Electrical Power and Energy Systems 100 (2018) 517–530

Contents lists available at ScienceDirect

Electrical Power and Energy Systems

journal homepage: www.elsevier.com/locate/ijepes

Fault analysis of solar photovoltaic penetrated distribution systems T

including overcurrent relays in presence of fluctuations
⁎
Browh Serge Tekpeti , Xiaoning Kang, Xinghua Huang
Shaanxi Key Laboratory of Smart Grid, Xi’an Jiaotong University, Xi’an 710049, China

A R T I C LE I N FO A B S T R A C T

Keywords: Intermittence from solar photovoltaic (PV) source can deteriorate overcurrent relays (OCRs) response to fault.
Accurate photovoltaic (PV) system This paper proposes a method for fault analysis with OCRs consideration to discuss relays faults currents fluc-
Battery Energy Storage System tuations in time against PV fluctuations. The simulated PV fluctuations are short time dynamic high and low
Fault analysis irradiations fluctuations and PV connection and disconnection based PV-dominated feeders. Two distribution
Photovoltaic (PV) fluctuation
systems (DSs) based accurate PV systems are modeled separately with and without Battery Energy Storage
overcurrent relay (OCR) fault current
System (BESS) for the purpose of proposed method. Series and parallels PV modules are arranged to model
accurate 6.5 kV PV farm power. Two level three-phase voltage source inverter (VSI), DC-DC boost converter and
LCL filter are applied for each PV system. Precise detailed control strategies are implemented in PSCAD software.
This paper proposed method complies with congestion which is frequently avoided in literature. The results
discussions based proposed method clarified PV fluctuations scenarios impacts on OCRs dynamic fault currents
profiles in presence and absence of BESS and clearly revealed and argued related important findings. In addition,
OCRs performance is shortly analyzed based particle swarm optimization (PSO) algorithm efficiently coded in
Matlab. The obtained results could be very profitable to power system relaying operators to handle protection
challenges with PV power fluctuations.

1. Introduction distribution feeders with high penetration of PV systems; accom-

Attention is needed towards PV fluctuations to conveniently address
relay malfunctions in electrical power system. Therefore, DSs fault
analysis involving OCRs against PV fluctuations scenarios such as dy-
namic irradiation fluctuations and PV connection and disconnection is
worth investigating, and this paper focuses on that aspect using a new
approach.
Mitigating PV fluctuations impact on DS protection is much needed.
Namely, research articles, standards [1,2] and books [3,4] are pro-
gressively contributing to that goal. Renewable generation impact on
protection systems [5,6]; fault current limiters to sustain OCRs with
different distributed generators (DGs) [7]; renewable energy and var-
ious energy storage systems [8–13]; low voltage ride through (LVRT)
capability [10,14,16]; fault current contribution from large PV systems
[17]; innovative control strategies [8–11,18–20]; smoothing PV power
output fluctuations [8,21]; power system frequency restoration [22];
mitigating temporary over-voltages [23] and PV grid connected systems
performance and fault analysis [12,19,24–29] are some of the pro-
mising works achieved.
Moreover, the authors in [26,27] discussed, fault analysis on
modating high PV penetration on distribution feeders, respectively,
where in [26], a method is proposed to estimate OCR operating time in
PV-dominated distribution feeder. In [30], a numerical comparison of
the effects of different types of DG units on OC protection systems in
MV distribution grids is proposed. In [31], a new protective method for
grid connected dispersed PV systems is developed to detect short circuit
fault in distribution line. Procedures for fault location and isolation to
solve protection selectivity problems in MV distribution networks with
dispersed generation is carried out in [32]. A method for evaluation of
OC protection in unbalanced DSs is available in [33].
In this paper based PV-dominated distribution feeders, we in-
vestigate OCRs maximum faults currents fluctuations analysis against
short time fault while considering solar radiation fluctuations and PVs
connection and disconnection in time. A method is proposed to address
OCRs dynamic faults currents profiles analysis in time. In addition,
other relevant aspects are carried out to highlight important findings.
Complex and accurate DSs models with and without BESS are em-
ployed in this paper to investigate fault analysis against dynamic high
and low irradiation fluctuation making use of OCRs in a technically
simple and special manner not proposed before. OCRs fault currents

⁎
Corresponding author.
E-mail addresses: sergetekpeti@stu.xjtu.edu.cn, tekpetiserge@yahoo.fr (B.S. Tekpeti).

https://doi.org/10.1016/j.ijepes.2018.03.003
Received 1 December 2017; Received in revised form 18 February 2018; Accepted 1 March 2018
Available online 13 March 2018
0142-0615/ © 2018 Elsevier Ltd. All rights reserved.
B.S. Tekpeti et al. Electrical Power and Energy Systems 100 (2018) 517–530

profiles in time are clearly analyzed based comparison in an interesting ⎡id ⎤ ⎡ cos(θ) cos(θ−2π /3) cos(θ + 2π /3) ⎤ ⎡ia ⎤
way. Few relays are implemented to ease analysis when taking into ⎢ iq ⎥ = 2/3 × ⎢− sin(θ) − sin(θ−2π /3) − sin(θ + 2π /3) ⎥ × ⎢ ib ⎥
account of all the relays in time based distinct PV fluctuations scenarios. ⎢i ⎥ ⎢ ⎥ ⎢i ⎥
⎣ 0⎦ ⎣ 1/ 2 1/ 2 1/ 2 ⎦ ⎣ c⎦
Three lines to ground (3L-G) symmetrical fault which is a severe fault in
(1)
power system is applied. Usually, inaccurate models are developed,
which this work overcomes as it mainly uses self-built power conver- v ⎡ cos(θ) ⎤
sion facilities and control strategies. In fact, it is obvious that, the more ⎡ a⎤
vabc = ⎢ vb ⎥ = vm × ⎢ cos(θ−2π /3) ⎥
⎢ ⎥
accurate the system the better the results precision. The control stra- ⎣ vc ⎦ ⎣ cos(θ + 2π /3) ⎦ (2)
tegies are dq current control and voltage mode PQ control based syn-
chronous rotating frame control strategy; Incremental Conductance (IC) The grid active power (P ) and reactive power (Q) follow (3) when vq is
maximum power point tracking (MPPT); phase locked loop (PLL); pulse not null, and (4) if vq = 0 due to PLL which assures θ synchronization.
width modulation (PWM); sinusoidal pulse width modulation (SPWM) PLL is a negative feedback control, with details available in [4].
and BESS control. OCRs performance is analyzed based PSO algorithm
⎧ P = vd id + vq iq
as case study.
The remainder of this paper is organized as follows. Section 2 ex- ⎩Q = vd iq + vq id
⎨ (3)
poses PV systems modeling with control strategies details and relevant
explanations including emphasizing this work goal. In Section 3, PV ⎧ P = vd id
⎨ Q = vd iq (4)
penetrated DSs models are explained to show this paper modeling ⎩
quality. Section 4 argues solar radiation fluctuation impact on PV where vd and vq are direct and quadrature voltages components, re-
system performance in distinct manners to highlight important findings spectively.The dq current control in rotating reference frame takes dq-
with descriptive figures. Section 5 provides the proposed simulation ∗
frame voltages (vdq ), currents (idq ) and dq-frame reference currents (idq )
method details. Section 6 gives PSCAD simulations results and provides from voltage mode PQ control to produce dq-frame reference voltages
∗
convincing discussions and comments. Section 7 presents relays per- (vdq ). Voltage mode PQ control is outer control loop whereas dq current
formance analysis based PSO algorithm. All the results are provided in control is inner control loop. Joined both control loops constitutes
form of Tables and figures. Section 8 concludes this paper. synchronous rotating frame structure.

2.2. Voltage mode PQ control

2. PV systems modeling description
DC bus voltage (vdc ) and reactive power (Q) PI controls are in (5).
The PV systems with and without BESS in Figs. 1 and 2, respec-
tively, apply to this paper proposed fault analysis method. ∗ 1
∗
⎧id = (Kpd1 + Kid1 × s )(vdc−vdc )
Figs. 1 and 2 similarly show 6.5 kV PV farm, DC-DC boost converter,
⎨iq∗ = −(Kpq2 + Kiq2 × 1 )(Qdc
∗
−Q)
two level VSI, LCL filter, IC MPPT with PWM, voltage mode PQ control, ⎩ s (5)
dq current control leading to SPWM, direct and inverse park transfor-
where Kp and Ki are proportional and integral gains of the controllers
mations, PLL, and 100 MVA grid linked to two delta-star transformers. ∗
(PI1 and PI2), whereas vdc and Q∗ are vdc and Q references, respectively.
However, Fig. 1 shows BESS with buck-boost converter interface. The
The variable id∗ also determines active power (P ) control. In other
PV farm voltage and current are vpv and ipv respectively, whereas other
words, voltage mode PQ control assures DC bus voltage control, active
appearing variables are defined further.
power control and reactive power control. Precisely, the control loop
outputs id∗ for both DC bus voltage control and active power control, and
2.1. Synchronous rotating frame control strategy generates iq∗ for reactive power control.

The relationship between natural (abc) and rotating (dq0) frames of
injected grid current (iabc ) formulated based park transformation is in
(1). The three-phase voltage (vabc ) is in (2). θ is the grid voltage phase
angle; vm the phase maximum voltage, whereas id , iq , and i 0 are direct,
quadrature, and zero components currents, respectively. Usually, i 0 is
neglected.
2.3. Boost converter and VSI control
Fig. 3 shows boost converter PWM control. The state of difference
between PV farm voltage (vpv ) and IC MPPT output voltage (vMPPT ) in (6)
decides control signal of switch T1 through PI 3 regulation based signals
comparator. The signal is 0 when (vMPPT ) is greater than (vpv ) and 1 else.

Point of Common
Coupling

+ DC-DC + 3-phase iabc 0.27/30 kV 30/110 kV Grid

BESS vpv buck-boost Cdc vdc 2-level LCL
converter VSI filter
vabc 110kV
S1~S6 100MVA
50 Hz
10 kHz SPWM
PV farm DC-DC signal PLL vabc
6.5 kV boost *
vabc
converter
abc dq
dq
*
abc
vdq
vpv Incremental vdc Voltage mode idq* vdq
PWM dq current dq
ipv conductance PQ control control iabc
Q (PI 1 and PI 2) idq abc
MPPT
Fig. 1. Modeled PV system with BESS.

518
B.S. Tekpeti et al. Electrical Power and Energy Systems 100 (2018) 517–530

Point of Common
Coupling

+ DC-DC + 3-phase iabc 0.27/30 kV 30/110 kV Grid

PV farm LCL
vpv boost Cdc vdc 2-level
6.5 kV filter
converter VSI vabc 110kV
100MVA
S1~S6 50 Hz
ipv Incremental 10 kHz
conductance PWM signal SPWM PLL vabc
vpv MPPT *
vabc
abc dq
dq
*
abc
vdq
vdc Voltage mode idq* v
dq current dq dq
PQ control control iabc
Q (PI 1 and PI 2) idq abc
Fig. 2. Modeled PV system without BESS.

ipv L1 D1 ipv
S22
+ + Reset
Rint ibat L2
Limiter vdc
PV vpv T1 ibat ibatlim 980V Ebat
Cpv Cdc vdc S11
Farm vbat
940Ah

SOC
Buck boost switching
vpv Filter IC MPPT vMP PT control logic
PI 3 Trigger vdc ibat
ipv Filter algorithm + COM pulse
*
vdc
*
ibat COM S22
vpv PI 4 PI 5
Fig. 3. Boost converter control with MPPT.
SOC SOC & vdc S11
vdc control
The DC bus capacitor (Cdc ) and PV farm output capacitor (Cpv ) are
modeled 10 4uF . Fig. 4. BESS buck-boost converter control.

1,vpv−vMPPT > 0
T1 = ⎧
part and exponential voltage part, which are 2nd and 3rd terms of (9),
⎩ 0,vpv−vMPPT < 0
⎨ (6)
respectively.
∗
Based produced vdq from current control loops, inverse park transfor- From Fig. 4, vdc is compared to vdc ∗
and controlled through PI4 to
∗
mation is applied to obtain equivalent reference voltages signals (vabc ) in ∗ ∗
generate reference battery current ibat . ibat is further compared to ibat and
abc natural frame. Lastly, SPWM for inverter control is realized. Indeed, ∗
controlled through PI5. ibat is defined in (10), where Kpb4 and Kib4 are
∗
signals (vabc ) are compared to 10kHZ high frequency signal to avoid PI4 proportional and integral gains, respectively. Combined SOC and vdc
switching loss and give the inverter 6 IGBT control signals S1−S6. The control is inspired from [12,35]. Both PI4 and PI5 controllers generated
LCL filter at VSI output has L = 2mH , C = 30uF and L = 1mH . signal is passed to the PWM circuit and later the logic circuit is used for
decision upon charge, discharge or halt modes of operation. The switch
2.4. BESS based buck-boost converter control S11 is triggered and S22 is zero during boost mode (discharge), whereas
S22 is triggered and S11 is zero during buck mode (charge) to absorb
The BESS buck-boost converter control is in Fig. 4. Lead-acid battery power from the DC bus. However, S11 and S22 become zero when no
is applied in this paper based [12,34,35]. The battery has terminal regulation signal is needed (halt mode).
voltage (vbat ) and state of charge (SOC ) define in (7) and (8), respec-
tively, as function of battery current (ibat ). ∗ 1 ∗
ibat = ⎛Kpb4 + Kib4 × ⎞ (vdc −vdc )
⎝ s⎠ (10)
vbat = Ebat −Rint ibat (7)
As defined in (11), fluctuations originated from PV (ΔP ),load or grid can
SOC = 100 1− ( (∫ i bat dt / Qbat )) (8)
cause extra energy (ΔE ) largely influenced by Cdc.. That is, vdc regulation
should be fast enough to maintain constant voltage and minimize power
where Qbat is battery capacity and Rint battery internal resistance. The loss.
battery open circuit voltage (Ebat ) in (9) is modeled using controlled
1
voltage source as shown in Fig. 4. ΔE = ∫T s
ΔP. dt =
2
2
[Cdc (vdc 2
1−vdc 0 )]
(11)
Ebat = E0−K × [(1−SOC )/ SOC ] × Qbat + Aexp[−B (1−SOC ) Qbat ] (9) 2 2
where and
vdc1 are DC bus voltages at the end and start of the period
vdc0
where E0 is the battery constant, which value lies between the full Ts , respectively.
charged voltage and exponential voltage of battery discharge curve. K The PV farm has 9.75 MW at standard test condition, which is 1000
is the polarization voltage; A and B are exponential voltage and ex- W/m2 irradiation and 25∘C temperature. The series and parallel com-
ponential capacity, respectively. The battery model has normal voltage binations of batteries produces total BESS power of approximately 920

519
B.S. Tekpeti et al.
kWh suitable to compensate for short term deviation in DC bus voltage.
As PV systems of Figs. 1 and 2 are determinant parts of focused DSs
in this paper, we preliminary conducted simulations on them to assure
each model works properly. However, the results are not mentioned
here to rather focus on this paper objective fulfillment. Unlike PV
models including BESS connection and disconnection cases in same
model, this paper applies two separate similar models with and without
BESS. This seems more accurate situation as no need to connect and
disconnect BESS to evaluate its performance. This paper analysis based
PV models of Figs. 1 and 2 could be profitable for protection engineers
to investigate overcurrent relays protection challenges against PV
fluctuations, such as unpredictable irradiation fluctuation and PVs
connection and disconnection in time, as discussed further. Thus, the
analysis could be useful to address new methods to enhance relay co-
ordination in DS [36–39]. The BESS performance is revealed based
pertinent analyses.
3. PV connected DSs models
Two DSs are considered for fault analysis in this paper, where the
difference between them is in the PV systems models used. The two DSs
have same scheme profile in Fig. 5 modeled in PSCAD. Precisely, for the
DS including BESS, PV1 and other PVs to be connected later at nodes
2–4, have separate identical PV system model of Fig. 1 with entire
control strategies. On the other hand, PVs of DS without BESS have
separate identical PV system of Fig. 2 with entire control strategies.
This paper is using accurate PV systems models with satisfied high
degree of complexity, congestion, and accuracy. In fact, quality models
are usually avoided over inaccurate models, such as constant voltage
and current controlled sources full of assumptions and uncertainty.
Though connected PV systems are identical, depending on type of DS
model, no common variable existed between them in implementation,
as each of them has its own variables, with subscripts 1–4, for PV1-PV4,
respectively, to comply with software requirement. The radial DS in
Fig. 5 includes eight OCRs named, R1-R8, and the lines are sketched
5 km and have two identical relays each. In addition, the grid is 110 kV
with 100 MVA power rating.
4. Solar radiation fluctuation impact correction
Considering Fig. 5 model, with only PV1 connected based Fig. 2
model, low irradiation fluctuation in 200–600 W/m2 induced 1 kA low
PV1 inverter current in Fig. 6(a). However, the use of BESS has cor-
rected that current up to 2.2 kA normal amplitude in Fig. 6(b).
Insignificant and high build up currents due to simultaneous 3L-G
faults in middle of lines L1-L4 are observed in Fig. 6(a) and (b), re-
spectively. The maximum fault currents are 1.2 kA and 4.8 kA in
Fig. 6(a) and (b), respectively. It is important to emphasize that irra-
diation fluctuations only have impact on inverter current and not on its
voltage, as high irradiation provides normal current amplitude,
PV 1
0.27 kV/
30 kV
R1 R2 R3 R4
30 kV/
Ld1 Ld2
110 kV Fault Fault
Ld3 Ld4
110 kV
100 MVA R5 R6 R7 R8
50 Hz
Grid
Fault Fault
Fig. 5. PV connected DS.
520
Electrical Power and Energy Systems 100 (2018) 517–530
whereas low irradiation results in low current.
Due to PV2 connection, apart from normal voltage drop during fault
from 0.3 s to 0.4 s, a slight but non-negligible voltage drop occurs at
PV1 inverter side, before and after fault, as shown in Fig. 6(d), com-
pared to normal voltage in Fig. 6(c). It could be caused by applied same
size PVs, as an extra case study justified that, PV2 with size higher than
PV1 rather induces over voltage, before and after fault, whereas a size
lower than PV1 induces more voltage drop, before and after fault. Same
observations apply to PV3-PV4 separate connections at nodes 3 and 4,
respectively, added to only PV1 initially connected in Fig. 5, with and
without BESS.In addition, BESS has regulated DC bus voltage and PV1
output during low irradiation profile in a robust and efficient way as
depicted in Fig. 7. Indeed, irradiation started maintained at 1000 W/m2
and decreased to 600 W/m2 at 0.5 s and, fluctuated randomly in
200–600 W/m2 with and without BESS. PV1 6.5 kV voltage is smoothed
in Fig. 7(b) compared to Fig. 7(a) voltage under minimum acceptable
value from 0.5 s, with and without BESS, respectively. DC bus voltage is
regulated in Fig. 7(d) compared to Fig. 7(c) voltage under minimum
acceptable value from 0.5 s, with and without BESS, respectively. The
batteries previously charged during PV working at high irradiation
around 1000 W/m2 are discharged against low irradiation to smooth
and regulate PV1 voltage and DC bus voltage, respectively. However,
PV1 current which is 1.5 kA at 1000 W/m2,varies almost same in BESS
absence and presence in Fig. 7(a) and (b), respectively, but fluctuations
instability is increased without BESS. Moreover, fault occurrence has
insignificant influence on PV current and PV and DC bus voltages as
clearly noticeable in fault duration from 0.3 s to 0.4 s in Fig. 7(a)–(d).
5. Proposed simulation method details
3L-G faults of 0.3 ohm severity are applied in the middle of lines L1-
L4, simultaneously, during all fluctuation cases. PV1 irradiation is
varied with and without BESS, in both high and low irradiation profiles,
separately, at all simulation time, for all studied cases.For both DSs
models, first, PV1 impact is investigated to collect relays maximum
fault currents in time. Later, PV2-PV4 are connected to nodes 2–4, re-
spectively, at same time, which is beginning of simulation, to collect
relays fault currents. PV2-PV4 irradiation is kept constant at 1000
W/m2 in this paper. PV2-PV4 connection is in Fig. 8(a). The case study
systems, PV2-PV3 disconnected whereas PV1 and PV4 connected; PV2
and PV4 disconnected whereas PV1 and PV3 connected; and PV3-PV4
disconnected whereas PV1-PV2 in service, as shown in Fig. 8(b)–(d),
respectively, are further carried out to do same. LVRT capability, which
is hot topic for PV is not applied in this paper as PVs connection and
disconnection scenarios are methodically proceeded manually. Except
PV1 which is always connected during all scenarios to dynamically
fluctuate its irradiation it time, the other PVs are either connected or
disconnected at time t = 0 s to simulate the expected scenario in Fig. 8.
Software limitations make long time irradiation fluctuations simu-
lation tedious. Besides, fast irradiation variations can be assessed
around an arbitrary instant in time. The faults started in 0.3 s and they
take 0.1 s. The maximum fault currents seen by each relay is collected
following fault current evolution in time. Precisely, every 0.02 s, the
maximum fault current of the relays are plotted, representatively, in
form of charts, from 0.3 s to 0.4 s. That is, from the collected dynamic
maximum faults currents in time, each relay will have highest and
minimum faults currents values in time. The highest fault current is the
high amplitude fault current in time whereas the minimum fault current
is the low amplitude fault current in time. This method ease fault
current comparison in time. Each relay fault current is represented with
a unique color to ease differentiation. During the fault period, the
plotted maximum fault currents correspond to maximum phase currents
at specified time. Based on this simple and effective method applied in
PSCAD, OCRs maximum faults currents are collected and plotted
faithfully.
B.S. Tekpeti et al. Electrical Power and Energy Systems 100 (2018) 517–530

PV1 Inverter side current PV1 Inverter side current

Iabc1 Iabc1
4.0 6.0

3.0
4.0

PV1 Inverter side current (kA)

PV1 Inverter side current (kA)

2.0

2.0
1.0

0.0 0.0

-1.0
-2.0
-2.0
-4.0
-3.0

-4.0 -6.0

Time(s) 0.300 Time(s) 0.300 0.310 0.320 0.330 0.340 0.350 0.360 0.370 0.380 0.390 0.400
0.310 0.320 0.330 0.340 0.350 0.360 0.370 0.380 0.390 0.400

(a) (b)
PV1 Inverter side voltage PV1 Inverter side voltage
Vabc1 Vabc1
0.250

0.200
0.20
0.150
PV1 Inverter side voltage(kV)
PV1 Inverter side voltage(kV)

0.10 0.100
0.050
0.00 0.000

-0.050
-0.10
-0.100

-0.20
-0.150
-0.200

-0.250
Time(s) 0.250 0.300 0.350 0.400 0.450 Time(s) 0.200 0.250 0.300 0.350 0.400 0.450 0.500

(c) (d)
Fig. 6. PV1 performance under random low irradiation variation in 200–600 W/m2
against 0.1 s simultaneous 0.3 ohm 3L-G faults in middle of L1-L4: (a) PV1 inverter current without
BESS (b) PV1 inverter current with BESS (c) PV1 inverter normal voltage (d) PV1 inverter voltage with PV2 added at node 2.

6. Results and relays faults currents fluctuations analysis 6.74 kA, 6.66 kA, 3.12 kA, 15.65 kA, 6.73 kA, 6.6 kA and 2.91 kA, re-
spectively, at 0.3 s for R4 and R8, whereas at 0.32 s for other relays in
Similar relays fault current profiles are observed against high and Fig. 11. However, with BESS included, R1-R8 have highest faults cur-
low irradiations fluctuations, where fluctuations in time seem to have rents of, 14.74 kA, 6.05 kA, 6.02 kA, 2.85 kA, 14.88 kA, 6.08 kA,
insignificant impact on faults currents seen by relays. However, PVs 6.01 kA, and 3.1 kA, respectively in Fig. 12, at same instants as in
connection and disconnection have very remarkable influence on faults Fig. 11.
currents seen by relays. Based on these facts, it is unnecessary to repeat In both cases, in terms of achieved highest fault currents, R4 and R8
approximately same values of fault currents for high and low irradia- have low faults currents probably due to far distance from PV1. As it is
tions fluctuations. That is, it is very useful for the purpose of this paper easily comparable, R1 and R5, R2 and R6, and R3 and R7, have very
to plot relays maximum faults currents variation in time when con- close highest faults currents, maybe due to the symmetry among relays
necting and disconnecting PVs at specific nodes, considering high and positions. Except R8, in majority, R1-R7 highest faults currents without
low irradiation fluctuations in time as a unique case. Indeed, as relays BESS are slightly greater than those with BESS. On the other hand,
faults currents profiles were almost identical against high and low ir- without BESS, R1, R3, R5 and R7 have minimum fault currents of,
radiation profiles, we choose to plot results with low irradiation profile 9.32 kA, 4.34 kA, 9.4 kA and 4.32 kA, respectively, achieved at 0.3 s for
as that profile usually gives PV systems weak performance. As a result, both R1 and R5 and at 0.4 s for both R3 and R7 in Fig. 11. However,
relays fault currents under low irradiation apply to high irradiation. with BESS, the corresponding relays mainly have limited minimum
Against high and low irradiations fluctuations in Figs. 9 and 10, fault currents of, 8.42 kA, 3.83 kA, 10.35 kA and 3.76 kA, respectively,
respectively, with separate PVs connection and disconnection cases, the at 0.3 s for R1 and at 0.4 s for the remaining relays in Fig. 12.
coming figures show relays maximum faults currents fluctuations dy-
namism in time in absence and presence of BESS. Considering the high
6.2. PV1-PV4 supplying 6.5 kV power each
6.5 kV power PVs penetration in this paper, quite high faults currents
could be observed, depending on, fault type, location and severity.
Here, PV2-PV4 are connected at time t = 0 s . Without BESS, R1-R8
have highest faults currents of, 15 kA, 12.12 kA, 12.07 kA, 15.57 kA,
6.1. PV1 supplying 6.5 kV power 12.94 kA, 15.92 kA, 5.12 kA and 15.00 kA, respectively, at 0.36 s for R6,
whereas at 0.32 s for other relays in Fig. 13. PV2-PV4 connection in-
Without BESS, relays R1-R8 have highest faults currents of, 15.6 kA, duced random faults currents and, as we can see, there are no particular

521
B.S. Tekpeti et al. Electrical Power and Energy Systems 100 (2018) 517–530

PV1 Volta
ge and Curren
t PV1 Volta
ge and Current
Vpv1 Ipv1 Vpv1 Ipv1
8.0 8.0

7.0 7.0
PV Voltage (kV) / Current (kA)

PV1 Voltage(kV)/Current (kA)

6.0 6.0

5.0 5.0

4.0 4.0

3.0 3.0

2.0 2.0

1.0 1.0

0.0 0.0

Time(s) Time(s) 0.00 0.10 0.20 0.30 0.40 0.50 0.60 0.70 0.80 0.90 1.00
0.00 0.10 0.20 0.30 0.40 0.50 0.60 0.70 0.80 0.90 1.00

(a) (b)
PV1 DC bus voltage PV1 DC bus voltage
Vdc1 Vdc1
8.0 8.0

7.0 7.0
PV1 DC bus voltage (kV)

6.0 6.0

PV1 DC bus voltage (kV)

5.0 5.0

4.0 4.0

3.0 3.0

2.0 2.0

1.0 1.0

0.0 0.0
Time(s) 0.00 0.10 0.20 0.30 0.40 0.50 0.60 0.70 0.80 0.90 1.00 Time(s) 0.00 0.10 0.20 0.30 0.40 0.50 0.60 0.70 0.80 0.90 1.00

(c) (d)
Fig. 7. PV1 performance under random low irradiation fluctuation in 200–600 W/m2
against 0.1 s simultaneous 0.3 ohm 3L-G faults in middle of L1-L4: (a) PV1 current and voltage
without BESS (b) PV1 current and voltage with BESS (c) PV1 DC bus voltage without BESS (d) PV1 DC bus voltage with BESS.

currents similarities among the relays. In majority, adding PV2-PV4 10.53 kA, 16.02 kA and 10.53 kA, respectively, at 0.3 s for R4, whereas
provoked relays build up currents and, for instance, R4, R6 and R8, at 0.32 s for other relays in Fig. 16. Without BESS, R2, R4 and R6 have
with highest currents of 3.12 kA, 6.73 kA and 2.91 kA, respectively, minimum faults currents of, 5.06 kA, 0.82 kA and 9.37 kA, respectively,
when PV1 connected, suddenly present build up currents of 15.57 kA, at 0.4 s for R2, 0.32 s for R4 and 0.3 s for R6 in Fig. 15. However, using
15.92 kA and 15.00 kA, respectively. It seems that PVs connections BESS, the corresponding relays have limited minimum faults currents
induce more faults currents seen by the closest relays, which are R4, R6 of, 4.31 kA, 0.69 kA and 8.58 kA, respectively, at cited same instants in
and R8, for PV2-PV4, respectively. Fig. 16.
Moreover, with BESS, R1-R8 have slightly smoothed highest faults
currents of, 14.56 kA, 10.97 kA, 10.94 kA, 16.13 kA, 12.19 kA, 16 kA, 6.4. PV1 and PV3 supplying 6.5 kV power each
4.46 kA and 14.74 kA, respectively, at 0.34 s for R6, whereas at 0.32 s
for other relays in Fig. 14, but build up currents still present. Without Here, PV3 and PV4 are connected and disconnected, respectively, at
BESS, R5 and R7 have minimum faults currents of, 6.22 kA and 2.45 kA, time t = 0 s.. Without BESS, R1-R8 have highest fault currents of,
respectively, at 0.3 s for both relays in Fig. 13. However, with BESS, 16.78 kA, 7.23 kA, 7.15 kA, 2.79 kA, 12.66 kA, 15.77 kA, 7.54 kA, and
both relays have 5.58 kA and 2.18 kA limited currents, respectively, at 2.57 kA, respectively, at 0.3 s for both R4 and R8, whereas at 0.32 s for
same instant in Fig. 14. other relays, as shown in Fig. 17.
Relay R6, which has 6.73 kA with PV1, now sees 15.77 kA build up
6.3. PV1 and PV4 supplying 6.5 kV power each current, whereas other relays have random decrease or increase in fault
currents. The use of BESS gives R1-R8 slightly smoothed highest fault
In this case, PV2-PV3 are disconnected at time t = 0 s . Without currents of, 16.05 kA, 6.51 kA, 6.46 kA, 2.86 kA, 11.83 kA, 16.04 kA,
BESS, R1-R8 have highest faults currents of, 17.65 kA, 7.61 kA, 7.53 kA, 7.13 kA and 2.67 kA, respectively, at 0.3 s for R4, R6 and R8, whereas at
2.96 kA, 10.80 kA, 11.8 kA, 11.66 kA and 16.25 kA, respectively, at 0.32 s for other relays in Fig. 18. Without BESS, R2 and R4 have
0.3 s and 0.34 s for R4 and R8, respectively, and at 0.32 s for other minimum faults currents of, 4.89 kA and 0.89 kA, respectively, at 0.3 s
relays in Fig. 15. for R2 and 0.32 s for R4 in Fig. 17. With BESS utilization, same relays
Here also the relays faults currents seem to have random variation have limited minimum faults currents of, 2.54 kA and 0.74 kA, re-
with noticeable more build up currents compared to the case with PV1. spectively, at cited same instants as shown in Fig. 18.
Relay R8 which has 2.91 kA with PV1, now sees 16.25 kA build up
current, whereas other relays have random decrease or increase in 6.5. PV1 and PV2 supplying 6.5 kV power each
faults currents. The model with BESS offers R1-R8 slightly smoothed
highest fault currents of, 16.59 kA, 6.75 kA, 6.70 kA, 3.05 kA, 10.9 kA, Lastly, PV2 and PV3 are connected and disconnected, respectively,

522
B.S. Tekpeti et al. Electrical Power and Energy Systems 100 (2018) 517–530

PV 1 PV 2 PV 1
0.27 kV/ 0.27 kV/ 0.27 kV/
30 kV 30 kV 30 kV
R1 R2 R3 R4 R1 R2 R3 R4

30 kV/ Ld1 Ld2 30 kV/ Ld1 Ld2

110 kV Fault Fault 110 kV Fault Fault
Ld3 Ld4 Ld3 Ld4
110 kV 110 kV
100 MVA R5 R6 R7 R8 100 MVA R5 R6 R7 R8
50 Hz 50 Hz
Grid Grid

Fault 0.27 kV/ Fault 0.27 kV/ Fault Fault 0.27 kV/
30 kV 30 kV 30 kV
PV 3 PV 4 PV 4

(a) (b)

PV 1
PV 1 PV 2
0.27 kV/
30 kV 0.27 kV/ 0.27 kV/
30 kV 30 kV
R1 R2 R3 R4
R1 R2 R3 R4

30 kV/ Ld1 Ld2

110 kV Fault Fault 30 kV/ Ld1 Ld2
Ld3 Ld4 110 kV Fault Fault
110 kV
100 MVA R5 R6 R7 R8 Ld3 Ld4
110 kV
50 Hz 100 MVA R5 R6 R7 R8
Grid
50 Hz
Fault Fault Grid
0.27 kV/
30 kV
Fault Fault
PV 3

(c) (d)
Fig. 8. PV dominated DSs case study systems: (a) PV1-PV4 connected (b) PV1 and PV4 connected (c) PV1 and PV3 connected (d) PV1 and PV2 connected.

at time t = 0 s . With no BESS, R1-R8 have highest fault currents of,
10.81 kA, 11.79 kA, 11.76 kA, 16.37 kA, 17.66 kA, 7.6 kA, 7.49 kA and
2.94 kA, respectively, at 0.3 s for R8, whereas at 0.32 s for other relays
in Fig. 19.
Relay R4, which has 3.12 kA with PV1, now sees 16.37 kA build up
current, whereas other relays have random decrease or increase in fault
currents. With PVs having BESS, R1-R8 have slightly smoothed highest
fault currents of, 10.9 kA, 10.57 kA, 10.54 kA, 15.96 kA, 16.58 kA,
6.76 kA, 6.68 kA and 2.83 kA, respectively, at 0.3 s and 0.34 s for R8
and R4, respectively, whereas at 0.32 s for other relays, as shown in
Fig. 20. Without BESS, R2, R4, R6 and R8 have minimum faults currents
of, 9.38 kA, 15.71 kA, 5.04 kA and 0.78 kA, respectively, at 0.3 s for R2,
0.4 s for both R4 and R6, and at 0.32 s for R8 as depicted in Fig. 19.
However, based BESS application, the corresponding relays have lim-
ited minimum faults currents of, 8.53 kA, 15.5 kA, 4.33 kA and 0.78 kA
respectively, at 0.3 s for R2 and R4, 0.4 s for R6 and 0.32 s for R8 in
Fig. 20.
Investigations revealed that relays mainly have highest faults cur-
rents achieved at 0.32 s. That is, against high and low irradiations
fluctuations, relays fault currents are in majority highest within 0.02 s

PV1 Solar radiation fluctuation

G1
1.00k
0.97k
Solar radiation (W/m^2)

0.95k
0.93k
0.90k
0.88k
0.85k
0.82k
0.80k
Time(s) 0.300 0.320 0.340 0.360 0.380 0.400

Fig. 9. PV1 high irradiation fluctuation in 800–1000 W/m2 .

523
B.S. Tekpeti et al. Electrical Power and Energy Systems 100 (2018) 517–530

PV1 Solar radiation fluctuation

G1
700

600
Solar radiation (W/m^2)
500

400

300

200
Time(s) 0.300 0.320 0.340 0.360 0.380 0.400

Fig. 10. PV1 low irradiation fluctuation in 200–600 W/m2 .

20
R1
R2
Currents (kA)

15
R3
10 R4
5 R5
R6
0 R7
0.3 0.32 0.34 0.36 0.38 0.4
R8
Time (s)
Fig. 11. Relays maximum fault currents fluctuations with PV1 without BESS.

of fault duration. In addition, though faults currents variation is
random, in majority we can observe a slight decrease in faults currents
profiles from 0.34 s to 0.4 s. Globally, the case with PV2-PV4 added to
existing PV1 induced high fault currents compared to other cases
showing similar performance at some instants. Importantly, the far the
relay the less the impact on its fault current with PV2-PV4 either con-
nected or disconnected at nodes 2–4, respectively.
Moreover, more in-depth results analyses, with high and low irra-
diations profiles, have shown that, although the impact of irradiation
fluctuation in time is slight, there seems to be slight build up currents at
some instants. However, PVs connection and disconnection scenarios
induce more build up currents in presence of fault and non-uniform
irradiation fluctuation.
Another separate case study, on both DSs models, with results not
plotted here, revealed that, fault resistance higher than applied 0.3 ohm
induces low relays fault currents, however, a resistance below 0.3 ohm
induces high fault currents. Moreover, against PVs connection and
disconnection, with impact depending on PVs locations and sizes, relays
and grid faults currents fluctuations are proportional to fault resistance
and so is the grid and transmission lines voltages drop.
It may seem that BESS combined with employed control strategies
has positive effect on system dynamism to slightly smooth the time
varying fault currents to minimum. These results could be improved
with a more robust BESS. However, this paper BESS which has com-
prehensive limited capability with few details provided, seems suffi-
cient to improve DC bus voltage and PV farm voltage quality, through
adjustment of voltage fluctuations, such as overshoot or undershoot,
and voltage sags and dips. Therefore, the use of batteries storage is
capable to enhance high PV connected DS performance.
Results similarities are observed when fluctuating irradiation of
each of PV2-PV4, separately, by adopting same procedure clearly de-
monstrated using PV1 as an example. That is, when varying respective
irradiation, each of PV2-PV4 impact is carried out firstly; then other
PVs are connected to respective nodes secondly; followed with other
cases, proceeded for PV2 as, PV1 and PV4 disconnected whereas PV2
and PV3 connected; PV1 and PV3 disconnected whereas PV2 and PV4
connected; and PV3-PV4 disconnected whereas PV1-PV2 in service etc.
We conducted all those case studies but due to space limitation and
resemblance of results the details are not provided.
Importantly, considering both DS models and both irradiation

20
R1
R2
Currents (kA)

15
R3
10 R4
5 R5
R6
0 R7
0.3 0.32 0.34 0.36 0.38 0.4 R8
Time (s)
Fig. 12. Relays maximum fault currents fluctuations with PV1 using BESS.

524
B.S. Tekpeti et al. Electrical Power and Energy Systems 100 (2018) 517–530

20
R1

Currents (kA)
15 R2
10
R3
R4
5 R5
R6
0
0.3 0.32 0.34 0.36 0.38 0.4 R7
R8
Time (s)
Fig. 13. Relays maximum fault currents fluctuations with PV1-PV4 without BESS.

20 R1
R2
Currents (kA)

15
R3
10 R4
5 R5
R6
0 R7
0.3 0.32 0.34 0.36 0.38 0.4
R8
Time(s)
Fig. 14. Relays maximum fault currents fluctuations with PV1-PV4 using BESS.

20
R1
R2
Currents (kA)

15
R3
10 R4
5 R5
R6
0 R7
0.3 0.32 0.34 0.36 0.38 0.4
R8
Time (s)
Fig. 15. Relays maximum fault currents fluctuations with PV1 and PV4 without BESS.

20
R1
15 R2
Currents (kA)

R3
10 R4
5 R5
R6
0 R7
0.3 0.32 0.34 0.36 0.38 0.4
R8
Time(s)
Fig. 16. Relays maximum fault currents fluctuations with PV1 and PV4 using BESS.

20 R1
R2
Currents (kA)

15
R3
10 R4
R5
5
R6
0 R7
0.3 0.32 0.34 0.36 0.38 0.4 R8
Time (s)
Fig. 17. Relays maximum fault currents fluctuations with PV1 and PV3 without BESS.

525
B.S. Tekpeti et al. Electrical Power and Energy Systems 100 (2018) 517–530

20
R1

Currents (kA)
15 R2
10
R3
R4
5 R5
0 R6
0.3 0.32 0.34 0.36 0.38 0.4 R7
Time (s) R8
Fig. 18. Relays maximum fault currents fluctuations with PV1 and PV3 using BESS.

20
R1
R2
Currents (kA)

15
R3
10
R4
5 R5
R6
0
0.3 0.32 0.34 0.36 0.38 0.4 R7
R8
Time (s)
Fig. 19. Relays maximum fault currents fluctuations with PV1 and PV2 without BESS.

20
R1
Currents (kA)

15 R2
R3
10
R4
5 R5
0 R6
0.3 0.32 0.34 0.36 0.38 0.4 R7
Time (s) R8
Fig. 20. Relays maximum fault currents fluctuations with PV1 and PV2 using BESS.

profiles, in the case PV1-PV4 connected, frequency unbalance profiles verified.

of PV1 and PV3 systems due to fault are depicted in Fig. 21(a) and (b),
respectively. Similar frequency unbalance profiles are noticed for PV2
and PV4 systems. Thus, frequency unbalance with PV needs serious
attention. The normal frequency is 50 Hz in this paper. From Fig. 21,
PV1 and PV3 systems frequency unbalance profiles look identical at
some instants.
7. OCRs performance analysis
The relay retained maximum fault current to evaluate its perfor-
mance with PSO is maximum of maximum faults currents of instants
0.3 s, 0.32 s, 0.34 s, 0.36 s, 0.38 s, and 0.4 s in each case. Precisely, in
each fluctuation case, the selected relay maximum fault current corre-
sponds to the highest fault current among the plotted relay maximum
fault currents in time. Those highest relays faults currents are the ones
mentioned and analyzed above while precising the times at which they
occurred, so no need to repeat them. This method seems reasonable and
reliable as irradiation is fluctuated in short time. The relationship main-
backup relays displayed in Table 1 is set fixed for all fluctuations cases.
PSO is a well-known optimization algorithm with details available in
[15]. PSO algorithm is suitably coded in Matlab based relevant details
given further. The formulation of relay coordination problem is also
available in [15] with clear and convincing explanations. Some im-
portant relays parameters are specified here. OCR protection with re-
quired equations is widely discussed in [3].
Essentially, PSO updates position and velocity of relays so-called
particles or agents through following (13), (14) and (15) until (12) is
t jback−tymain ⩾ CTI ;j,y = 1,…,n;j ≠ y (12)
Pym + 1 = Pym + vym + 1 (13)
where t jback and tymain are back-up and main relays operating times,
respectively. CTI is the coordination time interval which depending
upon relay type, circuit breaker speed and other parameters is chosen
equal 0.2 s in this paper. It represents an intentional time to assure
proper operation of main relay before back-up relay. n is the number of
relays involve in the network. Pym + 1 and Pym are the position of the agent
y in the iteration m + 1 and m , respectively, whereas vym + 1 and vym are
the velocity of the agent y in the iteration m + 1 and m , respectively. A
single agent velocity is defined in (14).
vym + 1 = wvym + c1 rand1 × (pbest y −Pym) + c2 rand2 × (gbest −Pym) (14)
where pbest y is the best solution of agent y and gbest that of the whole
swarm with all the particles in movement. Generally, c1 and c2 which
are the weighting factors, are equal whereas rand1 and rand2 are two
random numbers between zero and one. The defined parameter w in
(15) is the weighting function, with its maximum (wmax ) and minimum
(wmin ) values.
wmax−wmin
w = wmax− × iter
iter max (15)
where iter and iter max are respectively, current iteration and maximum
number of iterations during simulations.
PSO parameters are in Table 2. In addition, OCR time multiplier

526
B.S. Tekpeti et al. Electrical Power and Energy Systems 100 (2018) 517–530

PV1 frequency
freq1
62.5

60.0

PV1 frequency (Hz) 57.5

55.0
52.5

50.0
47.5
45.0

42.5
40.0
Time(s) 0.200 0.250 0.300 0.350 0.400 0.450 0.500 0.550 0.600

(a)

PV3 frequency
freq3
62.5
60.0
57.5
PV3 frequency (Hz)

55.0
52.5
50.0
47.5
45.0
42.5
40.0
37.5
Time(s) 0.200 0.250 0.300 0.350 0.400 0.450 0.500 0.550 0.600

(b)
Fig. 21. PV systems frequency unbalance due to 0.3 ohm 3L-G faults: (a) PV1 frequency unbalance (b) PV3 frequency unbalance.

Table 1 Table 3
Main relays and Backup relays. TMS, PS and OT limits.

Main relay Backup relay Main relay Backup relay Parameters Minimum Maximum

1 – 5 1 TMS 0.1 1.1

2 – 6 – PS 0.5 2.5
3 1 7 5 OT 0.1 s 4s
4 2 8 6

setting (TMS ) , plug setting (PS ) and operating time (OT ) limits are
Table 2 defined in Table 3 where the number of particles population is fixed 30
PSO parameters.
as only 8 OCRs are involved in the proposed method. The objective is to
Number of Number of Weighting Minimum Maximum
find optimized relays settings and objective function (OF ) . OF re-
population iterations factors (same weighting weighting presents minimum value of total OT of relays. In fact, OF is the fitness
(maximum) value) function function value which should be optimized based constraints [15].
Optimized OCRs TMS and PS values are in Table 4 and Table 5, for
30 100 2 0.4 0.9
DSs with and without BESS, respectively. Although DS with BESS pre-
sents faults currents limited to minimum, the relays settings values are
variably random in presence of PV fluctuations in both tables.

527
B.S. Tekpeti et al. Electrical Power and Energy Systems 100 (2018) 517–530

Table 4
Optimized relays settings with BESS.

Relays PSO PV1 PSO PV1-PV4 PSO PV1 and PV4 PSO PV1 and PV3 PSO PV1 and PV2

TMS PS TMS PS TMS PS TMS PS TMS PS

1 0.71737 0.83002 0.48824 2.2807 0.8881 2.4692 1.0105 1.8983 0.47461 1.2434
2 0.21211 1.8077 0.66029 2.1382 0.80121 1.8765 0.73283 1.4543 0.12978 0.62842
3 0.53044 0.59691 0.52049 2.1519 0.54225 1.8801 0.6891 0.63727 0.35058 2.0409
4 0.32259 0.51946 0.20091 0.60024 0.14913 0.80138 0.407 0.57853 0.83156 0.82998
5 0.93576 0.74242 0.48412 0.60876 0.44332 0.83852 0.17728 0.72505 0.78627 1.1835
6 0.852 0.66236 0.46556 2.3105 0.70243 1.8081 0.32392 2.1898 0.24063 0.92035
7 0.67325 1.2858 0.26886 0.84709 0.13598 1.0033 0.49894 1.7352 0.22652 1.9304
8 0.9695 2.0814 0.36072 1.5688 1.0369 0.9147 0.46516 2.3373 0.68294 1.8345
OF (s) 5.6866 2.7589 4.0619 3.8603 3.9109
Time (s) 61.492 61.804 54.704 55.935 59.154

Table 5
Optimized relays settings without BESS.

Relays PSO PV1 PSO PV1-PV4 PSO PV1 and PV4 PSO PV1 and PV3 PSO PV1 and PV2

TMS PS TMS PS TMS PS TMS PS TMS PS

1 0.74912 1.8736 0.84452 0.6907 0.2872 2.4397 0.71895 0.75192 1.0436 1.6463
2 0.83172 0.86702 0.29379 0.62815 0.81632 1.9887 0.19211 1.0921 0.83724 2.2673
3 0.74775 1.237 0.55912 0.54391 0.23754 2.4816 0.44842 1.4001 0.44278 1.5148
4 0.55092 1.7512 0.30075 2.2016 0.13144 0.87124 0.5064 1.08 0.41655 0.75244
5 0.64701 2.0605 0.23847 0.62785 0.98149 0.6102 0.67948 1.4634 0.49881 1.0298
6 0.39632 0.66225 0.81622 2.169 0.36294 0.6553 0.22647 1.6821 0.51517 1.3078
7 0.84469 2.3588 0.3603 1.4745 0.58228 1.0138 0.32807 1.518 0.17435 0.50445
8 0.28896 2.0514 0.93653 1.7483 0.78373 2.4802 0.96289 0.82271 0.8374 0.54089
OF (s) 5.3774 2.6692 3.465 3.6397 4.2813
Time (s) 73.906 55.749 51.514 55.604 65.307

9
PSO with PV1-PV4 Without BESS
8

7
Fitness value

2
10 20 30 40 50 60 70 80 90 100
Iteration
Fig. 22. Fitness function evolution with PSO in case PV1-PV4 connected without BESS.

Large volatility and strong randomness are characteristics of PV
power. Limiting fault current to minimum as in this paper could be very
profitable for protection systems. With PSO randomization character-
istics, it seems improper to compare relays settings of separate cases
among them. Nonetheless, a brief analysis is relevant. The reported
results are the best solutions achieved over 100 times compilations of
100 iterations each, for each fluctuation case.
From Tables 4 and 5, the optimized TMS and PS remained in spe-
cified limits in each fluctuation case. However, there are no particular
similarities among TMS and PS results. On the other hand, there seems
to be slight similarities among OF values when comparing Tables 5 with
4. For instance, when PV1-PV4 connected with and without BESS, best
OFs are 2.7589 s and 2.6692 s, respectively, whereas PSO algorithm
spent 61.804 s and 55.749 s, respectively, to run the calculations.
Considering all the cases, PSO algorithm spent minimum and maximum
times of 51.514 s and 73.906 s, respectively. The fitness function evo-
lution with PSO in case PV1-PV4 connected without BESS is plotted in
Fig. 22 showing best fitness value of 2.6692 s achieved after 17 itera-
tions. Other cases best fitness values are achieved variably between 10
and 45 iterations. Because few relays and few constraints were in-
volved, PSO algorithm is able to respond in less than 100 s in each case.
However, PSO is usually time consuming [15]. Importantly, in presence
of PV fluctuations, OF should be minimized to decrease the risk and
stress on the protected zones.

528
B.S. Tekpeti et al. Electrical Power and Energy Systems 100 (2018) 517–530

imax
*
vdc + id*
PI 1
-
vdc
imin
imax
Q - PI 2 iq*

+ imin
Q*
Fig. 23. Voltage mode PQ control.

vd
+ + +
PI 6
- -

- +
PI 7 +
+ +v
q
Fig. 24. dq current control strategy.

vd
vabc abc vq wPLL θ
dq0
PI 8
∫
v0 ( neglected )

Fig. 25. Phase locked loop diagram.

8. Conclusion A short relays performance analysis is achieved using PSO algorithm

This paper proposed a method to discuss interesting fault analysis of
PV connected DSs strictly based accurate PV systems. DSs are modeled
with complex inverter based PV systems with and without BESS.
Involved OCRs are intelligently considered to analyze their maximum
faults currents variations against various PV fluctuations scenarios.
Important findings are discussed mainly based impact of irradiation
fluctuation and the DS with BESS has shown great performance.
High and low irradiation fluctuations in time seem not to have
significant impact on relays faults currents but rather seem to induce
increase in current at some instants. However, PVs connection and
disconnection have significant influence on OCRs maximum faults
currents and, for that reason, LVRT with PV must be reinforced. Relays
faults currents depend on PVs locations and the higher the PV pene-
tration the higher the fault currents. The obtained faults currents with
DS having BESS are limited to minimum, which is satisfactory result.
as an example, not ignoring that metaheuristic based algorithms for
setting and coordination of OCRs are under question in practice. Urgent
measures are needed to accommodate OCRs response against PV pe-
netration. New strategies could be proposed making proper use of BESS
to, smooth PV performance, slightly limit OCRs fault currents and, al-
leviate associated issues such as voltage and frequency unbalance
leading to unnecessary islanding of PV systems, provided that the re-
quired entities are well coordinated. Above all, relays coordination
correctness against PV fluctuations is of utmost importance to accom-
modate OCRs with PV.
Acknowledgements
The authors would like to thank the School of Electrical
Engineering, Xi’an Jiaotong University, Xi’an, China, for its financial
support in this research.

Appendix A. Appendix

The modeled voltage mode PQ control based Eq. (5) is in Fig. 23 where the discussed input and output signals are clearly depicted. The generated

529
B.S. Tekpeti et al.
id∗ and iq∗ reference currents in dq frame are further used as inputs of dq current control loop as clearly shown in Fig. 24.
The dq current control is to deliver appropriate dq frame reference voltages vd∗ and vq∗ which expressions are in (16). Referring to Fig. 24, (16) is
easily reachable.
∗
⎧ vd = vd + Δvd−wLiq
∗
⎨ vq = vq + Δvq + wLid
⎩
∗ 1
⎧ Δvd = (id −id )(Kpd6 + Kid6 × s )
⎨ Δvq = (iq∗−iq)(Kpq7 + Kiq7 × 1
) (17)
⎩ s
where Δvd and Δvq in (17) are corresponding voltages of error in id and iq currents, respectively. The error
is an inductance chosen from LCL filter for decoupling actions; w is grid supplied angular frequency; Kpb6 and Kib6 are PI6 proportional and integral
gains, respectively, whereas Kpb7 and Kib7 are PI7 proportional and integral gains, respectively.
PLL is a negative feedback control which role is to output appropriate grid phase angle. To achieve that, PLL transforms inverter ac voltages vabc
into dc dq control voltages (vd,vq) and dynamically proceeds adjustment of the rotational speed wPLL to fulfill vd = 0 and finally output the correct grid
phase angle θ . The modeled PLL technique is shown in Fig. 25.
Appendix B. Supplementary material
Supplementary data associated with this article can be found, in the online version, at http://dx.doi.org/10.1016/j.ijepes.2018.03.003.
References
[1] IEEE Recommended practice for utility interface of photovoltaic (PV) systems, IEEE
Std 929–2000, 30 January 2000.
[2] IEEE guide for determining fault location on AC transmission and distribution lines,
IEEE Std C37. 114™- 2014, IEEE Power and Energy Society.
[3] Paithankar YG, Bhide SR. Fundamentals of power system protection. New Delhi:
Prentice-hall of India Priviate Limited; 2003.
[4] Sumathi S, Kumar LA, Surekha P. Solar PV and wind energy conversion systems.
Green Energy Technol 2015:1–807.
[5] Chen Li-Hsiung. Overcurrent protection for distribution feeders with renewable
generation. Int J Electr Power Energy Syst 2017;84:202–13.
[6] Hernández JC, De La Cruz J, Vidal PG, Ogayar B. Conflicts in the distribution
network in the presence of large photovoltaic plants: the case of ENDESA. Inter
Trans Electr Energy Syst 2013;23(5):669–88.
[7] Elmitwally A, Gouda E, Eladawy S. Optimal allocation of fault current limiters for
sustaining overcurrent relays coordination in a power system with distributed
generation. Alex Eng J 2015;54(4):1077–89.
[8] Liu Haoming, Peng Jianyu, Zang Qiyue, Yang Kelin. Control strategy of energy
storage for smoothing photovoltaic power fluctuations. Int Feder Auto Control
2015;48(28):162–5.
[9] Jayasekara Nadeeshani, Wolfs Peter, Masoum Mohammad AS. An optimal man-
agement strategy for distributed storages in distribution network with high pene-
trations of PV. Electr Power Syst Res 2014;116:147–57.
[10] Wang X, Yue M, Muljadi E. Modeling and control system design for integrated solar
generation and energy storage system with a ride-through capability. In: Energy
conversion conference and exposition on 15–20 September 2012, Raleigh, North
Carolina. p. 3727–34.
[11] Fiodar Kazhamiaka, Catherine Rosenberg, Srinivasan Keshav. Practical strategies
for operation in energy systems: design and evaluation. IEEE Trans Sustain Energy
2016;7(4):1602–10.
[12] Daud MZ, Mohamed A, Wanik Che MZ, Hannan MA. Performance evaluation of
grid-connected photovoltaic system with battery energy storage. In: International
conference on power and energy on 2–5 December 2012, Kota Kinabatu Sabah,
Malaysia. p. 337–42.
[13] Olabi AG. Renewable energy and energy storage systems. Energy 2017;136:1–6.
[14] Yongheng Yang, Frede Blaabjerg, Huai Wang. Low-voltage ride through of single-
phase transformerless photovoltaic inverters. IEEE Trans Ind App
2014;50(3):1942–52.
[15] Kheshti M, Tekpeti BS, Kang X. The optimal coordination of overcurrent relay
protection in radial network based on particle swarm optimization. In: IEEE PES
Asia-Pacific power and energy conference on 25–28 October 2016, Xi’an, China. p.
604–8.
[16] Saad Naggar H, El-Sattar Ahmed A, Mansour Abd El-Aziz M. Improved particle
swarm optimization for photovoltaic system connected to the grid with low voltage
ride through capability. Renewable Energy 2016;85:181–94.
[17] Subhashish Bhattacharya, Tapan Saha, Hossain MJ. Fault contribution from large
photovoltaic systems in building power supply networks. J Build Eng
2016;5:222–30.
[18] Hossain MJ, Saha TK, Mithulananthan N, Pota HR. Robust control strategy for PV
system integration in distribution systems. Appl Energy 2012;99:355–62.
[19] Lal Nandan Vivek, Singh SN. Control and performance analysis of a single-stage
utility-scale grid-connected PV system. IEEE Syst J 2017;11(3):1601–11.
[20] Nejib Hamrouni, Moncef Jraidi, Ahmed Dhouib, Adnen Cherif. Design of a com-
mand scheme for grid connected PV systems using classical controllers. Electr
Power Syst Res 2017;143:503–12.
530
Electrical Power and Energy Systems 100 (2018) 517–530
(16)
in id is id∗−id
whereas the error L in iq is iq∗−iq .
[21] Datta Manoj, Senjyu Tomonobu, Yona Atsushi, Funabashi Toshihisa, Kim Ch-Hwan.
Photovoltaic output power fluctuations smoothing methods for single and multiple
PV generators. Curr Appl Phys 2010;10(2):S265–70.
[22] Shenglong Yu, Lijun Zhang, Iu Herbert HC, Fernando Tyrone, Kit Wong Po. A DSE-
based power system frequency restoration strategy for PV-integrated power systems
considering solar irradiance variations. IEEE Trans Ind Inform 2017;13(5):2511–8.
[23] Ghoddami Hamidreza, Yazdani Amirnaser. A mitigation strategy for temporary
overvoltages caused by grid-connected photovoltaic systems. IEEE Trans Energy
Conver 2015;30(2):413–20.
[24] Roumpakias Elias, Stamatelos Anastassios. Comparative performance analysis of
grid-connected photovoltaic system by use of existing performance models. Energy
Convers Manage 2017;150:14–25.
[25] Danling Cheng, Mather Barry A, Richard Seguin, Joshua Hambrick, Broadwater
Robert P. Photovoltaic (PV) impact assessment for very high penetration levels.
IEEE J Photovoltaics 2016;6(1):295–300.
[26] Hossein Hooshyar, Baran Mesut E. Fault analysis on distribution feeders with high
penetration of PV systems. IEEE Trans Power Syst 2013;28(3):2890–6.
[27] Baran Mesut E, Hooshyar Hossein, Shen Zhan, Huang Alex. Accommodating high
PV penetration on distribution feeders. IEEE Trans Smart Grid 2012;3(2):1039–46.
[28] Tekpeti BS, Kang X, Kheshti M. Particular PV grid-connected system under sym-
metrical and asymmetrical faults analysis. In: Electrical power and energy con-
ference on 22–25 October 2017, Saskatoon, Canada. p. 1–6.
[29] Tekpeti BS, Kang X, Kheshti M. Fault analysis of accurate PV connected system
against dynamic fast solar radiation fluctuation. In: Electrical power and energy
conference on 22–25 October 2017, Saskatoon, Canada. p. 1–6.
[30] Firouz Y, Farhadkhani S, Lobry J, Vallee F, Khakpour A, Durieux O. Numerical
comparison of the effects of different types of different types of distributed gen-
eration units on overcurrent protection systems in MV distribution grids. Renewable
Energy 2014;69:271–83.
[31] Kobayashi Hiromu, Takigawa Kiyoshi. A new protective method for grid connected
dispersed PV systems to detect short circuit fault in distribution line. Sol Energy
Mater Sol Cells 1997;47(1–4):117–23.
[32] Conti Stefania, Nicotra Sebastiano. Procedures for fault location and isolation to
solve protection selectivity problems in MV distribution networks with dispersed
generation. Electr Power Syst Res 2009;79(1):57–64.
[33] Gimenez Ledesma Jorge Javier, de Araujo Leandro Ramos, Ribeiro Penido Debora
Rosana. A method for evaluation of overcurrent protection in unbalanced dis-
tribution systems. Int Trans Electr Energy Syst 2016;26(2):412–28.
[34] Tremblay Olivier, Dessaint Louis-A, Dekkiche Abdel-Illah. A generic battery model
for dynamic simulation of hybrid electrical vehicles. In: Vehicle power and pro-
pulsion conference on 9–12 September 2007, Arlington, Texas. p. 284–89.
[35] Muhamad DZ, Azah M, Hannan HA. Optimization of PI compensator parameters for
grid-tied photovoltaic with energy storage systems using simplex algorithm. Int Rev
Model Simulat 2012;5(2):751–60.
[36] Chabanloo RM, Mohammadzadeh N. A fast numerical method for optimal co-
ordination of overcurrent relays in the presence of transient fault current. IET Gener
Transm Distrib 2018;12(2):472–81.
[37] Shih MY, Conde A, Leonowicz Z, Martirano L. An adaptive overcurrent coordination
scheme to improve relay sensitivity and overcome drawbacks due to distributed
generation in smart grids. IEEE Trans Indus Appl 2017;53(6):5217–28.
[38] Salem MM, Elkalashy NI, Atia Y, Kawady TA. Modified inverter control of dis-
tributed generation for enhanced relaying coordination in distribution networks.
IEEE Trans Power Deliv 2017;32(1):78–87.
[39] Fani B, Hadi Bisheh, Karami-Horestani A. An offline penetration-free protection
scheme for PV-dominated distributed systems. Electr Power Syst Res 2018;157:1–9.