTROL

The most significant is compensation of harmonic disturbances in the recent years. By

attaching a repetitive controller, steady state performance can be achieved easily. At the
point of common coupling, the voltage may become unbalance due to system failure. In
order to get rid of the problem, a closed loop repetitive controller to compensate second
harmonic present in the system as shown in Fig. 5.4 is modelled using discrete fourier
transform (DFT) for compensation of harmonics. Rejection of periodic disturbances is
tracked by the repetitive controller by the periodic reference signals.

73
 Fig. 5.4 Representation of repetitive controller in the Discrete fourier transform (DFT)

Discrete Fourier transform is given by equation (5.3)

2  2 
FDFT 
N
 cos  N
k  Nh
h(i  N a ) Z i

(5.3)

Nh is the selected harmonics, N a is the leading steps of the stability.

With some conventional current control techniques, the proposed current control
techniques is correlated with the important characteristics. PI and PR are well suited for
linear system. However, repetitive, deadbeat and EPLL is suited for non-linear control.
Resonance is high in using integral controllers. Since 90 0 phase shift is necessary
response is slow in integral controllers. The system may be linear or non-linear, the
controllers are executed to the appropriate system according to the functionality.
Commonly, harmonic compensation is required in recent controllers. Repetitive and
EPLL is very good in it. The particular characteristics of the controller decide the system
reliability. Generally the steady state error elimination is good in all controllers. Dynamic
response, peak gain control and harmonic compensation is well suited in dead beat and
EPLL controllers as mentioned in Table 5.1

74
 Table 5.1 Performance comparison of different current control techniques

Proportional Proportional Repetitive Dead beat Enhanced

Integral (PI) resonant control phase
(PR) (RC) locked loop
(EPLL)
System Linear Linear Linear Non-linear Non-linear
Tracking Slow, since Slow Fast Fast Very fast
response 900 phase
shift
Steady state Very good Very good Very good Very good Very good
error response
elimination
Dynamic Fast Very fast Slow Very fast Very fast
response
Harmonic Poor Poor Good Very good Very good
compensation
Peak gain At very low At the most At resonant Very high Along with
frequencies of resonant frequency gain phase and
frequencies control frequency,
amplitude is
also
controlled

75
 Fig. 5.5 Frequency response analysis of repetitive controller

5.3.3 FREQUENCY RESPONSE ANALYSIS

Bode plot of the repetitive scheme as shown in Fig. 5.5 the magnitude increases from 31
dB at the resonant frequencies to -31 dB at the notches. However, if the gain k=0.75 is
lowered, the correlated maximum and minimum magnitudes are 16 dB and -16 dB,
respectively. An additional reduction to 0.5 results in a maximum and minimum
magnitudes of 9 dB and -9 dB, respectively. The robustness of the controller is known as
gain K decreases, the peak amplitude is reduced and the bandwidth increases. Since
repetitive controller is combined with a PI controller with its reference setting the
dynamic of the repetitive controller must be slower than PI controller. As observed in
Fig.5.5 the value of k=0.95 makes the repetitive controller unstable. Moreover k=0.75
makes the system faster than PI. The effectiveness of the controller is verified by
measuring the phase angle synchronization of dq PLL as 1Hz is equal to 6.28 rad/sec in
Fig. 5.6.

Fig. 5.6 Phase angle synchronization of dq PLL

76
5.3.4 SECOND ORDER GENERALIZED INTEGRATOR

The main attributes of interconnecting renewable energy to the grid by power converter
are the amplitude and phase angle of fundamental frequency component of the grid
voltage. However, detection of harmonic component is made similar in single phase grid
synchronization method is classified into the frequency domain and time domain.
In single phase applications, there is no need to generate a signal in quadrature to input.
In orthogonal signal generation methods are classified as PLL structure and inverse park
transformation methods. Time delay, all pass filter, Hilbert transform, Second order
generalized integrator (SOGI) fall under aforementioned category.

The transfer function of the filter is based on SOGI as shown in Fig. 5.7 with two outputs
as sinusoidal v` (s) and qv` (s) with the same amplitude and phase. It clearly states that
stability problem is due to infinite gain at ac frequency and no phase shift. This is made
finite by introducing resonance frequency (ω c) very high gain is achieved, thus being
capable to eliminate steady state error between the controlled signal.

Fig. 5.7 Representation of second order generalized Integrator

Under normal grid condition without any voltage harmonic distortion, SRF PLL with
high bandwidth can give an accurate and fast detection of the phase and amplitude of the

77
grid voltage. If the grid voltage is distorted with harmonics, the SRF PLL can still work if
its bandwidth reduced to cancel out the harmonics with slow process.
From the transfer functions of equation (5.2) and (5.3) bode plots is shown in Fig. 5.8 as
width depends on ki , low ki leads to narrow band and high k i leads to wider band. The
phase shift of 900 is achieved as shown in Fig. 5.8. For a grid connected application, ω
could be appropriately set corresponding to the grid frequency (50Hz) such that the line
current tracks the AC reference.

Fig. 5.8 Frequency response of BPF-SOGI with different integral gain

The SOGI-QSG can be straight forwardly applied to implement a PLL based on in-
quadrature signal generation. The synchronization system combining an ANF and a
conventional single-phase PLL is shown in Fig.5.9. In the adaptive notch filter (ANF)
block, integral is used as feedback to filter and is given to the input of PLL. Frequency
and phase angle of the reference signal generated by voltage controlled oscillator (VCO)
as x= cos(θ’) match with those of input signal V. Thus, signal oscillations at output of
phase detector (PD) are cancelled and input signal phase angle is detected by PLL.

78
 Fig. 5.9 Control structure of ANF- PLL

With the proposed control structure, and with the help of Matlab/Simulation, a 3 kW of
grid connected inverter is simulated. In the input side, boost converter is used to boost up
the voltage from the solar energy production. The MPPT is used to track the maximum
power point, according to the insolation or variable parameters changes. In the Fig. 5.10
it tracks the maximum point according to the changes in the variable parameters. When
the insolation is 1 the maximum current is 7.4 A, and insolation changes to 0.7 the
maximum current identified is 5.2 A with corresponding voltage of 388 V. In Fig. 5.11 it
depicts the enlarge view of PV output current, voltage and power. The robustness of
control structure in converter and MPPT control clearly shows constant PV voltage of
388 V, PV current as 7.9 A and power produced is nearly 2.8 kW. The switching pulse of
single phase full bridge inverter clearly shows exact and uniform pulse duration as in Fig.
5.12. The dc- link voltage is the input to the inverter, it must be constant without any
fluctuations. In Fig. 5.13 it shows constant voltage of 388 V given input to the inverter,
this shows the perfect control architecture. The figure shown in Fig. 5.14 depicts the
inverter output voltage and current as 388 V and 18 A. According to grid standard
policies, the power given to the grid should be harmonics free. The Fig. 5.15 shows the
filtered output current. There are more variations in the waveform before filter and after
filter waveforms.

79
 Fig. 5.10 Snapshot of MPPT tracking

Fig. 5.11 Enlarged view of power output waveform

80
 Fig. 5.12 Pattern of switching pulse

Fig.5.13 Enlarged view of voltage control waveform

81
 Fig. 5.14 Output of inverter current and voltage

Fig.5.15 Enlarged view of filtered output current waveform

82
5.4 PROPOSED INVERTER CURRENT CONTROL STRUCTURE

(Petrone et al., 2008) have discussed that electrolytic capacitor is found to be 30 times as
less reliable component than non-electrolytic capacitor. Accordingly, capacitor size
reduction is most mandatory to achieve high reliability with non-electrolytic film
capacitor.

The current passed through the grid by a power converter has a satisfactory link with
voltages at PCC. In order to avoid the glitches in the control of inverter it should be made
more robust with the accurate parameter analysis. The block diagram in Fig. 5.16 shows
conventional current control loop structure with grid side current feedback control.
Where, GC(s) denotes proportional resonant (PR) and harmonic compensator (HC)
controller, H(s) implies feedback gain of grid injected current, Ginv (s) indicates gain of
inverter and Gvi→i (s) denotes transfer function of filter.
g

Fig. 5.16 Inverter power control loop

5.4.1 DC LINK VOLTAGE CONTROL WITH FEEDBACK LINEARIZATION

(Xie et al., 2013), and (Yang et al., 2012) approached feed forward compensation
technique to smooth the fluctuations of DC – link voltage. (Lal et al., 2013) stated
feedback linearization can control the dc-link. (Lee et al., 2000) applied feedback
linearization in three phase PWM converters, and concluded that the dc-link electrolytic
capacitor can be reduced by fast voltage response.
The proposed inverter current control structure as in Fig. 5.17 shows the combination of
DC link voltage control and feedback linearization. It exhibits the calculation of inverter
current i1 from the duty ratio (D) needs grid voltage (Vg) that is related to i1 by the

83
undetermined grid impedance (Zgrid). To exhibit the undetermined grid impedance, Vg is
treated as a DC parameter. Compared to inverter current i1 that is, the current loop
crossover frequency is much higher than grid frequency. The Fig. 5.17 also shows duty
cycle (D) calculation from Vi using Vg and VDC as defined in equation (5.5). With the
proposed control the respective parameter makes the linear system with reduction in
capacitor ripple current and therefore reducing the size of the capacitor.

Fig. 5.17 Proposed inverter current control structure

The instantaneous current reference is used by the current compensator with the feedback
current to provide a duty ratio to the inverter. The DC bus voltage is fixed at the desired
set point and controlled current is injected into the grid. Here Vi controls the main
inductor current IL. The inverter current fed into the grid is given by equation (5.6).Where
K s represents the current feedback gain. Hence, the loop gain transfer function is
expressed as equation (5.4)

Gloop(s)  Gc (s)ksGp (5.4)

(ig  ig )  Gc ( s)  Vg
D (5.5)
VDC

VDC  D   Vg
ig  (5.6)
Z LLCL

84
5.5 CLASSICAL METHODS AND CONSTRAINTS ON LCL FILTER
DESIGN

Ripples and harmonics in the grid current can be reduced by filters connected at the
output of the inverter (Lee et al., 2000). The first order filter consists of one inductor
connected in series known as L-filter which is most commonly used (Wu et al., 2013). It
does not have a resonance problem as compared to higher order filters. The inductor
achieves reasonable attenuation of the current harmonics. The attenuation of 20
dB/decade is achieved for high frequency PWM converters.

In grid power converters, switching frequencies at intervals 3-20 kHz causes greater
harmonics and disturbs the electric grid. It is therefore essential to attenuate the switching
frequency harmonics and reduce the current ripples (Teodorescu et al., 2011) to fulfil the
standards of the IEEE 1547.2. (Akagi et al., 2004) examined high frequency component
currents would lead to electromagnetic interference (EMI) noise mainly due to the
parasitic capacitance disturbing the behaviour of high frequencies. The resonant
res
frequency should be in the range 100  res  . The switching frequencies between
2
2-15 kHz causes higher order harmonics and disturbances in grid and produces losses
(Liserre et al., 2005). Later, proposed the procedure to design the filter. In the filter
design, harmonic attenuation is required and damping methods are produced by (Wu et
al., 2014). The main objective is to reduce the filter size, Q- factor at the resonance
frequency and total power dissipation in the circuit.

5.5.1 PASSIVE DAMPED LCL FILTER

The advantage of LCL filter will be useful for high power converters. For the reduction
of switching frequency harmonics and to reduce the cost, LCL filters gain more popular
in grid connected PV inverter. (Liserre et al., 2004, 2006) inverter side inductor is to
reduce the resonance by using second order and fourth order digital filter. (Tong et al.,

85
2008; Liserre et al., 2004, 2006) discussed about real and reactive power supplied to the
grid, because reactive current limits is on the capacitor. The method of designing LCL
filters depends on rating of power converters, fundamental and switching frequency as
inputs with an integrated control design of filters. From the Fig. 5.18 (a). the inverter
output inductor L1 , the filter capacitor C f, and the grid-side inductor L2 constitute the
LCL filter of the inverter. Compared to L filter, LCL filter is proved to have a better
harmonic attenuation for the reduction of filter inductance volume which bypass the high
frequency harmonics through the capacitance branch. The scope is to lower the higher
order harmonics on the grid side and to reduce the oscillation effects. Therefore, in LCL
filter design, inductors should be properly designed in observing the current ripple, filter
capacitance and damping of resonance in filters.

Fig. 5.18 Conventional LCL filters a). LCL filter, b) passive damped LCL filter with Rd , c) passive
damped LCL filter with Rd - Cd

Transfer function of LCL filter is given by equation (5.7),

ig ( s)
GVi i g ( s)  

1 s  Rd C f Z LC s  Z LC
2 2 2


Vi ( s) L2 s 2  Rd C f res
2
s  res
2
 (5.7)

86
If Rd is zero then then equation (5.8),

GVi i g ( s) 
ig ( s )


1 s 2  Z LC
2


Vi ( s) L2 s 2  res
2
 (5.8)

Where,

2
Z LC  [ L1C f ]1

2
L2 Z LC
res
2

L

For improved passive damping as in equation (5.9)

GVi i g ( s) 
ig ( s)

Rd Cd s  1
Vi ( s) 
s L1L2 Rd Cd C f s  L1L2 (Cd  C f ) s 2  Rd Cd ( L1  L2 ) s  L1L2
3
 (5.9)

5.5.2 SELECTION OF FILTER CAPACITANCE (C F )

According to IEEE 1547 the primary term for the design of the filter is to reduce the
weight and size of individual filter parameters. (Liserre et al., 2005) determined the
selection of capacitor is between reactive power in C f and L1 . If the capacitance is high,
the reactive power flowing into it is more which leads to current demands in L1 . In the
design consideration large inductance (L1 ) and smaller capacitance (C f) lead to the
voltage drop across the inductor L1 . So, reactive power is chosen 15% of rated power and
from that the capacitance value can be chosen as in equation (5.10).
Pout
Cdclink  (5.10)
Vdc ripple  Vdc  0

Where, Prated is rated power and Vrated is grid RMS voltage.

87
5.5.3 SIGNIFICANCE OF Q-FACTOR ANALYSIS

The importance of damping is to lower the Q-factor at the resonant frequency without
affecting the frequency response at other frequencies. In Fig. 5.19 frequency response
analysis of higher order filter with and without damping is analyzed. The series LC
circuit gives a minimum impedance at resonance while parallel LC circuit gives a
maximum impedance at the resonant frequency. The value of Q-factor reduces in passive
damping at a dominant resonant frequency. The quality factor of Lf-Cf can be expressed
as in equation (5.11)

1 Lf
Q (5.11)
Rf Cf

For reducing the peak resonant, damping resistor is designed with an optimal Q-factor for
the stiff grid condition. Increase of grid inductance can reduce the passive damping effect
and cannot achieve the optimal Q-factor. Concurrently, the total power dissipation in the
damping circuit is also an important parameter.

Fig. 5.19 Frequency response of LCL filter

88
5.6 DESIGN OF MODIFIED LLCL FILTER

Based on the traditional LCL filter, a small inductor is inserted in the branch loop of the
capacitor, composing a series resonant circuit at the switching frequency. It can,
particularly, attenuate the switching-frequency current ripple components much better
than the LCL filter, saves the total inductance and thereby leads to size reduction. The
most convenient passive method is by adding physical resistors connected either in series
or in parallel with inductor or capacitor of the filter. It aims at reducing Q -factor at
dominant resonant frequency. Recently to reduce the inductor size a novel higher order
LLCL filter is proposed as shown in Fig. 5.20 a), b) with transfer function as in equation
5.11.

i g ( s)
Transfer function of vi ( s)
LLCL filter is given as in equation (5.12)

GVi i g ( s) 
ig ( s )

L C s
f f
2

1
Vi ( s) L L C
1 2 f  ( L1  L2 ) L f C f s3  ( L1  L2 ) s  (5.12)

For LLCL,
1
res 
 L1L2 
  Lf c f
 L1  L2 

5.6.1 PASSIVE DAMPED SCHEME OF LLCL FILTER

Modified LLCL filter topology is used to reduce the damping power loss and high
frequency harmonic attenuation as shown in Fig. 5.20 (a), (b). Because of a series
resonant circuit Lf-Cf is at switching frequency, the value of the inductor is much smaller
than that of L1 & L2 . From the Fig. 5.20 (b) passive damped LLCL2 & Fig. 5.18 (c)
improved passive damped LCL, the value of damping parameters Rd and Cd is same.
Most significantly, the total capacitance of (C d+Cf) is encouraged to be less than 5% of

89
apparent reactive power at rated load. From equation (5.12) and (5.13) it is to be noted
that the addition of poles and zeros in a system gives rise to stability issues and care
should be taken in designing the filters.

Fig. 5.20 Various passive damped LLCL filters with Rd and Cd

a). LLCL1 b). LLCL2

5.6.2 COMPARISON OF PASSIVE DAMPED FILTER

Bode plot of the transfer function of equation (5.7) & (5.12) is depicted as in Fig. 5.21
and it is clear that an LLCL filter based grid connected VSI has almost same frequency
response characteristic of LCL filter. In Fig. 5.22 the peak magnitude lies within 20 kHz,
the first resonant peak occurs at 7.6 kHz within half of the switching frequency and the
next peak occurs at 20 kHz. The damping technique minimizes the resonant peak within
the range. And it is worth mentioned that compared to LCL filter, the additional inductor
Lf does not bring any control difficulties, an additional grid inductance L2 is added to
widen the bandwidth.

Transfer function of improved passive damping LLCL1

ig ( s ) s3 A  s 2 B  sC  1
GVi i g ( s)  
Vi ( s) 
s5a  s 4b  s3c  s 2d  se  (5.13)

a  L1L2 L f C f Cd ;

b  (( L1L2  ( L1  L2 ) L f ) Rd Cd C f );

c  L1L2 (Cd  C f   L f C f ( L1  L2 ));

90
d  Rd Cd ( L1  L2 );

e  L1  L2 ;

A  L f C f Cd Rd ;

B  Lf C f ;

C  Rd Cd ;

Transfer function of improved passive damping LLCL2 :

ig ( s)s3 A'  s 2 B'  sC '  1
GVi i g ( s)  

Vi ( s) s 4b'  s3c'  s 2d '  se'  (5.14)

b'  (( L1L2  ( L1  L2 ) L f ) Rd Cd C f );

c'  ( L1L2 (Cd  C f )  L f (C f  Cd )( L1  L2 ));

d '  Rd Cd ( L1  L2 );

e'  ( L1  L2 );

A'  L f C f Cd Rd ;

B'  L f C f ;

C '  Rd Cd ;

91
 Fig. 5.21. Frequency response analysis of LCL and LLCL

Fig. 5.22. Frequency response analysis of damping techniques of LCL and LLCL

92
Fig. 5.23 Sensitivity analysis of parameter variation of Lf (±20%)

Table 5.2 Filter parameters in design

Parameters Case-I Case-II Case-III

Rd Rd - Cd Rd - Cd
damped damped damped
LCL LCL LLCL
L1 1.2mH 1.2mH 1.2mH

R1 0.1Ω 0.1Ω 0.1Ω

L2 1.2mH 1.2mH 0.22mH

R2 0.04Ω 0.04Ω 0.01Ω

Lf - - 32μH

Rf - - 0.2 Ω

Cf - 2μF 2μF

Rd 4Ω 30Ω 16.5Ω

Cd 4μF 2μF 2μF

93
 Fig. 5.24 Frequency response analysis of passive damped LLCL filter

The bode plot of LCL and LLCL filter parameters, as given in Table 5.2 is plotted in Fig.
5.23 with the sensitivity analysis on parameter variation of ±20%. The enlarge view of
magnitude of LLCL filter is also shown. Moreover, case I & case III as in Table 5.2
attenuate the resonance by -40 dB/decade, but in case II it is -60 dB/decade as in Fig.
5.24. The designed LLCL filter attenuates resonant harmonics around the switching
frequency and well suited for the single phase system.

5.7 EXPERIMENTAL AND HARDWARE PROTOTYPE

In order to verify the proposed damping technique and control structure, simulation is
carried out using MATLAB/Simulink software. The parameters are the same as the
designed in Table I. In order to verify the theoretical analysis, a 250 W experimental
prototype with the laboratory setup is constructed as in Fig. 5.25 based on a DSP
(TMS320F28035) controller is constructed. A programmable DC power supply (chroma
62012P-80-60) is used to emulate the renewable energy sources. The grid current is
sensed by current sensor ACS712ELCTR-20A, the type of IGBTs is IRFB4227PBF with
switching frequency (Fs) as 20 kHz. A DC link capacitor of united chemi-con KXG series

94
100 μF is used. Total harmonic distortion (THD) is measured by the Fluke-434-Power
quality analyzer and the waveform is obtained with agilent (MSO). Grid voltage and
frequency are normally 230 V, 50 Hz. The parameter specifications are listed in Table
5.3.

Fig. 5.25 Laboratory setup of grid connected PV system

95
Fig. 5.26 Measurement of PV voltage and current using programmable DC power supply

Fig. 5.27 Experimental steady state DC bus voltage

96
Fig. 5.28 Synchronization of inverter voltage and grid voltage

Fig. 5.29 Simulated inverter output voltage before filtering

97
 Table 5.3 List of parameter specification

S.No. Specification Range

1. PV voltage 35 V
2. PV current 2.198 A
3. Grid voltage (VRMS) 220-240 V
4. Switch: IGBT IRFB4227PBF
5. Controller: DSP TMS320F28035
6. Current sensor: ACS712ELCTR 20 A
7. Filters:
Capacitor- 10 µF
Inductor- 4 mH
8. Grid resistance 0.2 ohm
9. Mixed storage oscilloscope (MSO) 1 GHz, 5 Gsa/s, 4 channel
10. Programmable DC power supply 80 V, 60 A, 1200 W
11. THD maintained in % Less than 4

Fig. 5.30 Experimental waveform of voltage and injected current into grid with LCL filter

The PV power is given as input, it measured with the help of programmable DC power
supply as shown in Fig. 5.26 as given as Vpv =35 V and Ipv =2.198 A. With the feedback

98
linearization control, Fig. 5.27 shows the steady state experimental voltage response of
dc- link voltage of 380 V as input to the inverter. The grid voltage and inverter output
voltage synchronization in Fig. 5.28 shows this as a promising current control structure
with grid voltage and inverter voltage synchronized each other.The waveforms of inverter
output voltage 380 V before filtering is shown in Fig. 5.29. With the design of
conventional LCL filter, the results are validated through MSO and the waveform are
obtained as shown in Fig. 5.30 with the ripples and harmonics present in it.

Fig. 5.31 Experimental waveform of voltage and injected current into grid with LLCL filter, a) with LCL
filter as in case I and b) with LLCL filter as in case III

99
 Table 5.4 Comparison on different cases

Parameters Case I Case II Case III

LCL (Rd LCL (Rd- Cd LLCL
damped) damped)
THD Simulation 2.48 % 3.5 % 2.4 %
Experimental 3.82 % 4.04 % 3.5 %
Damping Calculated 2.51 0.89 0.45
Losses (W) Measured 2.57 0.91 0.52

Fig. 5.32 Experimental synchronized voltage and current

Fig. 5.33 Snapshot of THD measurement

100
To validate the proposed control structure and filter design, experimental results of both
LCL and LLCL filter output waveform is compared in Fig. 5.31.a), b). It shows that the
proposed design of LLCL filter has less distortion in the waveform compared to LCL
filter as the THD is 3.5% with the proposed LLCL filter and 3.82 % with the
conventional LCL filter as mentioned in Table 5.2. The results of the proposed grid
connected system is shown in experimental results as in Fig. 5.32 peak to peak voltage of
164.3 V and grid injected current of 2.2 A. The total harmonic distortion of the grid
current in the laboratary is THD = 3.5% as shown in Fig. 5.33 that meets the IEEE 519
standard by measuring with the power quality analyzer FLUKE-434.

5.7.1 RESULT AND DISCUSSION

From the results it is to be that, increasing grid inductance can reduce the passive
damping effect and increases the losses in the circuit in case I & II in Table 5.2.
However, in case III resonant peak occurs within the switching frequency with limited
magnitude attenuation and current harmonic around it satisfy IEEE 519-1992. The
current harmonic (>=35th) is less than 0.3% of fundamental current by the parameter
drift of Lf in range ±20% both in switching frequency and double of switching frequency.
The proposed improved control strategy enhances the performances by introducing
feedback linearization making linear structure in adding integrators to get fast and zero
tracking errors and reducing the capacitor value as 100 μF.

The total harmonic distortions of the grid side current in the three cases are measured
using simulation and experiments and are listed in Table 5.4. In case III, LLCL filter
(L1 =1.2 mH, L2 =0.22 mH, Lf = 0.032 mH) are used, of which the control performance of
the grid inverter is calculated as 2.4% in simulation and 3.5% in experimental. The
damping loss is calculated and is in Table 5.4. The power losses in the filter are mainly
caused by inverter-side inductor current ripples of both LCL and LLCL filter. The
damping power loss is mainly caused by volume of L2 , Lf and Rd with the calculated and
measured value of LLCL filter as 0.45 W and 0.52 W less than LCL filter.

101
5.8 CONCLUSION

For reducing the resonance in the filter circuit, damping resistors are added with an
additional inductance in LCL filter with improved control strategy is proposed in this
chapter. Comparative damping analysis of LCL and LLCL filter is designed with case
parameters defined. Addition of passive elements and variation in filter parameters does
not lose the stability. A Feedback linearization is used to control the duty ratio of inverter
control for fast dynamic process, by adding integral control to eliminate steady state error
that enhances the performance and shows satisfactory behaviour. The voltage transient is
improved to decrease of current ripple and the size of the capacitor is reduced. The
robustness of the system is verified using simulated and experimental results. The
waveform of output current without any distortion shows that the inverter with proposed
control structure can convert solar power to a high quality power into the utility grid.

102