Controller Design and Implementation of Solar Panel Companion Inverters

IEEE TRANSACTIONS ON INDUSTRY APPLICATIONS, VOL. 56, NO.
2, MARCH/APRIL 2020 2001
Controller Design and Implementation of Solar

Panel Companion Inverters
Prasanth Kumar Sahu and Madhav D. Manjrekar , Senior Member, IEEE
Abstract—A solar panel companion inverter (SPCI) is a single- utility grid. String inverters are smaller than central inverters
stage dc–ac power converter, installed with each solar panel. Each as they process power only from one string. An alternative
SPCI synthesizes an ac quasi-square wave voltage with variable approach is to employ a microinverter for every solar panel [4].
pulsewidth. The width of ac output voltage of each SPCI is pro-
portional to available power on the solar panel. Output voltages Such small inverters convert dc power produced by a single
of SPCIs are aggregated across multiple solar panels connected solar panel and synthesize required ac voltage. Because of its
in series, and a sinusoidal ac voltage is synthesized. In this arti- modularity, this topology has an additional advantage of scaling
cle, dynamic analysis and closed-loop current controller design of up to the required power level. Microinverter topology needs an
SPCI is presented. Simulation results demonstrating the dynamic additional power conversion stage to step up the module voltage
response of the closed-loop SPCI are included. The closed-loop
current control scheme is implemented on a hardware prototype of the solar panel (usually 30–35 V dc at maximum power point
to demonstrate the grid-tied operation of SPCI, and experimental operation), to match the grid voltage level. This additional dc–dc
results are presented. A maximum power point tracking (MPPT) power conversion stage introduces additional losses in the sys-
algorithm is implemented for sorted stair-case modulation (SSCM) tem. Also, the price on per watt basis is the highest for microin-
in MATLAB Simulink. Experimental results demonstrating MPPT verter topology, when compared to central and string inverter
operation are presented.
architectures.
Index Terms—Active power control, cascaded H- bridge For given operating conditions of sunlight and temperature,
inverter, current control, maximum power point tracking each PV module will have its own current–voltage (IV) and
(MPPT), multilevel inverter, photovoltaic (PV) inverter, PV power
conversion, reactive power control.
power–voltage (P–V) characteristics. In any PV inverter topol-
ogy, to ensure maximum utilization of the PV modules, it is
I. INTRODUCTION important that all PV modules are operated at their correspond-
ing maximum power points. Maximum power point tracking
OWARD optimizing capital costs of power electronics on
T a per watt basis, central photovoltaic (PV) inverters have
been designed for bulk power generation applications [1]–[3].
(MPPT) algorithms such as perturb and observe or incremental
conductance can be implemented to ensure that the maximum
power extraction is possible from PV inverters [7]–[16].
In a centralized architecture, solar panel strings are constructed
There can be two approaches toward controlling the grid-
by connecting individual panels in series, to build up dc voltage.
connected PV inverters: voltage-mode control and current-mode
These strings are then connected in parallel to scale up to the
control. Voltage-mode control has been utilized for flexible
required power level before feeding into a single central inverter.
ac transmission system controllers [17]–[20]. In this mode of
A variation to the centralized topology could be obtained by
control, the real and reactive powers are controlled by con-
introducing a dc–dc power conversion stage before the dc–ac
trolling the phase angle and amplitude of inverter output ac
conversion. The dc–dc power conversion stage can either be
voltage with respect to the voltage at point of common coupling
at the string level or at the panel level [4], [5]. An improved
(PCC). Voltage-mode control has a drawback that there is no
version of central inverter topology is called the string inverter
explicit control over the ac grid current, which means that the
topology [6]. In this topology, strings of panels are directly
inverter is exposed to conditions such as over currents [21].
connected to an inverter, which is then interfaced with an ac
In current-mode control, the ac grid current of the inverter is
controlled directly. In this mode, the active and reactive powers
Manuscript received April 14, 2019; revised August 12, 2019 and October 20,
2019; accepted November 18, 2019. Date of publication January 12, 2020; date
are controlled by controlling the phase angle and amplitude
of current version March 17, 2020. Paper no. 2019-IACC-0317.R2, presented of ac grid current with respect to the voltage at PCC. This
at the 2018 IEEE Energy Conversion Congress and Exposition, Portland, OR, ensures that the inverter is protected against sudden variations
USA, Sep. 23–27, and approved for publication in the IEEE TRANSACTIONS
ON INDUSTRY APPLICATIONS by the Industrial Automation and Control Com-
in the current. This approach also has the advantages of being
mittee of the IEEE Industry Applications Society. This work was supported robust and having a very good dynamic performance [21], [22].
by the Energy Production & Infrastructure Center (EPIC), University of North The control loops typically involve a comparison of parameters,
Carolina-Charlotte. (Corresponding author: Prasanth Kumar Sahu.)
The authors are with the Department of Electrical and Computer Engineering,
which are sinusoidal in nature. Controllers for such loops can
University of North Carolina-Charlotte, Charlotte, NC 28223 USA (e-mail: be proportional-integral (PI) or proportional-resonant (PR). Im-
[email protected]; [email protected]). plementing a PI controller to control the instantaneous value of
Color versions of one or more of the figures in this article are available online
at https://ieeexplore.ieee.org.
sinusoidal current is challenging to achieve a zero steady-state
Digital Object Identifier 10.1109/TIA.2020.2965867 error without a reference frame transformation [23]. On the
0093-9994 © 2020 IEEE. Personal use is permitted, but republication/redistribution requires IEEE permission.
See https://www.ieee.org/publications/rights/index.html for more information.
Authorized licensed use limited to: REGINA (Peru). Downloaded on April 13,2021 at 00:03:13 UTC from IEEE Xplore. Restrictions apply.
2002 IEEE TRANSACTIONS ON INDUSTRY APPLICATIONS, VOL. 56, NO. 2, MARCH/APRIL 2020
other hand, a PR controller can achieve a zero steady-state error

because of its high gain at fundamental frequency [24].
Solar panel companion inverters (SPCIs) have been proposed
as an alternative approach, which offers advantages of microin-
verters in realizing panel-level MPPT while simultaneously real-
izing system-level cost benefits of a central inverter [25]. SPCI
is a single-stage power conversion topology, whereas central
inverters, string inverters, and microinverters sometimes involve
two-stage power conversion to improve the yield. While there are
other multilevel inverter topologies such as diode clamped and
flying capacitor inverter topologies, they suffer drawbacks of be-
ing nonmodular and have complex control implementation [26],
[27]. SPCI is also a multilevel inverter topology but is modular
and control methodology is not required to be as complex as in
diode clamped or flying capacitor topologies. In previous work,
unsorted pulsewidth modulation, sorted stair-case modulation
(SSCM), and sorted pulsewidth modulation have been discussed
Fig. 1. Simplified circuit schematic of SPCIs connected to a 120-V ac rms
and the proposed SPCI system has been controlled in an open distribution grid [25]–[28].
loop such that only real power flows into the grid, at unity
power factor. The MPPT algorithm has been implemented to
demonstrate MPPT at the panel level, and active and reactive
power capabilities [28]–[30]. Although closed-loop current con-
trol was implemented in later manifestations, focus has been
on demonstrating the MPPT operation, whereas the controller
analysis and design of grid-connected SPCIs have not been
discussed in detail. In this article, the controller design and im-
plementation of grid-connected SPCI are presented in elaborate
details. Section III describing the design of the employed control
scheme is the main contribution in this article. Analysis similar to
the one presented in [21] is presented. Simulations are performed
in MATLAB Simulink to demonstrate the closed-loop current
control of grid-connected SPCI. Implementation of the MPPT
algorithm together with closed-loop current control operation
has been presented to demonstrate maximum power operation Fig. 2. Schematic of current controlled SPCIs interfaced with the grid at PCC.
in both simulations and hardware implementation.
Following Section I, this article presents description of SPCI
and its operating principles. Control scheme employed for the
synthesized. This is implemented by a sort and stack algorithm,
closed-loop operation of grid-connected SPCI is presented in
which also considers disparity in irradiation levels of solar
Section III. In Section IV, simulation results demonstrating
panels. To realize a required ac voltage, several such panels
the maximum power point operation of the PV modules are
are connected in series through their companion inverters. The
presented. Section V shows the experimental voltage and current
power circuit to synthesize a 120-V ac rms voltage is shown
waveforms obtained from a laboratory prototype. The article
in Fig. 1, wherein six panels, each capable of producing a dc
concludes with Section VI.
voltage of about 35 V, are connected in series through their
corresponding companion inverters and are interfaced with the
II. SYSTEM DESCRIPTION distribution grid via an isolation transformer. The magnitude of
the output ac waveform mag(V0 ) to be synthesized by the PV
SPCI is topologically identical to the cascaded H-bridge mul-
generation system is a function of the distribution grid voltage
tilevel inverter [31]. However, it is fundamentally different in
and the voltage drop across the interfacing reactance.
the way it is being implemented.
Each H-bridge inverter module is reduced to four power elec-
tronics devices, preferably Metal Oxide Semiconductor Field III. CONTROL SCHEME
Effect Transistors (MOSFETS), one dc-link capacitor, one micro- Fig. 2 shows a schematic diagram of current-controlled SPCI
controller, and a parasitic power supply harvesting energy from system. SPCIs are interfaced with the ac grid via a series RL
the solar panel itself. This inverter module is integrated with the branch at the PCC. An ac grid is stiff and is modeled by an ideal
solar panel (hence, the name SPCI), and depending on the dc single-phase voltage source Vgrid . The system should be able to
voltage of the panel, a quasi-square wave of required duration exchange active power P and reactive power Q with the ac grid.
or pulsewidth modulated waveform of required pulsewidth is The control objective of grid-connected SPCI is to synthesize a
SAHU AND MANJREKAR: CONTROLLER DESIGN AND IMPLEMENTATION OF SOLAR PANEL COMPANION INVERTERS 2003

sinusoidal ac output voltage (both magnitude and phase angle). diq dρ
The magnitude and phase angle of the synthesized output voltage L =− L id − Riq + Vspciq − Vgridq (11)
dt dt
should be such that a current of specific magnitude and phase
dρ
angle is injected into the grid. = ω. (12)
A closed-loop current control, referred to as current-mode dt
control, capable of tracking a sinusoidal reference current can If ω0 is the frequency of the ac grid, and ρ(t) = ω0 t + θ0 , then
be implemented. However, tracking a sinusoidal reference input (10) and (11) can be rewritten as
with the zero steady-state error is difficult. Hence, the current did
control scheme is implemented in the dq-frame, and a zero L = Lω0 iq − Rid + Vspcid − Vgridd (13)
dt
steady-state error can be achieved by including a PI compen-
diq
sator, since the control variables are dc quantities. Current L = −Lω0 id − Riq + Vspciq − Vgridq . (14)
control in the dq-frame is usually implemented in three-phase dt
systems by using Park’s transformation. For a single-phase The abovementioned equations (13) and (14) describe a first-
system, the measured current first needs to be converted to order linear system with Vgridd and Vgridq as constant inputs. If
the static αβ-frame [32]. First, a current in quadrature with Vspcid and Vspciq are dc quantities, id and iq are also dc quantities
the measured current is obtained by delaying the measured in the steady state. ρ(t) = ω0 t + θ0 is ensured by implementing
sinusoidal current by 90°. SPCI shown in this figure represents a a phase-locked loop algorithm.
lossless power processor, which converts dc power to ac power.
Real and reactive power can be controlled by controlling the B. Current Control Loop
amplitude and phase angle of the ac current with respect to ac The current control in the dq-frame is based on (13) and (14).
voltage of the ideal voltage source. In these equations, id and iq are state variables, Vspcid and Vspciq
are control inputs, and Vgridd and Vgridq are disturbance inputs.
A. Dynamic Model of Current Controller The dynamics of id and iq are coupled due to the presence of
Dynamics of the ac side of the SPCI system of Fig. 2 may be reactance terms (Lω0 ) in the (13) and (14). Let Vid and Viq be
described by two control inputs calculated as follows:
di −−→ −−→ Vid = Lω0 iq + Vspcid − Vgridd (15)
L = −Ri + Vspci − Vgrid (1)
dt Viq = −Lω0 id + Vspciq − Vgridq . (16)
where i is the grid side ac current phasor, R is the resistance of
−−→ Hence, (13) and (14) can be rewritten as
the interfacing inductor, L is the interfacing inductance, Vspci is
−−→ did
the phasor output voltage of SPCI, and Vgrid is the phasor grid L = −Rid + Vid (17)
voltage. Representing the voltage and current phasors of (1) in dt
αβ and dq frames of reference as follows: diq
L = −Riq + Viq . (18)
i = iα + jiβ = (id + jiq ) ejρ(t) dt
(2)
The abovementioned equations (17) and (18) represent two,
−−→
Vspci = Vspciα + jVspciβ = (Vspcid + jVspciq ) ejρ(t) (3) first-order linear systems. It may be seen from these equations
−−→ that the quantities id and iq can be controlled by controlling
Vgrid = Vgridα + jVgridβ = (Vgridd + jVgridq ) ejρ(t) . (4) Vid and Viq , respectively. Rewriting (17) and (18) in s-domain
representation, we get
Equation (1) can be rewritten using (2)–(4) as
d LsId (s) = −RId (s) + Vid (s) (19)
L idq ejρ(t) = − R idq ejρ(t) + Vspcidq ejρ(t)
dt LsIq (s) = −RIq (s) + Viq (s) . (20)
− Vgriddq e jρ(t)
(5) Transfer functions of the systems described by (19) and (20),
idq = (id + jiq ) (6) respectively, are
Vspcidq = (Vspcid + jVspciq ) (7) Id (s) 1

= (21)
Vid (s) Ls + R
Vgriddq = (Vgridd + jVgridq ) . (8)
Iq (s) 1
Equation (5) can be rewritten as = . (22)
Viq (s) Ls + R
d dρ
L (idq ) = −j L idq − R (idq ) + Vspcidq − Vgriddq . From (21) and (22), it can be said that the plants represented
dt dt in d-axis and q-axis are identical. Fig. 3 shows the simplified
(9)
block diagram of the open-loop system represented by (17) and
Equating the real and imaginary parts of (9), we get
(18).
did dρ The control objective for such systems would be to regulate
L = L iq − Rid + Vspcid − Vgridd (10)
dt dt the output currents id and iq at their corresponding specified
Loop gain becomes

kp
ld (s) = . (26)
sL
Hence, the closed-loop transfer function can be written as
Id (s) 1
= Gid (s) = . (27)
Idref (s) s (L/kp ) + 1
Fig. 3. Simplified block schematic showing open-loop control model of grid-
connected SPCIs. A parameter called time constant τi by
L
τi = . (28)
kp
Therefore, (27) can be rewritten as
Id (s) 1
= Gid (s) = . (29)
Idref (s) sτi + 1
The abovementioned equation (29) represents a first-order
transfer function with unity gain. Choice of τi is a design
problem. For a faster response, τi should be small. 1/τi defines
the bandwidth of the closed-loop control system. Bandwidth of
the control system should be selected such that it is at least ten
times smaller than the switching frequency. The compensator for
q-axis will be the same as that specified by (23) and calculated
Fig. 4. Simplified block schematic showing closed-loop current control in as
grid-connected SPCIs in the dq-frame. ki
kq (s) = kp + (30)
s
and the closed-loop transfer function for the q-axis current
reference values. Fig. 4 shows a closed-loop current control
control loop can be written as
block schematic wherein, id is controlled by comparing idref
with id to generate an error signal ed . This error signal is fed into Iq (s) 1
= Gid (s) = . (31)
a compensator kd (s), and a control input signal Vid is generated. Iqref (s) sτi + 1
Similarly, iq is controlled by comparing iqref with iq to generate
an error signal eq . This error signal is fed into a compensator C. Controller Design and Bode Plot
kq (s), and the control input signal Viq is generated. Interfacing inductance, L value is assumed to be 17 mH, and
Therefore, the corresponding compensators can also be iden- the series resistance is assumed to be 1 Ω. For six modules of
tical. For both the current control loops, the compensators can be SPCIs, the effective switching frequency is (4 × 6 × 60) =
a PI compensator, since the reference signals are dc quantities. 1440 Hz. Therefore, the bandwidth of the current controller
The PI compensator for the d-axis current control loop can be is chosen to be 1/τi = 144 Hz. Therefore, the time constant,
described by τi ≈ 7 ms. From (28) and (31), kp = 2.448 and ki = 144. There-
ki fore, the transfer functions of the compensators along d-axis and
kd (s) = kp + (23) q-axis control loops can be written as
s
where kp is a proportional gain, and ki is integral gain. The 2.448s + 144
kd (s) = kq (s) = . (32)
integral term of (23) guarantees that id tracks idref , with a zero s
steady-state error. The control system loop gain can be written The plant transfer functions from (23) and (24) can be written
as as
s + ki Id (s) Iq (s) 1
kp kp = = . (33)
ld (s) = . (24) Vid (s) Viq (s) 0.017s + 1
Ls s+ R L
Using the loop transfer function, a bode plot, as shown in
The plant pole is at s = −R/L. For an ideal case, resistance of Fig. 5, is generated. The gain margin is infinite, and the phase
the interfacing inductor is ≈ 0. For a nonzero resistance, though margin is 90°.
the pole is on the left half of s-plane, it is close to origin, and A closed-loop control system described by (32) and (33) has
hence, the natural response of the system is slow. Pole of the been built in MATLAB Simulink, to study the step response of
plant can be cancelled by zero of compensator by selecting kp the system. The input to the system is idref and output is id .
and ki such that The system is subjected to following sequence of events: idref
ki R is initially set to the 19.5 (which is the rated id value) until
= . (25)
kp L 0.15 s. At t = 0.15 s, idref is subjected to a step change from
Fig. 5. Bode plot for loop transfer function.
Fig. 7. Simplified circuit schematic of SPCIs, with ideal dc voltage sources

on the dc terminals and connected to a 120-V AC rms distribution grid.
TABLE I
SIMULATION PARAMETERS
Fig. 6. Step response of closed-loop current controller.
19.5 to 10. Fig. 6 illustrates the time response of the system to

the changes in the input signal. It may be seen that id tracks the
idref . The step response of id is a first-order exponential function
that reaches its final value after about 0.05 s. As the control loops
along d-axis and q-axis are same, the step response of iq when
subjected to a step change in iqref would also be the same as that have been performed to verify the control system performance
of id response. for step changes in input, for SSCM. Demonstration of controller
performance when both active and reactive power is injected into
IV. CLOSED-LOOP CURRENT CONTROL OF SPCIS the grid is shown in Fig. 8.
Initial references for grid current along the d-axis, “idref” and
A. Simulations
along q-axis, “iqref” are set to zero till t = 0.5 s. At t = 0.5 s,
The system described in Fig. 1 has been modified to study “idref” is set to 17 A. Reference current along the q-axis, “iqref”
the dynamic performance of grid-connected SPCI. As shown is zero till t = 1 s. At t = 1 s, “iqref” is set to 5 A. It may be
in Fig. 7, the dc terminals of each H-bridge inverter module seen that the actual current along the q-axis, “iq” reaches the
is connected to an ideal dc voltage source. The magnitude of steady state in about 0.2 s. At t = 2 s, “iqref” is again changed to
dc voltage of each dc voltage source is set to 36.3 V (volt- 8 A, and system reaches the steady state in about 0.2 s. Fig. 9(a)
age at a maximum power point for the solar panel modules shows the simulation waveforms of the ac quantities measured
considered in Section V). AC grid voltage is set to 120-V ac on the grid side, when only active power is injected into the grid,
rms. The control scheme, as described in Section III, as well with “idref” = 17 A. It may be seen that the grid current, “Igrid”
as the control parameters obtained have been used to study is in phase with the grid voltage, “Vgrid.” “Vspci” is the 13-level
the dynamic performance of grid-connected SPCI, with ideal output voltage synthesized by the SPCI. Fig. 9(b) and (c) shows
dc voltage sources connected to the dc terminals. Simulation the simulation waveforms of ac quantities measured for reactive
parameters are described in Table I. The interfacing inductance power injection. It may be seen that the grid current “Igrid” is
and resistance values assumed in Section III have been used. not in phase with the grid voltage “Vgrid” because in this case,
As mentioned in Section II, the d and q components of along with active power, reactive power is also being injected
current are cross-coupled due to the presence of reactance terms. into the grid. The magnitude of the grid current should be the
This cross coupling can affect the dynamic performance of the resultant of magnitude of the active and reactive components
controller. Hence, decoupling the term is necessary to improve of current. Also, in all the three figures, it may be seen that the
the performance of the controller. Feedforward compensation output voltage of SPCI is different from the grid voltage in terms
has been used to reduce the cross-coupling affect. Simulations of both magnitude and phase angle. This observation is justified
TABLE II
THD OF GRID CURRENT MEASURED AT PCC
Fig. 8. Simulation waveforms showing the controller performance of grid-

connected SPCI system with ideal dc sources connected to the dc terminals of
each H-bridge module, when both active and reactive power into the grid are Fig. 10. Laboratory prototype of SPCIs with six SPCI modules capable of
controlled. synthesizing 120-V ac rms.
by (1), which signifies that “Vspci” is the phasor sum of “Vgrid”

and the voltage trop across the interfacing reactance.
B. THD of Grid Current

Harmonic analysis is performed, and total harmonic distortion
(THD) of the grid current measured at the PCC is calculated
using the FFT Analysis tool of MATLAB Simulink. Table II
lists the percentage THD of the grid current measured during
the periods 0 s–1 s, 1 s–2 s, and 2 s–3 s, respectively, for the
simulations performed in Section IV.A. The THD in all the
three cases did not exceed 5% limit as suggested by IEEE 1547
Standards.
C. Experimental Results Demonstrating Closed-Loop Current

Control of Grid-Tied SPCI
An experimental prototype has been constructed in the labo-
ratory for demonstrating the operating principle of SPCI. Fig. 10
shows the laboratory prototype capable of synthesizing a 120-V
ac rms voltage. Each SPCI is built using IGBT inverter modules.
Two F28035 Piccolo microcontrollers operating in synchronism
are used to control the six SPCIs. An interfacing inductance is
used to interface the SPCI terminals to the grid. Three dc power
supplies are used to connect to the dc terminals of SPCI. Each
dc power supply has two channels capable of generating up to
30-V dc. Each of these channels are operated in isolated mode
and are connected to each of the dc terminals of SPCI. A
Fig. 9. Simulation results showing ac waveforms on the grid side of SPCIs, single-phase autotransformer is used as an ac grid so that the
with ideal dc voltage sources connected to the dc terminals of each H-bridge grid voltage can be adjusted to the required voltage level. A
module. (a) Reference currents along d-axis and q-axis are set to 17 A and 0 A, closed-loop current control operation of the grid-tied SPCI is
respectively. (b) Reference currents along d-axis and q-axis are set to 17 A and
5 A, respectively. (c) Reference currents along d-axis and q-axis are set to 17 A performed on the experimental prototype. First, the ac grid
and 8 A, respectively. voltage is adjusted to about 97-V AC rms. The output voltage
TABLE III
SIMULATION PARAMETERS
TABLE IV
OPERATING SCENARIOS
Fig. 11. Experimental waveforms showing active power injection into the grid
when dc power supplies are connected at the dc terminals of SPCIs and interfaced
to the ac grid. Vgrid is the grid voltage, Vpsci is the output voltage of SPCI, and
Igrid is the grid current injected into the grid.
The output of the MPPT block is a direct axis component of the

reference grid current, “idref .”
A. Simulations Demonstrating MPPT Algorithm

MATLAB Simulink is used to perform simulations and
demonstrate the maximum power operation of solar panels. To
Fig. 12. Schematic of SPCIs interfaced to the grid at PCC with MPPT and
closed-loop current control implementation.
be able to synthesize a 120-V ac rms voltage, with a maximum
power throughput of 1.7 kW, six solar panels are connected to
six H-bridges. Table III lists the simulation parameters used for
the demonstration.
generated by each inverter is based on the principle of sort and The MPPT algorithm is implemented together with SSCM.
stack algorithm. So, the module that has more available power The simulations are performed for various scenarios of uniform
would generate more power. The dc voltages of each power and nonuniform irradiance. As described in Table IV, six rep-
supply are gradually increased to about 30-V dc. resentative operating scenarios for SSCM have been selected to
To inject active power into the grid, reference current along demonstrate the efficacy of the MPPT algorithm.
the d-axis is set to 3.5 A, and reference current along the q-axis Figs. 13–15 show simulation results corresponding to uniform
is set to zero. Experimental waveforms demonstrating the active irradiation conditions. Fig. 13(a) shows P–V characteristics of
power injection into the grid may be seen in Fig. 11. It may be the solar panel model when operating in scenario 1. This scenario
seen that the grid current (Igrid) is in phase with the grid voltage corresponds to low available power and can be considered as
(Vgrid). The grid current waveform looks distorted because of representation of a cloudy day. The portion highlighted in red
the poor quality of grid voltage available in the lab. of the P–V curve shows the steady-state operating points of all
solar panels, upon reaching the maximum power point, when
the SSCM strategy is employed. Fig. 13(b) shows waveforms of
V. MAXIMUM POWER POINT TRACKING
grid voltage Vgrid , output voltage Vspci , and grid current Igrid
A block schematic of the grid-connected SPCI with MPPT for six cycles when operating in scenario 1, with SSCM strategy
and closed-loop current control implementation is shown in implemented. Similarly, Fig. 14 and Fig. 15 show simulation
Fig. 12. It may be noticed that this system is like the one results for scenario 2 and scenario 3, respectively. It may be
described in Section III, except that there is an MPPT block. observed that the magnitude of the grid current in scenario 3 is
The MPPT algorithm, as described in [26], is implemented. This the highest when compared to scenario 1 and scenario 2, with
block takes in a direct axis component of the grid voltage “Vd ,” SSCM strategy. This is because the system is operating under
and direct axis component of the grid current “Id ” as the inputs. conditions where the power availability is maximum. Also, in
Fig. 13. Simulation results when all panels are operating under scenario 1
(uniform irradiation of 200 W/m2 ) with MPPT algorithm implemented for
SSCM. (a) P–V curve showing the maximum power point operation of all the
PV panels highlighted in red. (b) Simulation waveforms of grid voltage Vgrid
inverter output voltage Vspci, and grid current Igrid.
PV panels highlighted in red. (b) Simulation waveforms of grid voltage Vgrid,
(950 W/m2 , 800 W/m2 , 650 W/m2 , 500 W/m2 , 350 W/m2 , 200 W/m2 ) with
PV panels highlighted in red. (b) Simulation waveforms of grid voltage Vgrid,
MPPT algorithm implemented for SSCM. (a) P–V curves showing the maximum
power point operation of all the PV panels highlighted in red. (b) Simulation
waveforms of grid voltage Vgrid, inverter output voltage Vspci, and grid current
Igrid.
all the three scenarios corresponding to uniform irradiance,

current is injected into the grid at unity power factor. Figs. 16–18 waveforms of grid voltage Vgrid , output voltage Vspci , and grid
show simulation results corresponding to nonuniform irradiation current Igrid for six cycles when operating in scenario 4, with
conditions. Fig. 16(a) shows P–V characteristics of all solar SSCM strategy. Similarly, Fig. 17 and Fig. 18 show simulation
panels in operating scenario 4. This represents a significant results for scenario 5 and scenario 6, respectively. For the consid-
disparity of irradiation incident on all the solar panels. The ered nonuniform irradiance scenarios, as disparity in irradiance
portion highlighted in red of P–V curves in Fig. 16(a) shows the increases, available power in the system decreases (assuming
steady-state operating points of all solar panels, upon reaching 1000 W/m2 as the highest irradiance possible). Hence, it may be
the maximum power point, with the SSCM strategy imple- observed from Figs. 16(b), 17(b), and 18(b) that the magnitude
mented. It may be seen that all the panels are operating at of grid current increases as the disparity in irradiance decreases.
their corresponding maximum power points. Fig. 16(b) shows Also, the current is injected into the grid at unity power factor.
TABLE V
EFFECT OF SOLAR IRRADIATION ON THD OF THE CURRENT
Fig. 19. PV panels utilized for demonstrating maximum power point extrac-
Fig. 17. Simulation results when all panels are operating under scenario 5 tion and grid tie operation.
MPPT algorithm implemented for SSCM. (a) P–V curves showing the maximum
power point operation of all the PV panels highlighted in red. (b) Simulation
waveforms of grid voltage Vgrid, inverter output voltage Vspci, and grid current
Igrid.
Fig. 20. Experimental waveforms captured during mid-day, when available

power is maximum, demonstrate active power injection into the grid from SPCIs.
more, the THD is below 5% (which is the limit as suggested by

IEEE 1547 standards).
C. Experimental Results With Solar Panels Connected to the

DC Terminals
Three SPCI modules that have been used for the prototype
Fig. 18. Simulation results when all panels are operating under scenario 6 described in Section IV-B are utilized to connect to three of
MPPT algorithm implemented for SSCM. (a) P–V curves showing the maximum the PV panels as maybe seen in Fig. 19. Voltage across the PV
power point operation of all the PV panels highlighted in red. (b) Simulation terminals inside the lab, when operating at the maximum power
waveforms of grid voltage Vgrid, inverter output voltage Vspci, and grid current point (with full sunlight) is 23.75-V dc. Three of the panels
Igrid.
produce a total dc voltage of 71.25-V dc. Hence, the grid voltage
is chosen to be 45-V ac rms. The grid voltage is set to 45-V ac rms
using the single-phase autotransformer. The MPPT algorithm
B. THD of Grid Current With PV Panels Connected has been employed together with the closed-loop current control
Harmonic analysis is performed, and THD of the grid current technique to inject active power into the grid. Fig. 20 shows
measured at the PCC is calculated using the FFT Analysis tool of the experimental waveforms correspondingly captured during a
MATLAB Simulink. Table V describes the percentage THD for sunny mid-day. Vgrid is the grid voltage, Vspci is the output
the simulations performed in Section V-A. Except for scenario 1, voltage measured across the terminals of grid-connected SPCI,
the THD for remaining five scenarios is below 5%. Scenario 1 and Igrid is the grid current injected. An ac rms current of 8.67 A
has high THD because the available power in this scenario is is injected into the grid, in phase with the grid voltage having an
very less. Irrespective of whether the irradiation is uniform or ac rms voltage of 45. This corresponds to an average power of
nonuniform, when the available power ranges from moderate to about 392 W. The experiment is performed later in the afternoon,
REFERENCES
[1] M. Díez-Mediavilla, M. Dieste-Velasco, M. D. C. Rodríguez-Amigo,
T. García-Calderón, and C. Alonso-Tristán, “Performance of grid-tied PV
facilities based on real data in Spain: Central inverter versus string system,”
Energy Convers. Manage., vol. 86, pp. 1128–1133, 2014.
[2] Y. Yang and F. Blaabjerg, “Overview of single-phase grid-connected
photovoltaic systems,” Elect. Power Compon. Syst., vol. 43, no. 12,
pp. 1352–1363, 2015.
[3] S. Z. M. Noor, A. M. Omar, N. N. Mahzan, and I. R. Ibrahim, “A review
of single-phase single stage inverter topologies for photovoltaic system,”
in Proc. IEEE 4th Control Syst. Graduate Res. Colloq., 2013, pp. 69–74.
[4] H. Bergveld et al., “Module-level DC/DC conversion for photovoltaic
systems,” in Proc. IEEE 33rd Int. Telecommun. Energy Conf., 2011,
pp. 1–9.
Fig. 21. Experimental waveforms captured in the afternoon, when available [5] A. Elasser, M. Agamy, J. Sabate, R. Steigerwald, R. Fisher, and M.
power is less than that of mid-day, demonstrate active power injection into the Harfman-Todorovic, “A comparative study of central and distributed
grid from SPCIs. MPPT architectures for megawatt utility and large scale commercial
photovoltaic plants,” in Proc. 36th Annu. Conf. IEEE Ind. Electron. Soc.,
2010, pp. 2753–2758.
[6] L. Hassaine, E. OLias, J. Quintero, and V. Salas, “Overview of power
inverter topologies and control structures for grid connected photovoltaic
when the available power is less compared to that of the mid-day. systems,” Renewable Sustain. Energy Rev., vol. 30, pp. 796–807, 2014.
Fig. 21 shows the experimental waveforms captured when the [7] A. K. Abdelsalam, A. M. Massoud, S. Ahmed, and P. N. Enjeti,
available power is less. It may be seen that the rms value of ac “High-performance adaptive perturb and observe MPPT technique for
photovoltaic-based microgrids,” IEEE Trans. Power Electron., vol. 26,
current injected into the grid is 6.54 A, which is less than the no. 4, pp. 1010–1021, Apr. 2011.
current injected during the mid-day. The grid current waveforms [8] D. Sera, L. Mathe, T. Kerekes, S. V. Spataru, and R. Teodorescu, “On the
look distorted because of the poor quality of single-phase grid perturb-and-observe and incremental conductance MPPT methods for PV
systems,” IEEE J. Photovolt., vol. 3, no. 3, pp. 1070–1078, Jul. 2013.
voltage waveform in the laboratory. [9] M. A. Elgendy, B. Zahawi, and D. J. Atkinson, “Assessment of perturb
In Section IV, the capability of SPCI to connect to 97-V ac and observe MPPT algorithm implementation techniques for PV pumping
rms voltage has been demonstrated. By increasing the number applications,” IEEE Trans. Sustain. Energy, vol. 3, no. 1, pp. 21–33,
Jan. 2012.
of SPCI modules, the ability to connect to higher voltage ac [10] S. Alsadi and B. Alsayid, “Maximum power point tracking simulation for
grid increases. Additionally, there is an advantage that an output photovoltaic systems using perturb and observe algorithm,” Int. J. Eng.
having superior distortion properties can be produced. The only Innovative Technol., vol. 2, no. 6, pp. 80–85, 2012.
[11] M. Killi and S. Samanta, “Modified perturb and observe MPPT algorithm
requirement for the SPCI system to be applicable to higher for drift avoidance in photovoltaic systems,” IEEE Trans. Ind. Electron.,
voltages is that the isolated dc sources should be available. vol. 62, no. 9, pp. 5549–5559, Sep. 2015.
The solar panel is one of the isolated dc sources. SPCI being [12] G. J. Kish, J. J. Lee, and P. Lehn, “Modelling and control of photovoltaic
panels utilising the incremental conductance method for maximum power
a modular topology, the system can be extended to connect to point tracking,” IET Renewable Power Gener., vol. 6, no. 4, pp. 259–266,
higher voltage ac systems, by connecting several PV panels to 2012.
each of the SPCI and increase the output ac voltage to higher [13] A. Safari and S. Mekhilef, “Simulation and hardware implementation of
incremental conductance MPPT with direct control method using cuk
levels. converter,” IEEE Trans. Ind. Electron., vol. 58, no. 4, pp. 1154–1161,
Apr. 2011.
[14] Z. Xuesong, S. Daichun, M. Youjie, and C. Deshu, “The simulation
VI. CONCLUSION and design for MPPT of PV system based on incremental conductance
method,” in Proc. Int. Conf. Inf. Eng., 2010, vol. 2, pp. 314–317.
Dynamic analysis and closed-loop control design for grid- [15] A. Safari and S. Mekhilef, “Incremental conductance MPPT method
connected SPCI has been performed. Simulation results demon- for PV systems,” in Proc. 24th Can. Conf. Elect. Comput. Eng., 2011,
strating the dynamic response of the grid-connected SPCI with pp. 000345–000347.
[16] M. A. G. de Brito, L. Galotto, L. P. Sampaio, G. D. A. E. Melo, and C.
dc voltage sources connected across the dc terminals of SPCI A. Canesin, “Evaluation of the main MPPT techniques for photovoltaic
have been presented. Experimental results demonstrating the applications,” IEEE Trans. Ind. Electron., vol. 60, no. 3, pp. 1156–1167,
closed-loop current control operation of grid-connected SPCI Mar. 2013.
[17] J. Rocabert, A. Luna, F. Blaabjerg, and P. Rodriguez, “Control of power
(with six SPCIs) have been included as well in the article. MPPT converters in AC microgrids,” IEEE Trans. Power Electron., vol. 27, no. 11,
together with the closed-loop current control algorithm has pp. 4734–4749, Nov. 2012.
been used to demonstrate maximum power operation of SPCI. [18] S. Li, L. Xu, and T. A. Haskew, “Control of VSC-based STATCOM using
conventional and direct-current vector control strategies,” Int. J. Elect.
Simulations have been performed to demonstrate MPPT, by Power Energy Syst., vol. 45, no. 1, pp. 175–186, 2013.
employing SSCM. It has been demonstrated that the algorithm is [19] A. Mohd, E. Ortjohann, D. Morton, and O. Omari, “Review of control
able to extract maximum power from the system for both uniform techniques for inverters parallel operation,” Elect. Power Syst. Res., vol. 80,
no. 12, pp. 1477–1487, 2010.
and nonuniform irradiations. Experiments have been performed [20] A. R. Bergen, Power Systems Analysis. Englewood Cliffs, NJ, USA:
by connecting three real solar panels to a grid-connected SPCI Prentice-Hall, 1986.
prototype with three SPCIs. The MPPT algorithm has been [21] A. Yazdani and R. Iravani, Voltage-Sourced Converters in Power Systems:
Modeling, Control, and Applications. Hoboken, NJ, USA: Wiley, 2010.
implemented on the three SPCI setup, and experimental wave- [22] M. Monfared and S. Golestan, “Control strategies for single-phase grid
forms captured during two different instants in a day have been integration of small-scale renewable energy sources: A review,” Renewable
presented. Sustain. Energy Rev., vol. 16, no. 7, pp. 4982–4993, 2012.
[23] S. Golestan, M. Monfared, J. M. Guerrero, and M. Joorabian, “A DQ Prasanth Kumar Sahu received the B.E. degree in
synchronous frame controller for single-phase inverters,” in Proc. 2nd electrical and electronics engineering from Andhra
Power Electron., Drive Syst. Technologies Conf., 2011, pp. 317–323. University, Visakhapatnam, India, in 2010, and the
[24] H. Cha, T.-K. Vu, and J.-E. Kim, “Design and control of proportional- M.S. and Ph.D. degrees in electrical engineering from
resonant controller based photovoltaic power conditioning system,” in the University of North Carolina at Charlotte, Char-
Proc. Energy Convers. Congr. Expo., 2009, pp. 2198–2205. lotte, NC, USA, in 2014 and 2019, respectively.
[25] P. K. Sahu and M. Manjrekar, “Control strategies for solar panel com- His research interests include applications of power
panion inverters,” in Proc. Appl. Power Electron. Conf. Expo., 2015, electronics for utilities, grid integration of renewable
pp. 3075–3082. energy and energy storage systems, power electronic
[26] P. K. Sahu, M. Manjrekar, and R. Bhatt, “Maximum power point tracking drives for motor control, real-time simulation and
for solar panel companion inverters,” in Proc. IEEE 18th Workshop Control control of power electronic converters.
Modeling Power Electron., 2017, pp. 1–7.
[27] P. Salodkar, N. Sandeep, P. Kulkarni, and R. Udaykumar, “A comparison of
seven-level inverter topologies for multilevel DC-AC power conversion,”
in Proc. IEEE Int. Conf. Power Electron., Drives Energy Syst., 2014,
pp. 1–6.
[28] M. Khazraei, H. Sepahvand, K. Corzine, and M. Ferdowsi, “A general- Madhav D. Manjrekar (Senior Member, IEEE)
ized capacitor voltage balancing scheme for flying capacitor multilevel received the B.E. degree in instrumentation and con-
converters,” in Proc. 25th Annu. Appl. IEEE Power Electron. Conf. Expo., trol from the Government College of Engineering,
2010, pp. 58–62. Pune, India, the M. Tech. degree in electronic design
[29] P. K. Sahu and M. Manjrekar, “Investigations on power transfer operating and technology from the Indian Institute of Science,
spaces of solar panel companion inverters,” in Proc. 9th IEEE Int. Symp. Bangalore, India, the M.S. degree in electrical en-
Power Electron. Distributed Gener. Syst., 2018, pp. 1–8. gineering from Montana State University, Bozeman,
[30] P. K. Sahu and M. Manjrekar, “A control scheme for panel level maximum MT, USA, and the Ph.D. degree in electrical and com-
power extraction of solar panel companion inverters,” in Proc. IEEE puter engineering from the University of Wisconsin,
Energy Convers. Congr. Expo., 2018, pp. 2524–2530. Madison, Madison, WI, USA, in 1993, 1995, 1997,
[31] Y. Cao and L. M. Tolbert, “11-Level cascaded H-bridge grid-tied inverter and 1999, respectively.
interface with solar panels,” in Proc. 25th Annu. IEEE Appl. Power He is an Associate Professor with the Department of Electrical and Computer
Electron. Conf. Expo., 2010, pp. 968–972. Engineering, University of North Carolina in Charlotte, Charlotte, NC, USA. He
[32] V. Khadkikar, M. Singh, A. Chandra, and B. Singh, “Implementation of also serves as an Assistant Director of the Energy Production & Infrastructure
single-phase synchronous dq reference frame controller for shunt active Center (EPIC) at the University. He holds 10 U.S. and international patents, and
filter under distorted voltage condition,” in Proc. Joint Int. Conf. Power has authored and coauthored more than 60 journal and conference papers.
Electron., Drives Energy Systems 2010 Power India, 2010, pp. 1–6. Dr. Manjrekar is the recipient of the multiple IEEE Prize Paper Awards.

Controller Design and Implementation of Solar Panel Companion Inverters

Uploaded by

Copyright:

Available Formats

Controller Design and Implementation of Solar Panel Companion Inverters

Uploaded by

Document Information

Original Title

Copyright

Available Formats

Share this document

Share or Embed Document

Sharing Options

Did you find this document useful?

Is this content inappropriate?

Copyright:

Available Formats

Controller Design and Implementation of Solar Panel Companion Inverters

Uploaded by

Copyright:

Available Formats

IEEE TRANSACTIONS ON INDUSTRY APPLICATIONS, VOL. 56, NO.

2, MARCH/APRIL 2020 2001

Controller Design and Implementation of Solar

other hand, a PR controller can achieve a zero steady-state error

Vspcidq = (Vspcid + jVspciq ) (7) Id (s) 1

Loop gain becomes

Fig. 5. Bode plot for loop transfer function.

Fig. 7. Simplified circuit schematic of SPCIs, with ideal dc voltage sources

Fig. 6. Step response of closed-loop current controller.

19.5 to 10. Fig. 6 illustrates the time response of the system to

Fig. 8. Simulation waveforms showing the controller performance of grid-

by (1), which signifies that “Vspci” is the phasor sum of “Vgrid”

B. THD of Grid Current

C. Experimental Results Demonstrating Closed-Loop Current

The output of the MPPT block is a direct axis component of the

A. Simulations Demonstrating MPPT Algorithm

all the three scenarios corresponding to uniform irradiance,

Fig. 20. Experimental waveforms captured during mid-day, when available

more, the THD is below 5% (which is the limit as suggested by

C. Experimental Results With Solar Panels Connected to the

You might also like