This article has been accepted for inclusion in a future issue of this journal.
Content is final as presented, with the exception of pagination.
IEEE TRANSACTIONS ON SUSTAINABLE ENERGY
A Three-Phase Grid Tied SPV System With Adaptive
DC Link Voltage for CPI Voltage Variations
Chinmay Jain, Member, IEEE, and Bhim Singh, Fellow, IEEE
Abstract—This paper deals with a three-phase two-stage grid
tied SPV (solar photo-voltaic) system. The first stage is a boost
converter, which serves the purpose of MPPT (maximum power
point tracking) and feeding the extracted solar energy to the DC
link of the PV inverter, whereas the second stage is a two-level
VSC (voltage source converter) serving as PV inverter which feeds
power from a boost converter into the grid. The proposed sys-
tem uses an adaptive DC link voltage which is made adaptive by
adjusting reference DC link voltage according to CPI (common
point of interconnection) voltage. The adaptive DC link voltage
control helps in the reduction of switching power losses. A feed for-
ward term for solar contribution is used to improve the dynamic
response. The system is tested considering realistic grid voltage
variations for under voltage and over voltage. The performance
improvement is verified experimentally. The proposed system is
advantageous not only in cases of frequent and sustained under
voltage (as in the cases of far radial ends of Indian grid) but also
in case of normal voltages at CPI. The THD (total harmonics dis-
tortion) of grid current has been found well under the limit of an
IEEE-519 standard.
Index Terms—Adaptive DC link, MPPT, overvoltage, solar PV,
two-stage, three phase, under voltage.
I. I NTRODUCTION
T HE electrical energy has a vital role in development of
human race in the last century. The diminishing conven-
tional primary sources for electricity production have posed an
energy scarcity condition in front of the world. The renewable
energy sources such as solar, wind, tidal etc are few of such
options which solve the problem of energy scarcity. The cost
effectiveness of any technology is prime factor for its commer-
cial success. The SPV (Solar Photovoltaic) systems have been
proposed long back but the costs of solar panels have hindered
the technology for long time, however the SPV systems are
reaching grid parity [1], [2].
The solar energy based systems can be classified into stan-
dalone and grid interfaced systems. The energy storage (con-
ventionally batteries) management is the key component of
standalone system. Various problems related to battery energy
storage standalone solar energy conversion systems are dis-
cussed in [3]–[6].
Considering the problems associated with energy storage
systems, the grid interfaced systems are more preferable, in
Manuscript received February 04, 2015; revised June 24, 2015; accepted
August 22, 2015. This work was supported by the Department of Science and
Technology (DST), Government of India, under Grant RP02583. Paper no.
TSTE-00078-2015.
The authors are with the Indian Institute of Technology Delhi, New Delhi
110016, India (e-mail: chinmay31jain@gmail.com; bsingh@ee.iitd.ac.in).
Color versions of one or more of the figures in this paper are available online
at http://ieeexplore.ieee.org.
Digital Object Identifier 10.1109/TSTE.2015.2496297
1949-3029 © 2015 IEEE. Personal use is permitted, but republication/redistribution requires IEEE permission.
See http://www.ieee.org/publications_standards/publications/rights/index.html for more information.
1
case the grid is present. The grid acts as an energy buffer, and
all the generated power can be fed into the grid. Several grid
interfaced SPV systems are proposed in past addressing various
issues related to islanding, intermittency, modeling etc [7]–[9].
With growing power system, the attention is moving from
centralized generation and radial distribution to distributed
generation. The distributed generation can bring in several
advantages such as reduction in losses, better utilization of dis-
tribution resources, load profile flattering etc [10]–[12]. The
SPV systems provide a good choice for distributed generation
system considering small scale generation from rooftop solar,
modularity of power converter and static energy conversion
process.
The initial investment in SPV systems is high because of
high cost of solar panels [13]. Therefore, considering the ini-
tial investments for any installed plant, the aim is to extract
maximum energy output from the given capacity.
To accomplish the objective of extraction of maximum
energy from an installed PV array several techniques are pro-
posed in the literature [14]–[18]. A review of MPPT (Maximum
Power Point Tracking) techniques is shown in [14]. An incre-
mental conductance (InC) based MPPT technique is shown in
[15]. An ANN based MPPT algorithm is shown in [16]. The
application of sliding mode controller to MPPT algorithm is
shown in [17]. A combination of fractional open circuit volt-
age and fuzzy based MPPT technique is shown in [18] wherein
a constant offset is added at the output of fuzzy controller to
improve the MPPT performance. The incremental conductance
based MPPT is fast, accurate and easy to implement. In this
paper, a composite InC based MPPT method is used. The com-
posite InC method is a combination of fractional Voc and InC
based method. The proposed MPPT technique limits the area of
search for MPP point hence improves the MPPT performance.
The tripping of the plant causes generation loss in case of
grid tied PV generation system. In general, grid tied VSCs
have under voltage and overvoltage protection. The nominal
range of set point for under voltage and over voltage is around
0.9 pu and 1.1 pu [19]. This range is very narrow because
of reasons such as converter may lose control, increase in
converter rating, and converter losses at low voltage etc. In
case of weak distribution system, a wide voltage variation is
observed. During peak loading condition, a sustained voltage
dip or under voltage is observed commonly. The practical range
of voltage variation is about ±15% of the nominal voltage.
Normally in such wide variation of distribution system the
shunt connected converter trips frequently. However, in case
of tripping of converter the PV generation is lost even when
PV power is available. Therefore, minimizing converter trips
 This article has been accepted for inclusion in a future issue of this journal. Content is final as presented, with the exception of pagination.
2 IEEE TRANSACTIONS ON SUSTAINABLE ENERGY
indirectly increases energy yield from the installed plant. The
proposed system is capable of working with wide range of
voltage variation hence avoids the generation loss.
The use of two stage SPV generation system has been
proposed by several researchers [20]–[23]. Conventionally a
DC-DC converter is used as first stage which serves the purpose
of MPPT. The duty ratio of DC-DC converter is so adjusted that
PV array operates at MPP point. The second stage is a grid tied
VSC (Voltage Source Converter) which feeds the power into
the distribution system. A single phase two stage grid tied PV
generation system with constant DC link voltage is shown in
[20]. Moreover, the three phase grid tied PV generation system
with constant DC link voltage control is also shown in [21],
[22]. The concept of loss reduction by adaptive DC link voltage
for VSC in hybrid filters is shown in [23], [24] wherein, the DC
link voltage is adjusted according to reactive power requirement
of filter. However, in the proposed system the DC link voltage
of VSC is made adaptive with respect to CPI voltage variation.
Moreover, the circuit topologies in both the systems are differ-
ent. Therefore, the work presented in [23], [24] is very different
from the proposed work.
For proper control of VSC currents, the DC link voltage ref-
erence is set more than peak of three phase line voltages. The
limitation for current control in single-phase grid connected
converter is shown in [25]. Considering the variation of CPI
(Common Point of Interconnection) voltage, the reference DC
link voltage is kept above the maximum allowable CPI voltage.
Therefore in case of fixed DC link voltage control for VSC, the
system always operates at a DC link voltage corresponding to
worst case condition.
In this paper, a simple control scheme is presented for grid
interfaced PV system with adaptive DC link voltage structure
for CPI voltage variation. A boost converter is used as the first
stage and a two level VSC is used as the second stage. Unlike
the earlier work with constant DC link voltage for VSC, the
presented work proposes an adaptive DC link voltage struc-
ture for the VSC and associated benefits. The adaptive DC link
voltage mainly reduces switching losses in all power devices
and high frequency ohmic losses in the interfacing inductor.
The maximum benefit of proposed DC link voltage structure
is found not only during under voltage (common in far radial
ends) but also under nominal grid voltage condition as the DC
link voltage is kept just necessary for proper current control
not according to worst case scenario. The claimed benefits of
the system are verified experimentally along with comparison
with conventional system. Moreover, the feed-forward term for
PV contribution is included to improve the dynamic response.
The PV feed forward term includes the effect of both CPI volt-
age variation and PV power variation. A linearised model of
DC link voltage control and effect of PV feed forward compo-
nent on the same is also analyzed in this paper. In the proposed
system both the input and output voltages of DC-DC boost con-
verter are adjusted in real time while keeping the objective of
MPPT intact. The performance of proposed control algorithm
is satisfactory under insolation change and sudden variation of
CPI voltage.
The two stage grid interfaced three phase systems are pro-
posed by several researchers, however, none of them have
Fig. 1. System configuration.
shown performance for such wide range of CPI voltage vari-
ations (350 V to 480 V for nominal of 415 V). The operation of
the system for a wide range of CPI voltage variation increases
the rating of the VSC and further the cost. However, small
increment in the cost of VSC can be justified on account of
large initial investment on PV array. The THD of grid current
and voltages has been found well under IEEE-519 standard
(less than 5%) under all operating conditions [26]. Moreover,
it should be noted that none of the ratings of the power devices
are compromised in the proposed system as compared to con-
ventional system, as in both the cases the ratings are decided
based on worst case scenario. Therefore, the system with pro-
posed control approach yields more energy output with the
same hardware resources.
II. S YSTEM C ONFIGURATION
The system configuration for the proposed system is shown
in Fig. 1. A two stage system is proposed for grid tied SPV
system. The first stage is a DC-DC boost converter serving for
MPPT and the second stage is a two-level three phase VSC.
The PV array is connected at the input of the boost converter
and its input voltage is controlled such that PV array deliv-
ers the maximum power at its output terminals. The output
of boost converter is connected to DC link of VSC. The DC
link voltage of VSC is dynamically adjusted by grid tied VSC
on the basis of CPI voltage. The three phase VSC consists of
three IGBT legs. The output terminals of VSC are connected to
interfacing inductors and the other end of interfacing inductors
are connected to CPI. A ripple filter is also connected at CPI
to absorb high frequency switching ripples generated by the
VSC. The values of various components and parameters used
in simulation and experimentation are given in Appendix A.
III. C ONTROL A PPROACH
The basic control approach for the SPV system is shown in
Fig. 2. The control of the system can be divided into two main
parts, which are control of the boost converter and control of a
grid tied VSC. The input voltage of a boost converter is adjusted
according to MPPT algorithm and the output voltage of boost
converter, which is also the DC link voltage of VSC is also
kept adaptive according to CPI voltage condition. In overall,
the proposed system is operated such that both the input and
output voltages of boost converter are adjusted according to
 This article has been accepted for inclusion in a future issue of this journal. Content is final as presented, with the exception of pagination.
JAIN AND SINGH: A THREE-PHASE GRID TIED SPV SYSTEM WITH ADAPTIVE DC LINK VOLTAGE
Fig. 2. Block diagram for control approach.
sensed variables of the circuit. The boost converter feeds the
power to the DC link of VSC, which then feeds that power into
the three-phase grid at unity power factor with respect to CPI.
A composite InC based MPPT technique is used to estimate the
reference PV array voltage and a PLL-less control is proposed
for the control of the VSC. The amplitude of the refrence grid
currents is estimated using a PV feedforward (PVFF) term and
a PI controller DC link voltage error. A set of unit vectors is
estimated from grid voltages to synchronize output currents of
VSC. The estimated reference grid currents are compared with
sensed grid currents and a hysteresis current controller is used
to generate switching logic for VSC. The detailed formulation
for control algorithm is presented in the later half of this section.
A. Maximum Power Point Tracking
A composite InC based MPPT algorithm is used. A range
of voltage for peak power is known with the knowledge from
fractional Voc MPPT, which is 0.7Vocmax to 0.9Vocmax , where
Voc is open circuit voltage and Vocmax is maximum open cir-
cuit voltage. The voltage for peak power is always searched in
this range for fast search of Vmpp . The InC algorithm works in
order to minimize the difference between the incremental con-
ductance and the conductance offered by the PV array. At first,
the reference PV array voltage is estimated based on the InC
principle then that reference voltage is used to estimate the duty
ratio of the boost converter. For calculation of an incremental
conductance ΔIP V and ΔVP V are estimated as,
ΔIP V = IP V (k) − IP V (k − 1)
ΔVP V = VP V (k) − VP V (k − 1)
where IPV (k) and VPV (k) are the instantaneous sampled current
and voltage of the solar array.
The governing equations for InC based MPPT algorithm
is as,
ΔIP V −IP V
= , at M P P (2a)
ΔVP V VP V
ΔIP V −IP V
> , Left of MPP onPP V v/s VP V curve (2b)
ΔVP V VP V
3
ΔIP V −IP V
< , Right of MPP on PP V v/sVP V curve (2c)
ΔVP V VP V
The reference PV array voltage (VP V ref ) is estimated using
(1) and (2). In order to keep the search in a limited region, the
VP V ref is bounded by corresponding upper and lower satura-
tion values (0.7Vocmax to 0.9Vocmax ). The output of that satura-
tion block is designated as new PV array voltage (VP V ref n ),
which is then used for estimation of reference duty ration.
The reference PV voltage (VP V ref n ) and sensed DC link
voltage (VDC ) are then used to estimate the duty ratio for the
boost converter. The governing equation for estimating duty
ratio is as,
VP V ref n (k)
Dref (k) = 1 − (3)
VDC (k)
This reference duty ratio is compared with saw-tooth wave-
form to generate switching logic for the boost converter.
B. Control Algorithm for VSC
The control approach for VSC is demonstrated in Fig. 2. The
main objective of the control algorithm for VSC is to regulate
the DC link voltage to the set reference value and to inject the
extracted power from PV array into the grid at unity power fac-
tor with respect to CPI. In order to control the output currents
of VSC (or grid currents), the appropriate reference grid cur-
rents are estimated. At first the amplitude of grid currents is
estimated and the estimated amplitude is then multiplied with
in-phase unit vectors (synchronization signals) to keep the grid
currents balanced and sinusoidal.
For control purpose, the CPI line voltages (vsab and vsbc ), DC
link voltage (vDC ) and grid currents (iga and igb ) are sensed.
The phase voltages are estimated from the line voltages. The
in-phase unit vectors are estimated from the estimated phase
voltages. For estimation of unit vectors (za , zb , zc ), the phase
voltages are divided by amplitude of three phase voltages (Vz ).
The amplitude of CPI voltage is estimated as,

2 + v2 + v2 )
2(vsa
sb sc
Vz = (4)
3
The unit vectors for all three phases are estimated as,
vsa vsb vsc
za = , zb = , zc = (5)
Vz Vz Vz
For proper control of VSC output current, the DC link volt-
(1a) age of VSC should be greater than amplitude of line voltage.
(1b) Moreover, in order to adaptively adjust the DC link voltage with
respect to CPI voltage variation, the reference DC link volt-
age is adaptively adjusted with amplitude of CPI voltage. The
reference DC link voltage is estimated as,
√
VDCref = µ 3VZ , where µ> 1 (6)
For proper current control the DC link voltage must be higher
than the amplitude of CPI line voltage. Therefore, reference DC
link voltage is kept around 10% higher than the peak of CPI
line voltage, considering the drop across switches, interfacing
 This article has been accepted for inclusion in a future issue of this journal. Content is final as presented, with the exception of pagination.
4 IEEE TRANSACTIONS ON SUSTAINABLE ENERGY
Fig. 3. Simulated performance for, (a) change in solar insolation without feedforward for PV contribution, (b) change in solar insolation with feedforward for PV
contribution, (c) normal to under voltage (415 V to 350 V), (d) CPI voltage variation from normal to over voltage (415 V to 480 V).
inductor, resistance of interfacing inductor and proper current
control under DC link voltage dynamics. Hence, the selected
value of µ in the proposed work is 1.1.
A PI controller is used to maintain the DC link voltage to
reference DC link voltage. The sensed DC link voltage is passed
through a low pass filter to suppress the switching noise. The
output of low pass filter is designated as VDC . The difference
between VDCref and VDC is the input to the PI controller.
The amplitude of grid currents consists of two main parts
which are the loss component and the contribution from the
PV array. The output of PI controller can considered as loss
component in steady state condition. The loss component is
estimated as,
Iloss (k) = Iloss (k−1) + Kp {ve (k) −ve (k−1)} + Ki ve (k)
(7)
A PV feed forward (PVFF) term for PV array contribution to
grid current is also estimated to provide fast dynamic response
for changes in solar insolation and grid voltages. The PV feed-
forward term is estimated as,
IP V g = (2PP V )/ (3VZ )
It can be observed from the above equation that in case of
insolation variation, the PV power (PPV = VPV ∗ IPV ) changes
and the instantaneous effect can be observed on PV contribution
term. Moreover, the in case of voltage variation at CPI, the grid
currents need to be adjusted to feed same solar power, which
eventually is adjusted due to term Vz in the PV contribution
term.
The grid currents are assumed coming out of CPI terminals
and considering the direction of grid currents, the losses are
drawn from the grid whereas the PVFF is fed into the grid.
Therefore, net amplitude of grid current is estimated as,
Irg = Iloss − IP V g
The estimated amplitude of grid current Irg is then multiplied
with unit vectors of corresponding phases to estimate reference
grid currents. The reference and sensed grid currents are then
given to current controller. The output of current controller is
the switching pulses to the VSC.
IV. S IMULATION R ESULTS
Fig. 3 shows the simulation results for different operating
condition. The simulation results for change in insolation level
and voltage variation at CPI are shown. The simulations are car-
ried out in MATLAB simulink and sim power system tool box.
The SPV array of 25 kW is considered for simulation study.
Other system parameters are given in Appendix.
A. Performance Under Sudden Change in Solar Insolation
Figs. 3(a)–(b) show performance of proposed system under
sudden change in insolation from 1000 W/m2 to 500 W/m2
with and without feed forward compensation respectively.
Before time t = 0.3 s, the system is working under steady state
condition with SPV insolation at 1000 W/m2 . The grid currents
are balanced and sinusoidal. At time t = 0.3 s, the insolation is
(8)
decreased from 1000 W/m2 to 500 W/m2 . The PV array cur-
rent decreases due to decrease in insolation and so is the PV
array power. It can be easily observed that the dynamic response
for sudden change in insolation level is better for proposed sys-
tem. The DC link voltage for only PI controller based system
shows more deviation and longer time to settle as compared to
proposed system with feed forward compensation based con-
trol approach. The system with proposed control approach soon
reaches the next state and it feeds the reduced power into the
grid. No appreciable effect is observed on the DC link voltage
of VSC.
(9) B. Performance for Under Voltage Operation
Fig. 3(c) shows the steady state and dynamics performance
of the system for under voltage operation at CPI. Before time
 This article has been accepted for inclusion in a future issue of this journal. Content is final as presented, with the exception of pagination.

JAIN AND SINGH: A THREE-PHASE GRID TIED SPV SYSTEM WITH ADAPTIVE DC LINK VOLTAGE 5

Fig. 4. Experimental data recorded by PV array simulator at (a) 1000 W/m2 ,

(b) 500 W/m2 .

t = 0.35 s, the system is operating at CPI voltage of 415 V. The

CPI voltage decreases from 415 to 350 V during 0.35 s to 0.4 s.
The adaptive nature of DC link voltage can be observed. The
DC link voltage also decreases with the decrease in CPI voltage.
The grid currents are maintained balanced and sinusoidal all the
time however, an increase in grid currents is observed to feed
the same PV power at reduced voltage. No appreciable effect is Fig. 5. Steady state performance (vsab with isa , isa harmonics spectrum,
observed on PV array voltage (vpv ), PV array current (ipv ), and vsab harmonics spectrum, power fed into grid respectively) for different CPI
voltage, (a)-(d) at 350 V, (e)-(f) at 415 V, (i)-(l) at 480 V.
PV array power (Ppv ).

C. Performance for Over Voltage Operation

Fig. 3(d) shows the steady state and dynamics performances
of the system for over voltage operation at CPI. Before time
t = 0.35 s, the system is operating at CPI voltage of 415 V.
The CPI voltage increases from 415 to 480 V during 0.35 s
to 0.4 s. The DC link voltage also increases with an increase in
CPI voltage, which shows the adaptive nature of DC link volt-
age. The grid currents are maintained balanced and sinusoidal
all the time however, a decrease in grid currents is observed to
feed the same PV power at the increased voltage. No appre-
ciable effect is observed on PV array voltage (vpv ), PV array
current (ipv ), and PV array power (Ppv ).

V. E XPERIMENTAL R ESULTS
The experimental results are shown in Figs. 4–7. A total
of seven quantities are sensed which are vpv , ipv , vDC , vsab ,
vsbc , iga , igb . The Hall-Effect voltage (LV-25) and current sen-
sors (LA-55p) are used. The sensed signals are conditioned
and given to ADC (Analog to Digital Converter) of controller
processor (dSPACE-1103). The outputs from controller are the
switching logics for all controlled switches. These results are
captured using power analyzer (Fluke 43B) and four chan-
nel digital storage oscilloscope (DSO6010A). Fig. 4 shows
readings  recorded by PV array simulator at different
 inso-
lation 1000 W/m2 and 500 W/m2 respectively . The MPPT
efficiency is found of order of 99.8%. Different other steady
state and transient performances are demonstrated in the latter
half of this section. The steady state results show the magni-
tude of various voltages, currents, powers and power quality
indices for full range of voltage. Moreover, the dynamic perfor-
mances for all salient disturbances (insolation change and CPI
voltage variation) are also demonstrated as important part of
Fig. 6. Dynamic performance for change in solar insolation level from
1000 W/m2 to 500 W/m2 and vice versa (a)-(b) decrease in insolation level,
(c)-(d) increase in insolation level.
experimental results. The experimental parameters are given in
Appendix A.
A. Steady State Performance Under Full Voltage Range
Fig. 5 shows the experimental results for complete range of
voltage variation under steady state condition at insolation level
of 1000 W/m2 . Figs. 5(a)–(d) show the operation of proposed
system at CPI voltage of 350 V. The CPI voltage (vsab ) and grid
current (iga ) are shown in Fig. 5(a). Only one of the grid cur-
rents is shown, as under steady state operation, the magnitude
of other two currents are equal hence they are not shown here.
Moreover, they can be observed in DSO waveforms in Fig. 6.
 This article has been accepted for inclusion in a future issue of this journal. Content is final as presented, with the exception of pagination.
6 IEEE TRANSACTIONS ON SUSTAINABLE ENERGY
Fig. 7. Dynamic performance for CPI voltage variation (a)-(b) decrease in
voltage from 415 V to 350 V, (c)-(d) increase in voltage from 415 V to 480 V.
Fig. 5(b) and (c) show the harmonic spectra of grid current (iga )
and CPI voltage (vsab ) respectively. The THD (Total Harmonic
Distortion) of grid current is found around 1.2% (within limit
of an IEEE-519 standard of 5%). Fig. 5(d) shows the power fed
into the grid. The negative sign of power shows that power is fed
into the grid and the DPF of −1 confirms the unity power fac-
tor operation with respect to CPI. Similarly Figs. 5(e)–(h) show
operation of proposed system at CPI voltage of 415 V. At nom-
inal voltage operation, the grid current is comparatively lesser
than under voltage operation, as the lesser current is required to
feed the same PV array power. The THD of grid current and CPI
voltage is well within limits under nominal voltage operation.
The power fed into the grid remains approximately the same.
Similarly Figs. 5(i)–(l) show the operation of proposed system
at CPI voltage of 480 V. At over voltage operation, the grid
current is comparatively lesser than under voltage and nominal
voltage operations. The power fed into the grid remains approx-
imately the same. The THDs of voltage and current are less than
5% under all these operating conditions.
B. Salient Dynamic Performance
The dynamic performance of proposed system is shown in
Figs. 6–7. Fig. 6 shows the performance of the system for
changes in solar insolation level. Figs. 6(a)–(b) show the perfor-
mance of the system for a decrease in solar insolation level from
1000 W/m2 to 500 W/m2 . Fig. 6(a) shows line voltage and grid
currents. The decrease in insolation causes the PV power to
decrease, which can be observed from decrease in magnitude
of grid currents. The grid currents are maintained balanced and
sinusoidal even under transition period. Fig. 6(b) shows other
salient variables such as DC link voltage, PV array voltage
and PV array current. The PV array current decreases due to
decrease in insolation level. A mark-able amount of power is
changed due to change in solar insolation however, no appre-
ciable effect is observed on the DC link voltage. Figs. 6(c)
and (d) show the performance for an increase in solar insola-
tion level from 500 W/m2 to 1000 W/m2 . Fig. 6(c) shows the
line voltage and three phase grid currents. A uniform increase
in grid current is observed for increasing insolation condition.
The grid currents are maintained balanced and sinusoidal even
under transition period. An increase in PV array current is also
observed during increasing insolation level. The DC link volt-
age is maintained well regulated at set reference value despite
of huge power change in incoming power at DC link of VSC.
Fig. 7 shows the performance of the system for variation
of CPI voltage. During this study, the PV array insolation is
kept constant at 1000 W/m2 . Figs. 7(a)–(b) show the perfor-
mance for CPI voltage changing from nominal to under voltage
(415 V to 350 V). Fig. 7(a) shows that the grid currents are
increased for decreased CPI voltage. The decrease in CPI volt-
age is conformed from decrease in magnitude of vsab as marked
in Fig. 7(a). However, the grid currents are maintained balanced
and sinusoidal throughout. Fig. 7(b) shows the DC link volt-
age and CPI voltage along with PV array voltage and current.
It can be easily observed that DC link voltage also decreases
with decrease in CPI voltage unlike constant DC link voltage
for all CPI voltage. This demonstrates the adaptive nature of
DC link voltage. No appreciable effect is observed on PV array
voltage and current. Figs. 7(c)–(d) show the performance for
CPI voltage changing from nominal to over voltage (415 V to
480 V). Fig. 7(c) shows that the grid currents are decreased
for increased CPI voltage. However, the grid currents are main-
tained balanced and sinusoidal throughout. The adaptive nature
of DC link voltage with CPI voltage can again be verified from
Fig. 7(d).
C. Comparison of Proposed System With Conventional System
The two-stage grid tied SPV systems have been proposed by
several researchers. Conventionally, the first stage is a DC-DC
converter which serves purpose of MPPT. The output of DC-
DC converter is connected to DC link of VSC. The DC link
voltage of VSC is kept constant and the reference DC link volt-
age value is decided on the basis of maximum CPI voltage. For
proper control of VSC currents, the DC link voltage should
be greater than the peak of maximum allowed grid voltage.
Therefore, the reference DC link voltage is kept around 10%
higher than the peak of maximum allowed grid voltage, consid-
ering the drop across interfacing inductor, switches and proper
current control under DC link voltage dynamics. Considering
a weak AC grid (far radial end) the over and under voltages
are quite common. Considering the range of voltage variation
of ±15%, the AC mains voltage may vary between 352 V to
477 V approximately for a nominal voltage of 415 V. The fixed
DC link voltage for operation under such a voltage range is
estimated as,
vDC = 1.414 ∗ Vg max ∗ 1.1 = 1.414 ∗ 477 ∗ 1.1
= 742.40 V  740 V
The adaptive DC link voltage structure helps in reducing the
power loss of the system. The two main parts of power loss
which get affected by DC link voltage are the switching losses
 This article has been accepted for inclusion in a future issue of this journal. Content is final as presented, with the exception of pagination.
JAIN AND SINGH: A THREE-PHASE GRID TIED SPV SYSTEM WITH ADAPTIVE DC LINK VOLTAGE
Fig. 8. Switching transient for shunt grid interfaced VSC.
Fig. 9. Basic principle for reduction in ripple current by keeping DC link
voltage near to amplitude of grid voltage.
and the loss in the interfacing inductor due to high frequency
ripple current. The basic principle and its effect on performance
are demonstrated experimentally follows.
Fig. 8 shows basic switching transients and power loss in
switching for any shunt grid interfaced VSC. It can easily be
observed that the switching loss in any shunt grid connected
system depends on DC link voltage of the VSC. In case when
fixed DC link voltage is selected for VSC, the switching losses
are higher not only during under voltage (quite common phe-
nomenon in rural grid) but also under nominal grid voltage.
The switching power losses in all 8 power switches (6 for
VSC and 2 for boost converter) are dependent on DC link volt-
age. Considering the above mentioned condition, an adaptive
DC link voltage based control for VSC is proposed to reduce
switching power losses. In the proposed adaptive control, the
DC link voltage is kept just sufficient for current control.
Another advantage of keeping the adaptive DC link voltage
is the reduction in high frequency ripple current in the inter-
facing inductor. Fig. 9 shows the basic principle for reduction
in ripple current by adaptive DC link voltage. The ripple in the
interfacing inductor depends on instantaneous voltage differ-
ence between the line voltage and the DC link voltage. For the
system with fixed DC link voltage based control structure the
instantaneous voltage difference across the interfacing induc-
tor is higher as compared to proposed DC link voltage control
structure. Therefore, high frequency ripple current is higher in
conventional system. Fig. 10 shows the experimental compari-
son between the current ripple with fixed and adaptive DC link
voltage structure. It can be easily observed that the high fre-
quency ripple content is higher in the system with fixed DC
link voltage structure. The high frequency ripple current is also
responsible high frequency ohmic losses. An overall experi-
mental comparison of losses between the two systems is shown
in Fig. 11.
In order to demonstrate the effectiveness of proposed control
algorithm, the experiments are carried out for both conven-
tional (fixed DC link voltage) and proposed (adaptive DC link
7
Fig. 10. Grid currents for phase a with (a) conventional DC link voltage
structure, (b) proposed DC link voltage structure.
Fig. 11. A comparision of power fed into grid at different CPI voltage by fixed
and adaptive DC link based system.
voltage) method keeping the same power from solar array emu-
lator. Fig. 11 shows the experimental comparisons of system
with fixed and adaptive DC link voltages. The solar insola-
tion is kept at 1000 W/m2 for all these operating conditions.
The fixed DC link voltage is kept at 740 V. It can be observed
that during under voltage (350 V) the power fed into the grid
by a system with adaptive DC link voltage is approximately
6 kW whereas the system using fixed DC link voltage feeds
approximately 5.5 kW. Moreover, even under nominal voltage
condition (415 V) the system with adaptive DC link voltage
feeds more power into the grid. The power fed at 480 V is
almost equal as the DC link voltage for both the systems are
almost equal.
VI. C ONCLUSION
A two-stage system has been proposed for three-phase grid
connected solar PV generation. A composite InC based MPPT
algorithm is used for control of the boost converter. The per-
formance of proposed system has been demonstrated for wide
range of CPI voltage variation. A simple and novel adaptive
DC link voltage control approach has been proposed for con-
trol of grid tied VSC. The DC link voltage is made adaptive
with respect to CPI voltage which helps in reduction of losses
in the system. Moreover, a PV array feed forward term is used
which helps in fast dynamic response. An approximate linear
model of DC link voltage control loop has been developed and
analyzed considering feed forward compensation. The PV array
feed forward term is so selected that it is to accommodate for
change in PV power as well as for CPI voltage variation. A
full voltage and considerable power level prototype has verified
the proposed concept. The concept of adaptive DC link voltage
 This article has been accepted for inclusion in a future issue of this journal. Content is final as presented, with the exception of pagination.
8 IEEE TRANSACTIONS ON SUSTAINABLE ENERGY
has been proposed for grid tied VSC for PV application how-
ever, the same concept can be extended for all shunt connected
grid interfaced devices such as, STATCOM, D-STATCOM etc.
The proposed system yields increased energy output using the
same hardware resources just by virtue of difference in DC link
voltage control structure. The THDs of the grid currents and
voltages are found less than 5% (within IEEE-519 standard).
The simulation and experimental results have confirmed the
feasibility of proposed control algorithm.
A PPENDIX
Parameters for simulation: Three-phase grid voltage
415 V, frequency = 50 Hz, supply inductance = 2.42 mH
and supply resistance = 0.76 Ω, interfacing inductor = 4 mH,
ripple filter R = 5 Ω, C = 5 µF, Kp = 1, Ki = 21, PV array
open circuit voltage: 500 V, PV array short circuit current:
70 A, PV array peak power: 25 kW, µ = 1.1.
Experimental Parameters: Three-phase grid voltage 415 V,
frequency = 50 Hz, interfacing inductor = 6 mH, ripple filter
R = 5 Ω, C = 5 µF, Kp = 1, Ki = 0.1, PV array open circuit
voltage: 500 V, PV array short circuit current: 16 A, PV array
peak power: 6.5 kW, µ = 1.1.
R EFERENCES
[1] M. Pavan and V. Lughi, “Grid parity in the Italian commercial and
industrial electricity market,” in Proc. Int. Conf. Clean Elect. Power
(ICCEP’13), 2013, pp. 332–335.
[2] M. Delfanti, V. Olivieri, B. Erkut, and G. A. Turturro, “Reaching PV grid
parity: LCOE analysis for the Italian framework,” in Proc. 22nd Int. Conf.
Exhib. Elect. Distrib. (CIRED’13), 2013, pp. 1–4.
[3] H. Wang and D. Zhang, “The stand-alone PV generation system with par-
allel battery charger,” in Proc. Int. Conf. Elect. Control Eng. (ICECE’10),
2010, pp. 4450–4453.
[4] M. Kolhe, “Techno-economic optimum sizing of a stand-alone solar
photovoltaic system,” IEEE Trans. Energy Convers., vol. 24, no. 2,
pp. 511–519, Jun. 2009.
[5] D. Debnath and K. Chatterjee, “A two stage solar photovoltaic based
stand alone scheme having battery as energy storage element for rural
deployment,” IEEE Trans. Ind. Electron., vol. 62, no. 7, pp. 4148–4157,
Jul. 2015.
[6] S. Krithiga and N. G. Ammasai Gounden, “Power electronic configu-
ration for the operation of PV system in combined grid-connected and
stand-alone modes,” IET Power Electron., vol. 7, no. 3, pp. 640–647,
2014.
[7] I. J. Balaguer-Álvarez and E. I. Ortiz-Rivera, “Survey of distributed gen-
eration islanding detection methods,” IEEE Latin Amer. Trans., vol. 8,
no. 5, pp. 565–570, Sep. 2010.
[8] C. A. Hill, M. C. Such, D. Chen, J. Gonzalez, and W. M. Grady, “Battery
energy storage for enabling integration of distributed solar power gen-
eration,” IEEE Trans. Smart Grid, vol. 3, no. 2, pp. 850–857, Jun.
2012.
[9] W. Xiao, F. F. Edwin, G. Spagnuolo, and J. Jatskevich, “Efficient
approaches for modeling and simulating photovoltaic power systems,”
IEEE J. Photovoltaics, vol. 3, no. 1, pp. 500–508, Jan. 2013.
[10] P. Chiradeja, “Benefit of distributed generation: A line loss reduction
analysis,” in Proc. IEEE/PES Transmiss. Distrib. Conf. Exhib.: Asia Pac.,
2005, pp. 1–5.
[11] A. Yadav and L. Srivastava, “Optimal placement of distributed gener-
ation: An overview and key issues,” in Proc. Int. Conf. Power Signals
Control Comput. (EPSCICON’14), 2014, pp. 1–6.
[12] K. A. Joshi and N. M. Pindoriya, “Impact investigation of rooftop Solar
PV system: A case study in India,” in Proc. 3rd IEEE PES Int. Conf.
Exhib. Innovative Smart Grid Technol. (ISGT Europe), 2012, pp. 1–8.
[13] E. Drury, T. Jenkin, D. Jordan, and R. Margolis, “Photovoltaic investment
risk and uncertainty for residential customers,” IEEE J. Photovoltaics,
vol. 4, no. 1, pp. 278–284, Jan. 2014.
[14] B. Subudhi and R. Pradhan, “A comparative study on maximum power
point tracking techniques for photovoltaic power systems,” IEEE Trans.
Sustain. Energy, vol. 4, no. 1, pp. 89–98, Jan. 2013.
[15] F. Liu, S. Duan, F. Liu, B. Liu, and Y. Kang, “A variable step size INC
MPPT method for PV systems,” IEEE Trans. Ind. Electron., vol. 55, no. 7,
pp. 2622–2628, Jul. 2008.
[16] M. Farhat, A. Flah, and L. Sbita, “Photovoltaic maximum power point
tracking based on ANN control,” Int. Rev. Model. Simul., vol. 7, no. 3,
pp. 474–480, 2014.
[17] J. Ghazanfari and M. M. Farsangi, “Maximum power point tracking using
sliding mode control for photovoltaic array,” Iran. J. Elect. Electron.
Eng., vol. 9, no. 3, pp. 189–196, 2013.
[18] M. Adly, H. El-Sherif, and M. Ibrahim, “Maximum power point tracker
for a PV cell using a fuzzy agent adapted by the fractional open circuit
voltage technique,” in Proc. IEEE Int. Conf. Fuzzy Syst. (FUZZ’11), 2011,
pp. 1918–1922.
[19] M. E. Ropp, J. G. Cleary, and B. Enayati, “High penetration and anti-
islanding analysis of multi-single phase inverters in an apartment com-
plex,” in Proc. IEEE Conf. Innovative Technol. Efficient Reliab. Electr.
Supply (CITRES’10), 2010, pp. 102–109.
[20] S. Deo, C. Jain, and B. Singh “A PLL-Less scheme for single-phase grid
interfaced load compensating solar PV generation system,” IEEE Trans.
Ind. Informat., vol. 11, no. 3, pp. 692–699, Jun. 2015.
[21] B. Singh, D. T. Shahani, and A. K. Verma, “Neural network con-
trolled grid interfaced solar photovoltaic power generation,” IET Power
Electron., vol. 7, no. 3, pp. 614–626, Jul. 2013.
[22] C. Jain and B. Singh “A frequency shifter based simple control for multi-
functional solar PV grid interfaced system,” in Proc. 37th Nat. Syst. Conf.,
2013, pp. 363–374.
[23] C.-S. Lam, W.-H. Choi, M.-C. Wong, and Y.-D. Han, “Adaptive DC-Link
voltage-controlled hybrid active power filters for reactive power compen-
sation,” IEEE Trans. Power Electron., vol. 27, no. 4, pp. 1758–1772, Apr.
2012.
[24] C.-S. Lam, M.-C. Wong, W.-H. Choi, X.-X. Cui, H.-M. Mei, and
J.-Z. Liu, “Design and performance of an adaptive low-DC-voltage-
controlled LC-hybrid active power filter with a neutral inductor in three-
phase four-wire power systems,” IEEE Trans. Ind. Electron., vol. 61,
no. 6, pp. 2635–2647, Jun. 2014.
[25] Y.-M. Chen, H.-C. Wu, Y.-C. Chen, K.-Y. Lee, and S.-S. Shyu, “The AC
line current regulation strategy for the grid-connected PV system,” IEEE
Trans. Power Electron., vol. 25, no. 1, pp. 209–218, Jan. 2010.
[26] IEEE Recommended Practices and Requirement for Harmonic Control
on Electric Power System, IEEE Standard 519, 1992.
Chinmay Jain (M’12) received the B.E. degree
in electrical engineering from Shri G. S. Institute
of Technology and Science, Indore, India, and the
M.E. degree in electrical engineering from Indian
Institute of Science, Bangalore, India, in 2009 and
2011, respectively. He is currently pursuing the Ph.D.
degree at the Department of Electrical Engineering,
Indian Institute of Technology, Delhi, India. His
research interests include power electronics, power
quality, grid interfaced solar PV systems, and design
of custom power devices.
Bhim Singh (SM’99–F’10) received the B.E. degree
in electrical engineering from the University of
Roorkee, Roorkee, India, and the M.Tech. and Ph.D.
degrees from the Indian Institute of Technology
(IIT) Delhi, New Delhi, India, in 1977, 1979, and
1983, respectively. In 1983, he joined, as a Lecturer,
the Electrical Engineering Department, University of
Roorkee, where he became a Reader in 1988. In
December 1990, he joined, as an Assistant Professor,
the Electrical Engineering Department, IIT Delhi,
where he became an Associate Professor in 1994, and
has been a Professor since 1997. His research interests include power electron-
ics, electrical machines and drives, renewable energy systems, active filters,
flexible ac transmission systems, high-voltage direct current, and power quality.
He is a Fellow of the Indian National Academy of Engineering; The National
Academy of Sciences, India; the Indian Academy of Sciences; The Institution
of Engineering and Technology, U.K.; The Institution of Engineers (India);
and the Institution of Electronics and Telecommunication Engineers and a Life
Member of the Indian Society for Technical Education and the System Society
of India.