Multifunctional Capability with Seamless Transition of

Three-Phase Grid using FLMF Control Approach

Vivek Narayanan, Member IEEE Seema, Member IEEE, Bhim Singh, Fellow IEEE
Electrical Engineering Electrical Engineering Electrical Engineering
IIT Delhi IIT Delhi IIT Delhi
[email protected] [email protected] [email protected]

Abstract—This paper presents a grid connected mode (GCM) and P&O method fails to track the peak power under those
standalone mode (SAM) of the photovoltaic (PV) - battery based conditions.
microgrid system. Whenever the grid fails, this system transfers When a solar PV array is operating in GCM, another issue to
from GCM to SAM without causing any interruption to the load be considered, is the power quality. If the distribution system is
supply. In the similar context, after the restoration of the grid, the
feeding nonlinear loads then the quality of grid currents,
system goes into GCM of operation. In the GCM, the voltage source
converter (VSC) of microgrid, operates in current control mode becomes poor. Moreover, due to the inductive nature of most of
(CCM) and in SAM, it operates in voltage control mode (VCM). A the loads, the power factor becomes further poor. So the power
bidirectional DC-DC converter (BDDC), controls the charging and quality improvement plays an important role in GCM. Several
discharging of the battery in GCM and in SAM. In SAM, the algorithms are available for the control of VSC and are reported
BDDC performs the extraction of PV array maximum power by in the literature [7-9]. Here, a fractional least mean fourth
regulating the DC link voltage to the PV array maximum power (FLMF) control is used for control of VSC. This is the one of the
point (MPP) voltage. In GCM, the VSC performs the extraction of algorithm from the adaptive control. It improves both steady
PV array maximum power. The BDDC is used for interconnecting state and dynamic responses in GCM of operation.
the solar PV array and the battery; hence the second order
A sudden failure of the grid is usually happened, while
harmonic is not fed to the battery so the effective life of the battery
is increased. The VSC performs the seamless transition from SAM disconnection of main grid, microgrid starts operating in SAM.
to GCM and vice versa and it also performs multi-functions such as Once the cause of disconnection is solved, the microgrid is
harmonics mitigation, balancing of grid currents and unity power resynchronized to the main grid and it operates in GCM. The
factor (UPF) operation in GCM. battery storage plays an important role in transition between
Keywords— Solar photovoltaic, battery, BDDC, power quality, GCM to SAM and vice versa [10]. It also handles the
SAM and GCM. intermittent variation of the solar power nature, since that
variation may lead to the instability of the utility grid. Moreover,
I. INTRODUCTION that functions as a backup power source in SAM. Because of
Nowadays the increased trends are going towards the sudden failure of the grid, during the transition between the
renewable energy sources based energy production due to the GCM to SAM, there may be large transient phenomenon
consideration towards the depletion of conventional sources and accompanied by oscillations and overshoots may occur. Several
environmental impacts [1]. Due to the ease of availability and control algorithms are developed for realising seamless
reducing trends in the cost of photovoltaic (PV) panels, the solar transition between the GCM to SAM and vice versa and are
energy has become good alternative for power generation reported in the literature [11-12].
compared to other renewable energy sources. The converter
plays a vital role in the integration of the energy generated from II. SYSTEM CONFIGURATION
the solar PV array to the electric grid [2]. Therefore, many Fig. 1 shows a configuration of proposed system. It consists
researchers are developing new converter systems for increasing of a solar PV array, a battery, a BDDC, a three-leg VSC, three-
the reliability and efficiency of the system. phase grid and nonlinear loads. A solid state switch is controlled
Mainly grid connected PV systems are classified into two to make the grid outage and to restore the grid.
types namely two stage configuration and a single stage
configuration and these are reported in the literature [3-4]. In III. CONTROL STRATEGY
two stage configuration, the PV array is integrated to the grid
through a DC-DC converter and a VSC. In this system, the DC- The control of proposed configuration is categorized into two
DC converter extracts peak PV power and VSC provides the sections: one is for the MPPT control and other one is for VSC
AC-DC conversion. In single stage configuration, the PV array control. The VSC control is divided into control for GCM and
directly feeds VSC and VSC performs the functions such as control for SAM. These control aspects are discussed here.
harvesting the solar PV maximum power and AC-DC A. MPPT Control
conversion. The overall efficiency is a critical in the system
One of the important consideration in this system, is the
design. As the single stage configuration avoids one DC-DC
extraction of maximum power of PV array irrespective VSC
conversion in the path between the PV array and VSC, so it
operates in CCM or VCM. Here an INC method is used as
contains less number of conversion stages, hence it provides
MPPT control for that purpose.
higher efficiency than a two stage configuration.
One of the important point to be considered in grid connected B. Control of VSC
PV system, is the extraction of peak power. Several algorithms The VSC control is categorized into two parts: the control of
are available for performing that task and these are compared in VSC in GCM and the control of VSC in SAM. These controls
the literature [5-6]. Among these controls, INC method is simple are described in following sections.
and even though the atmospheric condition changes quickly it
tracks the solar PV array peak power effectively whereas the
978-1-5386-4996-1/18/$31.00 ©2018 IEEE
Authorized licensed use limited to: Birla Institute of Technology. Downloaded on April 27,2023 at 05:17:31 UTC from IEEE Xplore. Restrictions apply.
 b) Calculation of Active Loss Components
In GCM of operation, the VSC performs the regulation of DC

Rf

Rf
link voltage. The MPPT control gives the reference DC voltage
(Vdc*) corresponding to the MPP voltage. This reference DC

Cf

Cf
voltage is compared with the sensed DC link voltage (Vdc). At
the jth instant it is given as,
Vde ( j ) = Vdc* ( j ) - Vdc ( j ) (4)
This error voltage is utilized for regulating the DC link voltage

ivsca
ivscb

ivscc
of VSC. That is achieved by the PI controller as,
wci ( j + 1) = wci ( j ) + K pd (Vde ( j + 1) - Vde ( j )) + K id Vde ( j + 1) (5)
Where, wci is the active loss component and Kid, Kpd are the gains
of the PI controller.
The weight of the PV array power is given as,
2 Ppv ( j )
w pv ( j ) = (6)
3Vt
Where, Ppv is the power output of a PV array.
c) Calculation of Fundamental Active Weights of Load
Currents.
The load current consists of active, reactive and harmonics
components. For generating the reference grid currents, it needs
to separate the fundamental active weight component from the
Fig. 1. Proposed multifunctional PV system load current.
1) VSC Control in GCM of Operation The fundamental active weight of load current of phase ‘a’ is
estimated by using FLMF control as [13],
The control of VSC in GCM, mainly focuses on generating
the unit templates from the sensed grid voltages, estimation of {
wia ( j + 1) = wia ( j ) + μ eia3 ( j )uia ( j ) β + (1 − β ) wia1−v ( j ) } (7)
active weight required for regulating the DC link voltage, Where, eia(j) is the error of adaptive component, µ is the step
extraction of fundamental active weight of load currents, size and β is the mixing parameter. The step size, µ decides the
generation of reference grid currents and switching signals steady state accuracy; a smaller step size µ is desirable for
generation. The structure of VSC control in GCM of operation, getting the lower steady state error.
is shown in Fig. 2. eia ( j ) = iLa ( j ) - uia ( j ) wia ( j ) (8)
a) Calculation of Unit Templates
Where, wia(j), iLa(j) and uia(j) are the active weight, load current
The grid phase voltages are generated from the sensed grid line and the in-phase unit template of ‘a’ phase, respectively at the jth
voltages (vgab, vgbc) as, instant.
2v gab + vgbc -vgab + vgbc -vgab - 2vgbc Similarly, the phase ‘b’ and ‘c’ reference active current
vga = , vgb = , vgc = (1)
3 3 3 components are estimated as,
The amplitude (Vt) of the grid voltage, is estimated as, {
wib ( j + 1) = wib ( j ) + μ eib3 ( j )uib ( j ) β + (1 − β ) wib1− v ( j ) }
(9)
2 2
Vt = 2
(vga + vgb 2
+ vgc ) (2) wic ( j + 1) = wic ( j ) + μ eic ( j )uic
3
( j ) {β + (1 − β ) w
ic
1− v
( j )} (10)
3
Knowing the grid phase voltages and amplitude of grid voltages, d) Estimation of Reference Grid Currents
The performance at UPF operation is achieved at the grid side.
the in-phase unit templates are estimated as,
In UPF operation, the reference grid currents are generated by
vga vgb vgc
uia = , uib = , uic = (3) utilising the total active weight. Total active weight (wgi) is
Vt Vt Vt estimated as,

Fig. 2. VSC control algorithm in GCM

Authorized licensed use limited to: Birla Institute of Technology. Downloaded on April 27,2023 at 05:17:31 UTC from IEEE Xplore. Restrictions apply.
wgi = wLi + wci - w pv (11)
Where,
wLi is the average active weight of load currents and it is given
as,
( w + wib + wic )
wLi = ia (12)
3 Fig. 5. BDDC control
wci is responsible for DC link voltage regulation and wpv is the On the restoration of the grid, the inputs of the AND gate
solar PV array weight. becomes 1, then the output of AND gate becomes 1 and it is
The reference in-phase grid currents are given as, given to the solid state switch, there by the grid is connected to
*
iga = wgi uia (13) the system. If any of the conditions not met, then the signal 0 be
the output of the AND gate, thereby disconnect the grid from the
*
i = wgi uib
gb (14) system. Every time, it checks the grid voltage magnitude, grid
*
i = wgi uic (15) frequency and the phase error. If any of them, goes beyond the
gc
prescribed value then a signal 0 goes to the switch and the
e) Generation of Switching Pulses system performs in the SAM of operation.
The comparison between the reference grid currents (iga*, igb*and
igc*) and the sensed grid currents (iga , igb and igc) yields error 4) BDDC Control
signals. Those error signals are fed to a hysteresis controller, The control of BDDC in GCM and SAM, is shown in Fig. 5.
which generates the switching signals for the VSC in GCM of In GCM, it is operating in the constant CCM and in SAM, it is
operation. operating to regulate the DC link voltage corresponding to the
2) Control in SAM of Operation MPP voltage. In GCM, the comparison between the reference
In the SAM, reference three phase load voltages are battery current (Ib*) and the sensed battery current (Ib), yields
generated as, error signal and this error signal is given to the PI controller and
the output of the PI controller is considered the duty cycle signal
2π 2π
vLa* = Vt * sin( wo t ), vLb* = Vt * sin( wo t − ), vLc* = Vt* sin( wo t + ) D2 that decides the switching of BDDC in GCM of operation. In
3 3 SAM, the reference DC link voltage (Vdc*) is compared with the
(16) sensed DC link voltage (Vdc) and the error signal is given to a PI
Where, Vt* is a magnitude of the reference voltages, and wo is the controller. The PI controller gives reference battery current. This
nominal frequency of 314 rad/sec. reference battery current is compared with the sensed battery
These reference voltages (vLa*, vLb*, and vLc*) are compared with current and the error is given to another PI controller. This PI
sensed load voltages (vLa, vLb, and vLc) and the resulting error controller output is considered the duty cycle signal D1 that
voltages are fed to the PI controller. The output of PI controller decides the switching of BDDC in SAM of operation.
is the reference load currents (iLa*, iLb* and iLc*). These reference
When signal SX=1, it is operates in GCM and when SX=0, it is
load currents are compared with the sensed load currents and the
operates in SAM. This duty cycle is compared with a saw-tooth
error signals are fed to the hysteresis controller for generating
wave, which gives the switching signal to the converter.
the switching signal to VSC. The block diagram of VSC control
in the SAM, is shown in Fig. 3. When the grid is restored, the IV. TEST RESULTS
reference voltage phase angle is replaced by θnew for the grid
synchronization. The generation of θnew signal is shown in Fig. 4 The performance of proposed system is studied for a grid
(a). voltage of 230 Vrms (line), 50 Hz in the laboratory.
3) Synchronization Control A. Steady State Performance at Nonlinear Load in GCM and in
The control for the synchronization, is depicted in Fig. 4 (b). SAM
If the grid voltage amplitude (Vt) is within the prescribed limit Fig. 6 shows the operational behaviour of proposed
then the signal 1 is given to the AND gate. Similarly the configuration in GCM at steady state condition. It includes the
difference between the phase angle θg and θL is given to the sine battery voltage (Vb), the battery current (Ib), battery power (Pb),
function. If sin (θg-θL) is within the limit, then another signal 1 is grid voltage (vg), grid current (ig), grid power, harmonic spectra
given to AND gate. Also, another signal 1 is given to the AND of grid current and grid voltage, VSC current (ivsc), VSC power,
gate, if the grid frequency (fg) lies within the limit. load current (iL), load power and the harmonic spectrum of load
current. The performance is studied for a solar irradiance of
1000W/m2. The VSC is controlled such that maximum power is
harvested from the PV array by regulating the DC link at MPP
voltage, balanced sinusoidal grid currents are obtained at UPF
(a) condition. When the grid fails, the BDDC regulates the DC link
voltage for extracting the PV array maximum power. The load
current is distorted and having THD of 27.3%. Even though the
load current is distorted, the quality of the grid currents are
improved by proper control of VSC. The harmonic levels of grid
voltage and grid current, are 1.9% and 4.8%, respectively. Those
values are within the IEEE-519 standard specified limit of
THDs. Fig. 7 shows the operational behaviour of proposed
(b) configuration in SAM of operation at the steady state condition.
Fig. 4 (a)-(b) Synchronization control

Authorized licensed use limited to: Birla Institute of Technology. Downloaded on April 27,2023 at 05:17:31 UTC from IEEE Xplore. Restrictions apply.
Fig. 3. VSC control algorithm in SAM
particular time duration, the solar irradiance is kept constant and
the battery is changed from charging mode to the discharging
mode of operation. When the battery starts discharging then the
excess power is fed to the grid. Even though with the solar
irradiance variations, the DC link voltage is maintained to the
MPP voltage, thereby the maximum power is effectively
harvested from the PV array. The VSC is controlled such that
(a) (b) (c) UPF grid currents are obtained.
C. Performance under Varying Irradiance in SAM
The performance at varying irradiance in SAM of operation,
is depicted in Fig. 9. Here the performance is studied for a
constant load. Fig. 9 includes PV current (Ipv), load voltage (vL),
battery current (Ib) and the load current (iL). Even though there is
disturbance in the solar irradiance, the load voltage is maintained
(d) (e) (f) at the reference value. During the reduction in the solar
irradiance, the battery discharging current is increased and
during an increase in the solar irradiance, the battery discharging
current is decreased and it is depicted in Fig. 9.
D. Performance at Load Perturbation in GCM
The operational behaviour of proposed configuration at load
perturbation in GCM, is depicted in Fig. 10 and Fig. 11.
(g) (h) (i)

(a) (b) (c)

(j) (k)
Fig. 6. Steady state performance in GCM under nonlinear load (a) battery voltage
and battery current, (b) battery power, (c) grid voltage and grid current, (d) grid
power, (e) harmonic spectrum of grid current, (f) harmonic spectrum of grid
voltage (g) VSC current, (h) VSC power, (i) load current, (j) load power, (k)
harmonic spectrum of load current
It includes the battery voltage (Vb), battery current (Ib), battery
power (Pb), load voltage, load current, load power and the
harmonic spectrum of the load voltage. In the SAM of operation, (d) (e)
the reference load voltages are generated by appropriate Fig. 7. Steady state performance in SAM under nonlinear load (a) battery voltage
controls. The THD of load voltage is observed about 4%, which and battery current, (b) battery power, (c) load voltage and load current, (d) load
power, (e) harmonic spectrum of load voltage
is within the IEEE-519 standard specified limit.
B. Performance at Varying Irradiance and Battery Current in
GCM
The performance of proposed configuration at varying irradiance
and varying battery current, is shown in Fig. 8. It shows PV
power (Ppv), grid power (Pg), load power (PL) and the battery
power (Pb). Here the load is kept constant. The variation in the
solar PV output power, causes the variation in the grid power.
The increase in solar irradiance, causes the power output from a
PV array to increase thereby the power that delivered to the grid,
is increased. Similarly the decrease in the solar irradiance causes
reduction in the power that delivered to the utility grid. After a
Fig. 8. Performance under solar irradiance, battery current perturbation in GCM

Authorized licensed use limited to: Birla Institute of Technology. Downloaded on April 27,2023 at 05:17:31 UTC from IEEE Xplore. Restrictions apply.
Fig. 10 includes PV power (Ppv), load power (PL), grid power
(Pg) and the battery power (Pb). Here the performance is studied
at constant irradiance. The variation in load power, causes the
variation in the grid power that is depicted in Fig. 10. Here the
battery is operated in constant current discharging mode. Hence
the PV array and the battery together, meet the load demand. If
the total power generation that is the sum of the PV power and
the battery discharged power, becomes excess the load demand
then the remaining power is fed to the grid otherwise the grid
provides remaining power to meet the load demand. Figs. 11 (a)-
(b) depict the performance of proposed configuration at
unbalanced nonlinear load. Fig. 11 (a) includes grid voltage (vg),
Fig. 9. Performance under varying solar irradiance in SAM
grid current (ig), load current (iL), and VSC current (ivsc). Fig. 11
(b) shows PV current (Ipv), PV voltage (Vpv), battery current (Ib)
and load current (iL). The grid currents are balanced even though
the load is unbalanced by the proper control of VSC. Even
though during the disturbance in load unbalance, the DC link is
well maintained to the MPP voltage. During the disturbance of
load unbalance, the active weight component of load (wLi)
reduces, so the surplus PV power is fed to the grid since the
generated PV power is higher than the load demand.
E. Seamless Transition of the Microgrid from GCM to SAM and
Vice Versa
Figs. 12 (a)-(b) depict the seamless transition capability of Fig. 10. Performance under load perturbation in GCM
proposed system from GCM to SAM and vice versa. Fig. 12 (a)
shows the load voltage (vL), grid voltage (vg), battery current (Ib) during the load unbalance. Fig. 14 shows that the oscillation in
and the grid current (ig). In the GCM, the load voltage is same as weight with proposed FLMF control is lesser than that with
the grid voltage. When the grid fails, the control is shifted LMF control. Fig. 15 shows that the settling time required for
automatically from the grid connected control to standalone proposed FLMF is less than the conventional LMF control,
control and the load voltage is maintained to the reference which requires increased settling time. Thus the proposed
voltage thereby not causing any interruption in supplying the control improves steady state performance along with the
load demand. During the GCM, the battery is operated in dynamic response.
charging mode and operates in discharging mode when the grid
fails. Fig. 12 (b) shows the load voltage (vL), grid voltage (vg), V. CONCLUSION
load phase angle (θL) and the grid phase angle (θg). When the The multifunctional capability of proposed system has been
grid fails, θg becomes zero, and the control is in SAM. When the investigated in this work. This system presents seamless
grid is restored, θL tracks θg for synchronization control. When transition from GCM to SAM when the grid fails and vice versa.
the grid voltage magnitude (Vt), the phase angle difference and The PV array maximum power is effectively harvested in both
the grid frequency (fg), become within the specified value, and GCM and SAM of operation. In this proposed system, instead of
then the main grid connects to the system. feeding the battery directly to the DC link, a BDDC is used in
Figs. 13 (a)-(b) show the functionality of the MPPT control. the path between the PV array and the battery, hence the battery
It shows the P-V and I-V characteristics of a PV array at is free from the second order harmonic that results in due to the
1000W/m2 and 500W/m2 irradiance. It shows that the PV array unbalanced load. Hence the battery life is increased. The
maximum power is effectively tracked with proper control. proposed FLMF control provides faster response than the
F. Comparative Analysis of the Proposed Control with conventional LMF control. The performance of the system at
Conventional LMF Control varying irradiance and under unbalanced nonlinear loads, has
demonstrated the effectiveness of proposed FLMF control.
Fig. 14 and Fig. 15 show the ‘a’ phase load current active
weight (wia) and the load average active weight (wLi) variations

(a) (b)
Fig. 11 (a-b) Performance under unbalanced load in GCM

Authorized licensed use limited to: Birla Institute of Technology. Downloaded on April 27,2023 at 05:17:31 UTC from IEEE Xplore. Restrictions apply.
 (a) (b)
Fig. 12 (a-b) Performance during the synchronization
[4] N. Beniwal, I. Hussain and B. Singh, “Hybrid VSS–LMS–LMF based
adaptive control of SPV-DSTATCOM system under distorted grid
conditions,” IET Ren. Pow. Gen., vol. 12, no. 3, pp. 311-322, 2018.
oscillation in weight [5] S. Amin, S. Khan and A. Qayoom, “Comparative analysis about the study
wia=2 A/div
of maximum power point tracking algorithms: A review,” 2018 Inter. Conf.
on Comp., Math. and Eng. Tech., 2018, pp. 1-8.
[6] A. Kchaou, A. Naamane, Y. Koubaa and N. K. M'Sirdi, “Comparative
wia=2 A/div study of different MPPT techniques for a stand-alone PV system,” 17th
LMF Inter. Conf. on Sci. and Tech. of Auto. Cont. and Comp. Eng., 2016, pp.
FLMF 629-634.
load rejection
iL=5 A/div [7] B. Singh, D. T. Shahani and A. K. Verma, “IRPT based control of a 50 kW
grid interfaced solar photovoltaic power generating system with power
load injection quality improvement,”4th IEEE Inter. Symp. on Pow. Elect. for Distr. Gen.
Sys., 2013, pp. 1-8.
Fig. 14. Comparison in response of wia signal for FLMF and LMF algorithms
[8] V. Narayanan and P. Jayaprakash, “Performance comparison of three-phase
load injection four-wire grid integrated solar photovoltaic system with various control
iL=5 A/div algorithms,” Inter. Conf. on Intel. Comp., Instr. and Cont. Tech., 2017, pp.
406-414.
settling time difference [9] R. K. Agarwal, I. Hussain and B. Singh, “LMF-Based Control Algorithm
for Single Stage Three-Phase Grid Integrated Solar PV System,” IEEE
wLi=2 A/div Trans. on Sust. Ene., vol. 7, no. 4, pp. 1379-1387, 2016.
[10] P. Arboleya et al., “Efficient energy management in smart micro-grids:
zero grid impact buildings,” IEEE Trans. Smart Grid, vol. 6, no. 2, pp.
1055–1063, 2015.
LMF [11] R. Bisht, S. Subramaniam, R. Bhattarai and S. Kamalasadan, “Active and
FLMF reactive power control of single phase inverter with seamless transfer
between grid-connected and islanded mode,” IEEE Pow. and Ene. Conf. at
Illinois, 2018, pp. 1-8.
Fig. 15. Comparison in response of wLi signal for FLMF and LMF algorithms [12] J. Zhang, Q. Wang, C. Hu and T. Rui, “A new control strategy of seamless
ACKNOWLEDGEMENT transfer between grid-connected and islanding operation for micro-grid,”
12th IEEE Conf. on Ind. Elect. and Appl., 2017, pp. 1729-1732.
The authors are very grateful to the Dept. of Science and Tech., [13] S. Khan, N. Ahmed, M. A. Malik, I. Naseem, R. Togneri and M.
Govt. of India, for funding this project under Grant Number: Bennamoun, “FLMF: Fractional least mean fourth algorithm for channel
RP03195 (FIST Proj.) and RP03443G (INDO-US Proj.). estimation in non-Gaussian environment,” Inter. Conf. on Inf. and Comm.
Tech. Conv., 2017, pp. 466-470.
REFERENCES APPENDICES
[1] E. Goutard, “Renewable energy resources in energy management PV array Vmpp= 380 V, Impp= 10.4 A, Pmpp= 4 kW
systems,” 2010 IEEE PES Innov. Smart Grid Tech. Conf. Europe, 2010, DC-link voltage Vdc= 380 V
pp. 1-6. DC link capacitor Cdc=3mF
[2] C. Chakraborty, H. H. C. Iu and D. Dah-Chuan Lu, “Power converters, Experimental Battery 240 V, 35 Ah
control, and energy management for distributed generation,” IEEE Trans. parameters Grid voltage VLL= 230V (rms)
on Ind. Elect., vol. 62, no. 7, pp. 4466-4470, 2015. Step size µ = 2e-4
[3] B. Singh, C. Jain and A. Bansal, “An Improved Adjustable Step Adaptive Mixing parameter β = 0.5
Neuron-Based Control Approach for the Grid-Supportive SPV System,”
IEEE Trans. on Ind. Appl., vol. 54, no. 1, pp. 563-570, 2018.

(a) (b)
Fig. 13. MPPT performance of PV array at (a) 1000 W/m2 (b) 500 W/m2

Authorized licensed use limited to: Birla Institute of Technology. Downloaded on April 27,2023 at 05:17:31 UTC from IEEE Xplore. Restrictions apply.