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

net/publication/328980665

Novel Voltage Balancing Algorithm for Single-phase Cascaded Multilevel

Inverter for Post-module Failure Operation in Solar PV Applications

Article in IET Renewable Power Generation · November 2018

DOI: 10.1049/iet-rpg.2018.5483

CITATIONS READS

7 115

4 authors, including:

Rahman Syed md Meraj

Qatar University Qatar University
35 PUBLICATIONS 74 CITATIONS 37 PUBLICATIONS 123 CITATIONS

SEE PROFILE SEE PROFILE

L. Ben-Brahim
Qatar University
131 PUBLICATIONS 2,003 CITATIONS

SEE PROFILE

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

1MW solar QZS inverter View project

1 - MW PV Power R&D Using SiC based qZS Cascaded Multilevel Inverter and Battery Energy Storage View project

All content following this page was uploaded by Rahman Syed on 21 February 2019.

The user has requested enhancement of the downloaded file.

IET Renewable Power Generation

Research Article

Novel voltage balancing algorithm for single- ISSN 1752-1416

Received on 9th March 2018
Revised 17th October 2018
phase cascaded multilevel inverter for post- Accepted on 14th November 2018
E-First on 13th December 2018
module failure operation in solar photovoltaic doi: 10.1049/iet-rpg.2018.5483
www.ietdl.org

applications
Syed Rahman1, Mohammad Meraj1, Atif Iqbal1 , Lazhar Ben-Brahim1
1Department of Electrical Engineering, Qatar University, Doha, Qatar
E-mail: atif.iqbal@qu.edu.qa

Abstract: Four cascaded quasi-impedance source inverter (qZSI) modules are required for achieving nine-level output voltage
waveform. In case of one module failure, number of levels in output voltage are reduced to seven. This leads to decrease in the
output voltage magnitude and increased THD (higher than conventional seven-level qZSI). This is due to the dominance of
unwanted harmonic component introduced in the harmonic spectrum. To restore voltage magnitude and optimize THD
performance, novel voltage balancing algorithm is proposed. To validate the control algorithm for off-grid and grid connected
system, simulation results of the multilevel qZSI are discussed in two categories: (i) DC voltage source powered multilevel qZSI
for RL load subjected to module failure and (ii) solar powered multilevel qZSI connected to utility grid subjected to module failure
and variable solar irradiation. Pre-fault and post-fault performance of the system with the proposed control algorithm is
discussed for both categories, which validates the effectiveness of the algorithm. Hardware results for proof-of-concept are
discussed for DC voltage source fed cascaded qZSI connected to RL load during pre-fault and post-fault conditions. FPGA
Virtex-5 is used for hardware implementation of the control algorithm. The results validate the improvement in output voltage
both quantitatively and qualitatively.

1 Introduction from solar panel in all conditions [16]. Injection of active and
reactive power from qZSI powered from PV panel is also discussed
With the development of semiconductor technology, it is now in the literature [17]. It is difficult to achieve optimum operation
possible to operate medium/high-power converters at higher with conventional PI control. For achieving this, model predictive
switching frequencies. This impacts the component size, waveform control is implemented [18]. Cascaded multilevel topologies of
quality, cost, and transient response of the system. For medium and quasi-Z-source have been recently developed for integrating PV
large power applications, the rating of the switching components distributed generation plants to the utility grid [19]. For achieving
required is very high which is difficult to realise with a single better performance with active and reactive power control,
switch. To overcome this, multilevel topologies have been integration of cascaded qZSI with thyristor-based components is
developed which can handle higher power while lowering stress on also presented [20].
switching devices, dv/dt, filter requirement, and cost. The Level-shifted pulse width modulation (LS-PWM) and phase-
modularity of multilevel topologies increases the reliability as well shifted pulse width modulation (PS-PWM) are applicable for all
as the efficiency of the overall power conversion process [1–3]. the above-mentioned topologies including the CHB. The basic
The key multilevel topologies available in the literature are: (i) difference between the working operations is in LS-PWM, the
diode-clamped inverter, (ii) capacitor-clamped inverter, (iii) unsymmetrical switching operation occurs between the cascaded
cascaded H-bridge inverters, and (iv) hybrid inverter topologies modules, whereas in PS-PWM, the symmetrical switching
etc. [4–6]. Out of the numerous topologies available, CHB operation happens between cascaded modules. LS-PWM cannot be
topology has been widely used due to its modularity, low applied to the quasi-Z-source inverter because of the non-
component stress, and low cost [7–9]. Symmetric and asymmetric availability of the constant DC voltage of the middle and lower
hybrid multilevel cells have also been proposed [10, 11]. modules. Hence, PS-PWM is implemented for achieving multilevel
Comparison of their performance is also reported. Topology for output with qZSI [21, 22].
achieving higher voltage levels with a reduced number of switches In the literature, reliability aspects of most popular multilevel
and power supplies have also been proposed [12]. inverter topology – cascaded H-bridge – have been discussed. In
Variation in the DC input voltage of multilevel is prone to occur this circuit, switching components are among the least reliable
due to various reasons. When photovoltaic (PV) panels are components in the electric system. The various reasons for the
connected to the input of these modules, the output power from the component failure in inverters have been discussed [23]. This
solar panel is subjected to change due to variations in solar discussion has been extended to post-fault operation in the case of
irradiance, humidity, dust accumulation, and partial or full shading multilevel inverters in electric drives and power system
[13]. The output voltage of the MLI must consist of identical levels applications [24]. The power system application includes
to ensure better total harmonic distortion (THD) performance. STATCOM and renewable power integration to the grid. In the
Recently developed quasi-impedance source inverters (qZSIs) case of bridge failures in STATCOM applications, the DC-link
are a better option for integrating renewable power generated from capacitor voltage reference is increased to meet the deficit caused
PV panels to the utility grid. The unique capability of boosting the by bridge failure [25]. This is possible because there is no active
input voltage with continuous input current gives better power injection from the multilevel inverter into the grid. In the
performance compared to the conventional H-bridge converters. It case of renewable power integration, various methods have been
eliminates the need for the extra DC–DC converter as required suggested to improve the THD performance of the current injected
boosting is possible with the quasi-impedance network [14, 15]. into the grid [26, 27]. In the case of electric drives, bypassing of
Like the conventional DC–DC converter operation, it is possible to faulty bridges and extraction of power from the remaining bridges
operate the qZSI at MPPT to ensure extraction of maximum power
IET Renew. Power Gener., 2019, Vol. 13 Iss. 3, pp. 427-437 427
© The Institution of Engineering and Technology 2018
 improvement in output voltage performance while adopting the
proposed control algorithm.

2 System description
The system consists of four PV panels, four cascaded qZSI
modules connected to a load consisting of a grid and an inductor.
However, as the system is somewhat complex, two steps are used
in the validity of the proposed controller. The first step is proof of
concept (PoC), in which the system is simplified to demonstrate the
feasibility of the design concept. In this step, the system is reduced
to DC power sources instead of the PV panels, and an RL load
instead of the grid. In this step, a simulation and a hardware
implementation were carried out. After the PoC step, the validation
of the proposed design by using the PV panels as voltage sources
and the power harvested is injected to the grid. This step was only
validated in simulation due to the hardware limitations and
regulation in the university lab.
Thus, two combinations of load and source are presented in
Fig. 1a. The two combinations are:

i. Source 1 (PV panel) and load 1 (utility grid): Here, with this
combination of load and source, the system represents solar
panel powered multilevel qZSI system connected to utility
grid. Here, the main aspect is to inject grid current at unity
power factor when the system is subjected to module failure
and variable environmental conditions.
ii. Source 2 (DC voltage source) and load 2 (RL load): Here, with
this combination of load and source, the system represents DC
voltage source powered multilevel qZSI system connected to
RL load. Here, the main aspect is to control the load current.

Bypass mechanism consists of two current bidirectional relays –

one series relay Ssn (normally open) and one parallel relay Spn
(normally closed), where ‘n’ denotes the module number to which
the relay is connected. In the case of module failure, the faulty
module will be removed by activating the bypass mechanism. The
Fig. 1 Nine - Level qZSI connected to load with bypass mechanism phase-shifted PWM technique is used for achieving multilevel
provided with every module operation. For achieving the desired levels, the phase shift required
(a) Nine-level qZSI connected to load (RL load/utility grid) provided with bypass between the carrier signals is 180°/N (where N is the number of
switches, modules in multilevel inverter).
(b) qZSI module consisting of source (DC voltage source/PV panel), quasi-network
followed by H-bridge inverter
2.1 DC-bus voltage requirement
has also been discussed. Shifting of neutral point for three-phase Equation relating the DC-bus voltage and AC output voltage for
system to operate with a reduced number of bridges has also been multilevel inverter is given by
reported [28]. Modification in the switching strategy (namely space
vector modulation) for fault-tolerant operation of multilevel CHB V rms* 2
with few devices added to the basic structure have been proposed. V dc = (1)
M*N
This is to ensure that maximum achievable output voltage is
obtained even with switch fault [29]. Replacement of faulty where Vdc is the input voltage to H-bridge; M the modulation
module with a redundant module to ensure fault-tolerant operation
index; N the number of cascaded modules; Vrms the rms value of
is also reported [30].
In this paper, deterioration in the performance of multilevel H-bridge output voltage.
cascaded qZSI due to module failure is analysed. The failure of In conventional inverter, Vdc is the input voltage applied to the
qZSI module is due to semiconductor fault (as semiconductor H-bridge. However, with qZSI applied input voltage (Vin) is
components are the most vulnerable component in any converter boosted to Vdc by controlling the shoot-through duty cycle as
circuit). Quantitative and qualitative analysis is carried out in terms shown in Fig. 1b. This relation is given by [15]
of reduction in rms value and THD performance, respectively.
Harmonic analysis is carried out to ascertain the increase in THD V in
of output phase voltages. Based on these analysis and working V dc = (2)
1 − 2D
theory of qZSI, novel control algorithm is proposed to improve the
post-fault performance. To validate the performance and robustness where Vin is the input voltage applied to qZS and Vdc the input
of the proposed control, simulation results of qZSI system with voltage to the inverter bridge.
isolated RL load powered from DC voltage source are presented. D is the shoot-through duty cycle.
To prove its readiness for solar PV application, the performance of Substituting (2) in (1), the following relation is obtained
grid-connected solar powered multilevel qZSI system with the
proposed control algorithm is analysed under module failure and Vrms ∗ 2
variable solar irradiation conditions in Simulink. Experimental Vin = ∗ 1 − 2D (3)
results for DC voltage source powered multilevel qZSI connected M∗N
to RL load are presented. Experimental results validate the
effectiveness of the proposed control algorithm. Results suggest Thus, depending upon required rms voltage, and available number
of modules, qZSI operation and the output voltage can be
controlled by means of varying the shoot-through duty cycle.

428 IET Renew. Power Gener., 2019, Vol. 13 Iss. 3, pp. 427-437
© The Institution of Engineering and Technology 2018
Fig. 2 Triangle of coupled constraints

3 Deterioration in multilevel inverter output Output voltage waveform for cascaded multilevel qZSI system
voltage waveform during post-fault operation consisting of ‘N’ modules and operating with ‘M’ modulation
index is given by the below equation
3.1 Reduction in voltage magnitude
In the case of a module failure in a multilevel system, bypass vab t = NV γCos ω0t
switches change their state and remove the faulty module from the 4V γ inf inf
1
π m∑ ∑ 2m
circuit. Owing to this, voltage supplied by the faulty module + J2n − 1(mπ)
becomes zero. This means, in the system shown in Fig. 1a, failure = 1 n = − inf
(6)
of one module leads to reduction in voltage magnitude by 25%. In × Cos m + n − 1 π
conventional CHB-based inverters, there is no scope of restoring N
i−1 π
the post-fault voltage to pre-fault value since CHB operate only in × ∑ Cos 2m ωct + N
+ 2n − 1 ω0t
buck mode. However, operation of qZSI is possible in both buck i=1
and boost modes of operation. Utilising this feature, boosting of the
input voltage can be controlled (increased) to restore the voltage where V γ = (V in /(1 − 2D))*M is the peak value of AC rms voltage;
magnitude to the pre-fault value. However, there are certain J2n − 1 the Bessel function of the first kind; ω0 the fundamental
constraints which must be considered before selecting this frequency in rad/s; ωc the carrier frequency in rad/s; m = 1, 2, 3…;
approach, as discussed in the subsequent section. n = 1, 2, 3 ….
In (6), the second coefficient corresponds to the harmonic
3.2 Constraints for voltage boost in qZSI introduced in the spectrum due to PWM operation. Owing to
Owing to module failure, the voltage deficit must be shared among application of unipolar switching, double the switching frequency
the healthy modules in operation to restore Vrms to pre-fault components (represented by 2mωc) are introduced in the spectrum.
condition. This leads to an increase in the DC-bus reference of Usually in cascaded qZSI switching techniques, carrier signals are
available healthy modules. usually phase shifted with respect to each other to eliminate 2mωc
To boost the applied input voltage to this new Vdc new, shoot- components. This can be achieved by giving a phase shift of
through duty cycle must be increased 180°/N with respect to each other. This means, for nine-level
inverter N = 4. Substituting this in second summation term of (6)
V in gives the below equation
V dc new = (4)
1 − 2Dnew N
2m i − 1 π
∑ Cos 2mωct + 2n − 1 ω0t +
N
However, Dnew may or may not be greater than Dmax old. If Dnew > i=1
Dmax old, then the value of Dmax must be increased to achieve the 2mπ 4mπ
= Cos β + Cos β + + Cos β + (7)
new DC-bus reference voltage. An increase in Dmax leads to a N N
decrease in the value of modulation index (M) as for the optimum 2m N − 1 π
performance of multilevel inverter, this equation must be satisfied +⋯ + Cos β +
N
M = 1 − Dmax (5) where
Any decrease in modulation index leads to an increase in the DC- β = 2mct + 2n − 1 ω0t (8)
bus voltage reference required as evident from (1). This is a
cumulative process and it stops when Dnew < Dmax old. These
Equation (7) becomes zero for all m ≠ kN, where k = 1,2,3…. This
coupled constraints as shown in Fig. 2 must be considered while holds true for multilevel inverter operation. Thus, for nine-level
designing the control algorithm. qZSI, when the carrier phase shift of 45° is maintained then
dominating harmonic component will be 2Nωc instead of 2ωc.
3.3 Deterioration in quality of output voltage waveform From this fact, (6) can be written as
Module failure in nine-level qZSI leads to seven-level output
voltage waveform. This seven-level voltage waveform has higher 4V γ inf inf
1
THD content when compared with conventional seven-level qZSI.
vab t = NV γCos ω0t + ∑ ∑ J
π m = 1 n = − inf 2m 2n − 1
(Nmπ)
(9)
To optimise the quality of post-fault voltage waveform, harmonic
analysis of multilevel qZSI is discussed in this section. Fourier × Cos Nm + n − 1 π Cos 2mNωct + 2n − 1 ω0t
analysis of both pre-fault and post-fault voltage waveform is done
to determine the reason for higher THD. Similarly, this discussion can be extended to seven-level inverter
consisting of three modules with phase shift of 60°. In the case of

IET Renew. Power Gener., 2019, Vol. 13 Iss. 3, pp. 427-437 429
© The Institution of Engineering and Technology 2018
module failure, the voltage supplied by that module becomes zero. For utility grid supplied with solar panels, MPPT operation
The nine-level qZSI voltage will be reduced to seven-level qZSI. must be ensured to inject maximum possible current into the utility
Substituting m = 1 and N = 4 in (7) gives (10) where this grid at unity power factor. To achieve this, MPPT algorithm is
summation term does not reduce to zero for all values of m added in the slave controller and output of this loop is the shoot-
through duty cycle (D < Dmax).
4
2 i−1 π The shoot-through duty cycle (by either control loops) per
∑ Cos 2ωct + 2n − 1 ω0t +
N unitised modulating signal (Vref) and carrier signal are used for
i=1
generation of switching pulses as shown in Fig. 3.
π (10)
= Cos β + Cos β + + Cos β + π + 0
2 4.3 Impact of increase in shoot-through duty cycle on voltage
π quality
= Cos β + ≠0
2
In the control algorithm, the increase in shoot-through duty cycle is
Conclusively, during post-module failure operation, dominating suggested to restore the post-fault voltage to pre-fault value.
harmonic component will be 2ωc instead of 2Nωc due to improper However, it must be ensured that the increase in shoot-through
duty cycle must not deteriorate the harmonic performance. To
phase shift. As lower harmonics have higher THD contributions, validate this, generation and implementation of shoot-through
THD of post-module failure output voltage is higher. pulses must be understood.
In Fig. 3b, first generation of shoot-through pulses is shown by
4 Control algorithm comparing the carrier signal with 1 – D and D – 1. Secondly,
Based on the investigation done in Section 3, the following points generation of switching pulses with sine-triangle comparison is
can be concluded: shown. Later, logical OR operation is performed between these two
For post-fault operation: signals to generate resultant switching pulses for insulated gate
bipolar transistors.
i. Voltage magnitude can be restored to pre-fault condition by Consider the instant shown in Fig. 3c, when these conventional
increasing shoot-through duty cycle, but it should not increase pulses are applied to H-bridge, bottom switches of both legs are
THD content of output voltage waveform. turned ON which means the load current is free-wheeling through
the bottom switches and the output voltage generated at this instant
ii. 2ωccomponent introduced in the harmonic spectrum must be
is zero.
eliminated and the THD of the waveform must be at least in It is at this instant that shoot-through pulses are applied, which
the proximity of conventional seven-level qZSI output voltage. results in boosting of applied input voltage. However, the duration
of shoot-through pulses must always be less than this free-wheeling
The purpose of control algorithm for either RL load (with DC duration. If it exceeds the free-wheeling duration, then the output
voltage input) or utility grid load (with solar panel as input) is to voltage becomes zero (due to shoot-through operation). This
control the load current. Thus, generalised control algorithm for changes the output voltage waveform and thereby changes the
current control of either load–source combination is shown in THD content of the voltage. To ensure application of shoot-through
Fig. 3a. The control structure consists of a single centralised pulses only at instant of zero output voltage, the relation M = 1 −
controller and four slave controllers placed individually for each Dmax must always be satisfied. This fact is later verified in the
module.
simulation results.
4.1 Master controller
5 Simulation results
The master controller consists of the current controller followed by
Simulation results for the system are discussed in three
a novel voltage balancing algorithm as shown in Fig. 3a. Sensed
subsections. The first two subsections discuss the performance of
load current is compared with load current reference to generate
RL load and the third subsection discusses the performance with
control signal. For RL load, this control signal is passed as input to
utility grid load.
the voltage balancing algorithm. However, for utility grid load, this
control signal is added with utility grid voltage and filter voltage
drop to generate reference signal and passed on as input to the 5.1 Variation of qZSI output voltage parameter versus shoot-
voltage balancing algorithm. through duty cycle
Voltage balancing algorithm senses the state of operation and The parameters used for the simulation of multilevel qZSI powered
input voltage of each qZSI module. Based on the inputs and rms with DC voltage source are given in Table 1.
voltage required, DC-bus reference and maximum DC-bus voltage For achieving higher voltage boost, the duration of shoot-
generation possible with Dmax is calculated. If through pulse application should be longer. At D = 0, no shoot-
V dc max k + 1 > V dc ref k + 1 , then no updation in variables is through pulses are available as no boosting is required. Shoot-
required. Else, the maximum shoot-through duty cycle is updated through pulses are always applied when the output voltage is zero.
and the value of dependent variables (M, Vdc ref, and θ) are This provides the required voltage boost without any reduction in
determined. These signals are sent as per unitised modulation the output voltage. This operation holds true for operation in the
signal (Vref), Dmax, and phase carrier shift θ(n) to the slave range D ≤ 1 − M. This means that the output voltage wave shape
controller for generation of switching pulses. and number of pulses remain the same for variation of D from 0 to
0.2 (here M = 0.8). To validate this, single module of qZSI is
4.2 Slave controller operated with RL load and Vdc, Vpeak, and VRMS of output voltage
is plotted along with THD as shown in Figs. 4a and b. It should be
The structure of the slave controller is different for RL load current observed that as the shoot-through duty cycle increases, the output
control and grid current control as shown in Fig. 3a. For RL load voltage of quasi-network (Vdc) increases without any increase in
current control, two cascaded PI controllers are used for generation the THD content of the voltage waveform.
of shoot-through duty cycle. First, PI control loop controls the DC- However, when D is increased beyond 1 − M, application of
bus voltage and generates inductor current reference. This becomes shoot-through pulses occurs at non-zero output voltage. Owing to
the input for inductor current control loop. The output of this the application of shoot-through pulses, the output voltage remains
control loop is the shoot-through duty cycle (D). However, the zero. The output voltage wave shape will not be like the one
value of D is governed by saturation limit of D < Dmax (obtained obtained during operation in D < 1 − M. The output voltage
from the master controller). waveform now deviates from sinusoidal behaviour as the output
voltage is forced to zero due to application of shoot-through pulses.

430 IET Renew. Power Gener., 2019, Vol. 13 Iss. 3, pp. 427-437
© The Institution of Engineering and Technology 2018
Fig. 3 Control Algorithm for cascaded qZSI consisting of Master - Slave Control strategy
(a) Master controller–slave controller strategy for load current control,
(b) Typical switching pulses for qZSI for D = 0.1,
(c) Switching signal for H-bridge inverter switches

Owing to this, THD content of the signal will be increased as give nine-level output voltage. This multilevel inverter is
reflected in Fig. 4b for D values of 0.25 and 0.3. connected to RL load. Phase-shifted carrier PWM technique is
implemented to achieve nine-level output voltage. The
5.2 Multilevel inverter operation connected to RL load with required phase shift is 45°. The dynamic performance of the
the proposed control algorithm system is shown in Fig. 4c. During pre-fault condition, the
output voltage waveform has nine levels. Harmonic spectrum
of the output voltage is shown in Fig. 5a. The voltage
i. Pre-fault operation: All the four cascaded qZSI module are magnitude is 240 V peak. It should be observed that in the
operating: Four modules of qZSI are connected in cascade to harmonic spectrum, after the fundamental frequency

IET Renew. Power Gener., 2019, Vol. 13 Iss. 3, pp. 427-437 431
© The Institution of Engineering and Technology 2018
 component, the most dominating harmonic is 8 kHz (2*N*ωc five-level. This reflects as increase in the load current drawn.
= 2*4*1 = 8 kHz). This is in accordance with the harmonic Waveform of the output voltage along with harmonic spectrum
analysis. is shown in Fig. 5b. Owing to module failure, the voltage
ii. Post-fault operation without the proposed control algorithm: magnitude has now reduced from 240 to 176 V, which is
At t = 3 s, qZSI module failure as shown in Fig. 4c. Owing to almost 25% reduction. Harmonic spectrum also shows increase
this, nine-level output voltage waveform is now reduced to in THD content from 17.52 to 30.14%. Here, the dominant

Table 1 Parameter specification for each module of qZSI

Parameter Specification
quasi-inductor (L1 = L2) 4.5 mH
quasi-capacitor (C1 = C2) 6.0 mF
Dmax 0.2
input voltage (Vin) 65 V
modulation index (M) 0.8
switching frequency 1 kHz
load impedance 60 Ω, 50 mH

Fig. 4 Simulation results for performance variation as a function of duty cycle (D)
(a) Variation of voltages with respect to D,
(b) Variation of THD with respect to D,
(c) Response of cascaded qZSI from t = 0 to 12 s

432 IET Renew. Power Gener., 2019, Vol. 13 Iss. 3, pp. 427-437
© The Institution of Engineering and Technology 2018
 harmonic component is 2 kHz (=2ωc) in the harmonic b. Modifying the phase shift to improve THD performance:
spectrum. The second step of implementation is the optimisation of
iii. Post-fault operation with the proposed control algorithm: the harmonic spectrum by eliminating harmonic
components (=2*N*ωc = 6 kHz for three modules). By
a. Restoring voltage magnitude to pre-fault value: In this implementing this step, improvement in the harmonic
step, the voltage balancing algorithm is activated at t = 6 component is observed. This is evident from Fig. 5d
s. Voltage deficit due to module failure must be where the voltage magnitude is 240 V and harmonic
compensated using other healthy operating modules. The components <6 kHz are eliminated. THD of the output
sensed signals start the voltage balancing algorithm and voltage is reduced from 30.53 to 23.58%.
V dc max > V dc ref comparison is done. Based on this,
shoot-through duty cycle and other variables are updated. Thus, by implementing the proposed control algorithm,
In Fig. 4c, the increase in DC-bus reference voltage of restoration of voltage magnitude and optimisation of harmonic
each healthy module is observed. spectrum of output voltage is achieved.
To verify the effectiveness of the implemented
algorithm, harmonic spectrum is analysed as shown in 5.3 Solar PV fed grid-connected multilevel qZSI system with
Fig. 5c. Voltage magnitude is restored back to 240 V; the proposed control algorithm
however, there is no change in the THD content of the
waveform. This validates the fact that when the equation To validate the proposed control algorithm for solar PV
Dmax ≤ 1 − M is satisfied, there is no change in the THD. applications, cascaded multilevel qZSI is connected to the utility
grid. Injection of power from PV panels to the utility grid is always
at unity power factor.

Fig. 5 Harmonic profile of load voltage

(a) With four modules operating,
(b) Post one module failure,
(c) Post-fault with rms voltage restored,
(d) Post-fault with rms voltage restored and modular frequency eliminated

Fig. 6 PV characteristics and performance of 240W solar panel

(a) Rating of PV panel at different solar irradiations,
(b) Response of grid-connected multilevel qZSI system when subjected to module failure and irradiation changes

IET Renew. Power Gener., 2019, Vol. 13 Iss. 3, pp. 427-437 433
© The Institution of Engineering and Technology 2018
i. Characteristics of customised PV module: The characteristics overall current supplied by the cascaded multilevel inverter. To
of the PV module used for simulation of the system are shown address this problem, during post-module failure operation,
in Fig. 6a. Rating of each PV panel is 120 W with MPPT point solar irradiation level of module 1 is decreased from 1 to 0.8
being 30 V, 4 A. Specifications of grid-connected system are kW/m2 at t = 6 s. To mitigate this disturbance, the control
given in Table 2. algorithm shifts the operating point of the operating modules
ii. Response of the system during pre and post-module failure: so that the output current of all qZSI modules becomes equal.
Simulation results of the solar PV fed grid-connected system Owing to this, the resultant inverter output current injected into
are shown in Fig. 6b. During pre-fault condition, the four the grid at unity power factor is reduced as shown in Fig. 6b
modules are operating at MPPT point of 240 W, respectively. and its zoomed waveform is shown in Fig. 7. It can also be
During this time, the grid current injects active power into the observed that variation in environmental condition will not
grid. At t = 2 s, module failure is observed. Owing to this, affect the quality of the output voltage waveform.
voltage deficit is created in the system. If the operation is
allowed in this mode, then the multilevel qZSI inverter draws
huge reactive current from the utility grid due to this voltage 6 Experimental results
deficit. To avoid this, the first step of the algorithm is
implemented at t = 2 s. In this step, inverter output voltage is The experimental set-up of nine-level qZSI with RL load is
restored to pre-fault value by increasing the boosting of the developed with the same parameters as used for simulation. To
input voltage. This is reflected by increased DC-bus voltage as validate the effectiveness of the control algorithm, the performance
shown in Fig. 6b. However, the power deficit created due to of the multilevel qZSI is tested both with and without proposed
module failure will be reflected as decrease in grid current control algorithm. Results obtained for multilevel inverter
magnitude as shown in Fig. 6b and its zoomed version is operation in pre-fault and post-fault condition are discussed below.
shown in Fig. 7.
The operation of grid-connected system at unity power 6.1 Pre-fault operation and qZSI module failure
factor is ensured during this mode as shown in Fig. 7. The
second step of the algorithm involves the correction of Multilevel inverter output voltage consisting of nine-level phase
remaining healthy modules’ carrier phase shift from 180°/4 to voltage is shown in Fig. 8a. The zoomed version shows the levels
180°/3. This improves the THD content of the output voltage formed along with the load current. For emulation of qZSI module
quality at t = 4 s. The response of the system is shown in failure and bypass mechanism, bottom switches of the H-bridge are
Figs. 6b and 7. given continuous pulses and pulses of top switches are withdrawn.
Owing to module failure, the number of voltage levels in the
iii. Response of multilevel qZSI at different solar irradiations
waveform is reduced. Owing to this, the voltage magnitude
levels: Solar irradiation of a given panel is always proportional
reduces, and the harmonic spectrum is deteriorated. Voltage
to its current rating. A decrease in solar irradiance is
magnitude and harmonic spectrum of the output voltage and load
proportionally reflected in the output current of the panel.
current during pre-fault condition are shown in Fig. 8b.
When solar PV fed multilevel inverters are used, the
Dominating frequency component after the fundamental frequency
occurrence of different solar irradiation levels in different
in both harmonic spectra is 8 kHz (=2*N*ωc).
module is quite common. This may cause reduction in the
Harmonic spectra during post-fault condition are shown in
Fig. 8c where the voltage magnitude is reduced and now the
Table 2 PV panel specifications for grid-connected dominating frequency component is 2 kHz (2*ωc). Without
multilevel qZSI proposed control algorithm, the performance of the multilevel
Parameter Specification inverter will continue to be in this state.
power rating 240 W To improve the post-fault operation, the magnitude of voltage
MPPT voltage 60 V and current must be restored to pre-fault operation. This is
MPPT current 4A
achievable in the case of qZSI as input voltage can be boosted. The
next step is the implementation of the proposed control algorithm.
grid voltage, rms 220 V
Implementation is carried out in two steps to ensure that the impact
grid current, rms 4.2 A [pre-fault] of shoot-through duty cycle is not felt on the harmonic spectrum of
3.1 A [post-fault] output voltage and current.

Fig. 7 Zoomed waveforms of inverter voltage, grid current, and grid voltage ensuring operation at unity power factor during all conditions

434 IET Renew. Power Gener., 2019, Vol. 13 Iss. 3, pp. 427-437
© The Institution of Engineering and Technology 2018
6.2 Post-fault operation with proposed control algorithm Thus, with the proposed control algorithm consisting of master
controller and slave controllers improves both quality and quantity
of the output voltage waveform of multilevel inverter.
i. Restoring voltage magnitude to pre-fault condition: Voltage
balancing algorithm is now activated, and DC-bus reference,
maximum shoot-through duty cycle, and modulation index are 6.3 Steady-state operation waveform of single qZSI during
updated till the inequality Vdc max − Vdc ref > 0 is satisfied. post-fault and pre-fault operation
Dynamic response of the system during this transition is shown In the pre-fault condition, four qZSI modules are supplying power
in Fig. 9a. To confirm restoration of voltage magnitude, rms to the RL load, whereas in the post-fault condition, only three qZSI
value is measured as shown in Fig. 9c. To validate that there is modules are connected to the load. To keep the same voltage and
no impact on the harmonic spectrum, Fig. 9c also shows power level, power drawn from remaining healthy modules will be
harmonic spectra of both signals. The dominating component increased during post-fault condition. This must be reflected in the
after fundamental frequency is still two times the switching steady-state performance of each qZSI module. To observe this,
frequency. DC-bus voltage and inductor current of single qZSI module for
ii. Improving the quality of output voltage waveforms: The next pre-fault and post-fault condition are shown in Figs. 10a and b,
step is to improve the harmonic spectra of the output signals. respectively. Fig. 10a shows qZSI waveforms during pre-fault
Transient response of the multilevel qZSI for this step is shown condition. Inverter output voltage consists of nine levels and its
in Fig. 9b where the quality of voltage waveform is improved, rms value is 170 V with load current value being 2.8 A. After the
although the voltage peak remains the same. module failure, the DC-bus voltage and inductor current drawn are
increased by ∼35% to address the power deficit as shown in
To observe this improvement reflection on the harmonics Fig. 10b.
content of the signal, harmonic spectrum is presented in Fig. 9d
where the dominating frequency component after fundamental 7 Conclusion
frequency is now 6 kHz instead of 2 kHz. For conventional seven-
level qZSI, dominating harmonic component must be 6 kHz as In this paper, deterioration in the multilevel qZSI output voltage
shown. performance due to module failure is discussed. The impact on the
voltage magnitude (quantity) and THD content of the output
voltage (qualitatively) is analysed by Fourier analysis. Based on

Fig. 8 Steady state performance and harmonic spectrum of inverter voltage and current during pre - fault and post - fault operations
(a) Transient response for failure of qZSI module with zoomed waveforms of steady-state operations in pre-fault and post-fault conditions,
(b) Harmonic spectrum of pre-fault operation,
(c) Harmonic spectrum of post-fault operation

IET Renew. Power Gener., 2019, Vol. 13 Iss. 3, pp. 427-437 435
© The Institution of Engineering and Technology 2018
Fig. 9 Transient response for two steps of control algorithm implementation
(a) Transient response for increase in shoot-through duty cycle to restore the voltage to pre-fault value,
(b) Transient response for minimisation of THD for seven-level inverter operation by modifying the phase shift during post-fault operation,
(c) Harmonic spectrum of output voltage and output current after voltage boost in post-fault condition,
(d) Harmonic spectrum of post-fault operation with dominant frequency component of 6 kHz

Fig. 10 Steady state performance of qZSI module during pre-fault operation and post-fault algorithm implementation
(a) Steady-state waveform of nine-level inverter in pre-fault operation,
(b) Steady-state waveform of seven-level inverter restored to pre-fault operation

the outcome of the Fourier analysis and considering the voltage voltage source and grid-connected solar powered multilevel qZSI
boost restrictions in qZSI, a novel voltage balancing control system are presented. The system is subjected to module failure
algorithm is presented. To verify the use of this control algorithm, and variable environmental conditions (for grid-connected system).
simulation results for both isolated RL load powered with DC The response of the system shows achievement of pre-fault voltage

436 IET Renew. Power Gener., 2019, Vol. 13 Iss. 3, pp. 427-437
© The Institution of Engineering and Technology 2018
magnitude and optimum THD value for isolated RL load. For grid- [13] Touati, F.A., Al-Hitmi, M.A., Bouchech, H.J.: ‘Study of the effects of dust,
relative humidity, and temperature on solar PV performance in Doha:
connected system, stable post-fault operation is achieved at comparison between monocrystalline and amorphous PVS’. Int. J. Green
reduced grid current with the proposed algorithm. Post-module Energy, 2013, 10, (7), pp. 680–689
failure system is also subjected to variable solar irradiation [14] Peng, F.Z.: ‘Z-source inverter’, IEEE Trans. Ind. Appl., 2003, 39, (2), pp.
condition, which gives satisfactory results. Lastly, experimental 504–510
[15] Vinnikov, D., Roasto, I.: ‘Quasi-Z-source-based isolated DC/DC converters
results are presented for multilevel qZSI powered with DC voltage for distributed power generation’, IEEE Trans. Ind. Electron., 2011, 58, (1),
source connected to RL load. Post-fault operation performance pp. 192–201
with and without proposed control algorithm are shown, which [16] Abu-Rub, H., Iqbal, A., Ahmed, S.M., et al.: ‘Quasi-Z-source inverter-based
clearly indicate the improvement in THD performance and voltage photovoltaic generation system with maximum power tracking control using
ANFIS’, IEEE Trans. Sustain. Energy, 2013, 4, (1), pp. 11–20
magnitude. Comparison of steady-state performance for the same [17] Meraj, M., Rahman, S., Iqbal, A., et al.: ‘A hybrid active and reactive power
output power during pre-fault and post-fault are also shown. control with quasi Z-source inverter in single-phase grid-connected PV
systems’. IECON 2016 – 42nd Annual Conf. of the IEEE Industrial
Electronics Society, Florence, 2016, pp. 2994–2999
8 Acknowledgments [18] Liu, Y., Ge, B., Abu-Rub, H., et al.: ‘Model predictive direct power control
for active power decoupled single-phase quasi-Z-source inverter’, IEEE
This publication was made possible by NPRP-EP grant # Trans. Ind. Inf., 2016, 12, (4), pp. 1550–1559
[X-033-2-007] from the Qatar National Research Fund (a member [19] Rahman, S., Meraj, M., Iqbal, A., et al.: ‘Failure mode analysis for single-
of Qatar Foundation). The statements made herein are solely the phase multi-level qZSI interfacing PV system to utility grid’. 2017 11th IEEE
responsibility of the authors. Int. Conf. on Compatibility, Power Electronics and Power Engineering (CPE-
POWERENG), Cadiz, 2017, pp. 504–509
[20] Rahman, S., Meraj, M., Iqbal, A., et al.: ‘Thyristor based SVC and multilevel
9 References qZSI for active and reactive power management in solar PV system’. 2017
11th IEEE Int. Conf. on Compatibility, Power Electronics and Power
[1] Rodriguez, J., Lai, J-S., Peng, P.Z.: ‘Multilevel inverters: a survey of Engineering (CPE-POWERENG), Cadiz, 2017, pp. 528–533
topologies, controls, and applications’, IEEE Trans. Ind. Electron., 2002, 49, [21] Liu, Y., Ge, B., Abu-Rub, H.: ‘An active power decoupling quasi-Z-source
(4), pp. 724–738 cascaded multilevel inverter’. IECON 2016 – 42nd Annual Conf. of the IEEE
[2] Rodriguez, J., Bernet, S., Wu, B., et al.: ‘Multilevel voltage-source-converter Industrial Electronics Society, Florence, 2016, pp. 6453–6458
topologies for industrial medium-voltage drives’, IEEE Trans. Ind. Electron., [22] Trabelsi, M., Abu-Rub, H., Ge, B.: ‘1-MW quasi-Z-source based multilevel
2007, 54, (6), pp. 2930–2945 PV energy conversion system’. 2016 IEEE Int. Conf. on Industrial
[3] Abu-Rub, H., Holtz, J., Rodriguez, J., et al.: ‘Medium-voltage multilevel Technology (ICIT), Taipei, 2016, pp. 224–229
converters – state of the art, challenges, and requirements in industrial [23] Lu, B., Sharma, S.K.: ‘A literature review of IGBT fault diagnostic and
applications’, IEEE Trans. Ind. Electron., 2010, 57, (8), pp. 2581–2596 protection methods for power inverters’, IEEE Trans. Ind. Appl., 2009, 45,
[4] Tolbert, L.M., Peng, F.Z., Habetler, T.G.: ‘Multilevel converters for large (5), pp. 1770–1777
electric drives’, IEEE Trans. Ind. Appl., 1999, 35, (1), pp. 36–44 [24] Rodríguez, J., Hammond, P.W., Pontt, J., et al.: ‘Operation of a medium-
[5] Fazel, S.S.: ‘Investigation and comparison of multi-level converters for voltage drive under faulty conditions’, IEEE Trans. Ind. Electron., 2005, 52,
medium voltage applications’ (Technische Universität, Berlin, 2007) (4), pp. 1080–1085
[6] Krug, D., Bernet, S., Fazel, S.S., et al.: ‘Comparison of 2.3-kV medium- [25] Song, W., Huang, A.Q.: ‘Fault-tolerant design and control strategy for
voltage multilevel converters for industrial medium-voltage drives’, IEEE cascaded H-bridge multilevel converter-based STATCOM’, IEEE Trans. Ind.
Trans. Ind. Electron., 2007, 54, (6), pp. 2979–2992 Electron., 2010, 57, (8), pp. 2700–2708
[7] Corzine, K.: ‘Operation and design of multilevel inverters’. Developed for the [26] Yu, Y., Konstantinou, G., Hredzak, B., et al.: ‘Power balance of cascaded H-
Office of Naval Research, 2003, pp. 1–79 bridge multilevel converters for large-scale photovoltaic integration’, IEEE
[8] Khersonsky, Y., Lee, R.H.: ‘Step switched PWM sine generator’. Google Trans. Power Electron., 2016, 31, (1), pp. 292–303
Patents, 2003 [27] Yu, Y., Konstantinou, G., Hredzak, B., et al.: ‘Power balance optimization of
[9] Corzine, K., Familiant, Y.: ‘A new cascaded multilevel H-bridge drive’, IEEE cascaded H-bridge multilevel converters for large-scale photovoltaic
Trans. Power Electron., 2012, 17, (1), pp. 125–131 integration’, IEEE Trans. Power Electron., 2016, 31, (2), pp. 1108–1120
[10] Zambra, D.A.B., Rech, C., Pinheiro, J.R.: ‘A comparative analysis between [28] Lezana, P., Ortiz, G.: ‘Extended operation of cascade multicell converters
the symmetric and the hybrid asymmetric nine-level series connected H- under fault condition’, IEEE Trans. Ind. Electron., 2009, 56, (7), pp. 2697–
bridge cells inverter’. 2007 European Conf. on Power Electronics and 2703
Applications, Aalborg, 2007, pp. 1–10 [29] Aleenejad, M., Iman-Eini, H., Farhangi, S.: ‘Modified space vector
[11] Najmi, E.S., Ajami, A.: ‘Modular symmetric and asymmetric reduced count modulation for fault-tolerant operation of multilevel cascaded H-bridge
switch multilevel current source inverter’, IET Power Electron., 2016, 9, (1), inverters’, IET Power Electron., 2013, 6, (4), pp. 742–751
pp. 51–61 [30] Moamaei, P., Mahmoudi, H., Ahmadi, R.: ‘Fault-tolerant operation of
[12] Tsang, K.M.., Chan, W. L.: ‘Single DC source three-phase multilevel inverter cascaded H-bridge inverters using one redundant cell’. 2015 IEEE Power and
using reduced number of switches’, IET Power Electron., 2014, 7, (4), pp. Energy Conf. at Illinois (PECI), Champaign, IL, 2015, pp. 1–5
775–783

IET Renew. Power Gener., 2019, Vol. 13 Iss. 3, pp. 427-437 437
© The Institution of Engineering and Technology 2018

View publication stats