A novel hybrid islanding detection method for inverter-based DGs using SFS and ROCOF

Abstract

This paper is aimed at proposing a new hybrid method for the islanding detection of Distributed Generation (DG) units. Hybrid method operation is based on the combination of an active and a passive method, for which, Optimized Sandia Frequency Shift (SFS) method is used as the selected active method, and Rate of Change of Frequency relay (ROCOF) is used as the passive method. In order to demonstrate the effectiveness of the proposed technique on islanding detection, several simulation studies based on IEEE 1547 and UL1741 anti-islanding test requirements are carried out on a system equipped with the proposed hybrid method. The evaluation of simulation results reveals that the control system, based on the proposed hybrid algorithm, meets the DG islanding protection requirements efficiently. In other words, not only it holds all the benefits of both SFS and ROCOF, but also it removes their drawbacks by providing smaller Non Detection Zone, improved system's power quality and higher speed of response, and provides some other advantages. Moreover, it will be demonstrated that the proposed hybrid method is capable of accurately operating under multiple DG units, load switching in the grid connected mode, as well as different load quality factor conditions. Index Terms—Inverter-based distributed generation (DG), islanding detection technique, ROCOF relay, SFS method, hybrid method, negligible Non Detection Zone (NDZ), power quality improvement of system, multi DG system.

Full text

0885-8977 (c) 2015 IEEE. Translations and content mining are permitted for academic research only. Personal use is also permitted, but republication/redistribution requires IEEE
permission. See http://www.ieee.org/publications_standards/publications/rights/index.html for more information.
This article has been accepted for publication in a future issue of this journal, but has not been fully edited. Content may change prior to final publication. Citation information: DOI
10.1109/TPWRD.2015.2406577, IEEE Transactions on Power Delivery
Abstract—This paper is aimed at proposing a new hybrid
method for the islanding detection of Distributed Generation
(DG) units. Hybrid method operation is based on the
combination of an active and a passive method, for which,
Optimized Sandia Frequency Shift (SFS) method is used as
the selected active method, and Rate of Change of Frequency
relay (ROCOF) is used as the passive method. In order to
demonstrate the effectiveness of the proposed technique on
islanding detection, several simulation studies based on IEEE
1547 and UL1741 anti-islanding test requirements are carried
out on a system equipped with the proposed hybrid method.
The evaluation of simulation results reveals that the control
system, based on the proposed hybrid algorithm, meets the
DG islanding protection requirements efficiently. In other
words, not only it holds all the benefits of both SFS and
ROCOF, but also it removes their drawbacks by providing
smaller Non Detection Zone, improved system’s power quality
and higher speed of response, and provides some other
advantages. Moreover, it will be demonstrated that the
proposed hybrid method is capable of accurately operating
under multiple DG units, load switching in the grid connected
mode, as well as different load quality factor conditions.
Index Terms—Inverter-based distributed generation (DG),
islanding detection technique, ROCOF relay, SFS method,
hybrid method, negligible Non Detection Zone (NDZ), power
quality improvement of system, multi DG system.
I. INTRODUCTION
oday, the performance of distributed generations (DGs)
and microgrids depends, to an extent, upon the DGs’
operational conditions and the protection systems. Many
challenges arise when connecting DG to the power grid.
One of the most important challenges, widely debated over
the years, is islanding phenomenon. According to IEEE
standard1547, islanding is defined as a condition in which a
part of the power system becomes isolated from the rest of
network, and the separated part is operated with at least one
DG [1]. Unintentional islanded condition is an undesirable
mode of operation, which may pose existential threat to the
DG, the utilities’ systems, as well as the consumer’s
devices. Moreover, it may threat the safety of utility’s staff,
as well as customers. The reasons for islanding occurrence
and its consequences are discussed in [1], [2], and [3]. As
an important DG protection requirement, it should be
equipped with a proper anti-islanding protection scheme.
Over the last few years, many islanding detection methods
have been proposed which can be classified into three main
categories: communication based methods, passive
methods, and active methods. Communication based
methods operate based on transferring data between the
electric utility and the DG unit, which are reliable and
effective, but their main drawback is the cost of
implementation [3]. The most common communication
based methods used for islanding detection are discussed in
[4-5]. Passive islanding detection methods are based on
controlling the system parameters and monitoring their
changes at the Point of Common Coupling (PCC) [6].
Some commonly used passive methods are Under/Over
Voltage Protection and Under/Over Frequency Protection
methods [7]-[8], Voltage Phase Jump Detection method
(PJD) [9], Rate Of Change Of Frequency protection
method [10], Rate Of Change Of Active Power method
[11], and voltage unbalance and total harmonic distortion
(VU/THD) technique [12]. Passive methods are known as
islanding detection methods which cause no disturbance in
the system. The main drawback of passive methods is their
large Non Detection Zone (NDZ) which makes them fail
when there is a small power mismatch between DG and the
load. In other words, when load and DG are closely
matched in islanded system, the change of voltage as well
as frequency at PCC will be small, and the mentioned
passive methods fail to detect islanded conditions [8].
However, active islanding detection methods are based on
the occurrence of disturbance on the terminal of DG, which
will have a significant change when DG gets disconnected
from the power grid [13]; therefore, active methods stay
effective, even if there is a small power mismatch between
load demand and DG capacity. The dominant active
techniques are: Slip-Mode frequency Shift method (SMS)
[14], Active Frequency Drift method (AFD) [4], Sandia
Frequency Shift method (SFS) [15], and Sandia Voltage
Shift method (SVS) [15]. The drawbacks of active methods
are the reduction of power quality, and the lower speed of
detection in comparison with passive methods [13]. As it
can be seen, active and passive methods have their own
advantages and drawbacks. Combining these two methods
in order to get benefit from all the advantages, results in a
new category of islanding protection techniques called
hybrid anti-islanding methods [15]. Some of the recent
literature focused on hybrid islanding detection techniques
are: A robust hybrid islanding detection method based on
the combination of current injection method as the active
method, and a frequency relay, a voltage relay and the THD
method as the passive method, is proposed in [15].
Reference [16] combines the principles of positive
feedback (active) technique, and voltage unbalance and
total harmonic distortion (passive) technique as an efficient
A Novel Hybrid Islanding Detection Method for
Inverter-Based DGs Using SFS and ROCOF
Mahdiyeh Khodaparastan, Hesan Vahedi, Student Member, IEEE, Farid Khazaeli,
Hashem Oraee, Senior Member, IEEE
T

0885-8977 (c) 2015 IEEE. Translations and content mining are permitted for academic research only. Personal use is also permitted, but republication/redistribution requires IEEE
permission. See http://www.ieee.org/publications_standards/publications/rights/index.html for more information.
This article has been accepted for publication in a future issue of this journal, but has not been fully edited. Content may change prior to final publication. Citation information: DOI
10.1109/TPWRD.2015.2406577, IEEE Transactions on Power Delivery
method for synchronously rotating DGs. In [17], a digital
signal processor (DSP) calculates the rate of change of
frequency, rate of change of voltage, then correction factor
of distributed synchronous generator decides whether the
trip conditions are met. A hybrid method based on
combining the average rate of voltage change, and the real
power shift is proposed in [18], in which an average rate of
voltage change (passive technique) has been used to initiate
a real power shift (active technique).
In this paper, a novel and robust hybrid islanding
detection method, based on SFS method as the active
method,and a combination and coordination of ROCOF
and under/over frequency relays as the passive method, is
proposed. The performance of SFS method depends on its
positive feedback gain [19]. The Imperialist Competitive
Algorithm (ICA) [20] is used to find the optimal value of
positive feedback gain. To evaluate the effectiveness of the
proposed method, a system equipped with the proposed
hybrid method is studied under IEEE standard 1547 [1] and
UL1741 test conditions [21]. Different load quality factors
and multiple DG units are considered for the evaluation of
system’s operation. The rest of paper is organized as
follows: Section II describes the system under study. SFS
method, ROCOF method, and the proposed hybrid
technique are discussed in section III. Section IV is
focussed on the simulation studies and results. Finally, in
section V, the paper is concluded.
II. SYSTEM DESCRIPTION
The system under study, which is set up to study the
islanded conditions, is depicted in Fig. 1. As it can be seen,
the main grid is modelled by a balanced three-phase
voltage source with external RL impedance (Rsand Ls).
Moreover, DG is modelled by a constant current source
feeding a DC/AC converter. This DG can be used to model
a photovoltaic power station, a fuel cell or any kind of
inverter-based distributed generations. The H-bridge
inverter is implemented using Insulated Gate Bipolar
Transistors (IGBTs), and a pulse width modulation (PWM)
technique for the switching signals of the inverter. The
constant parallel RLC branch is used as the load model.
The main grid is connected to the load as well as DG
through the circuit breaker CB. The breaker CB is used in
order to yield the islanded condition [22]. The series power
filter, which compensates current harmonics, is modelled
by Lf, and the equivalent resistance of this inductance is
modelled by Rf.
Fig. 1. Single-line diagram of the system.
DG interface controller, which is shown in Fig. 2, is a
constant current controller providing the sinusoidal pulse
width modulator’s with the required parameters in order to
produce the proper switching pulses of inverter. The
controller transforms the three-phase inverter output
currents into d-q components. The Idcomponent is used for
controlling the DG’s output active power, and the Iq
component is used for controlling the DG’s output reactive
power. Regarding the fact that preferred operation is at
unity power factor, the reference value of q-axis current (Iq-
ref) is set to zero [6].
Fig. 2. Block diagram of the inverter controller.
III. HYBRID METHOD FOR ISLANDING DETECTION
Active and passive methods have some negative and
positive aspects. Passive methods do not have any impact
on power system; however, they have a large NDZ [3].
Active methods are typically accurate, but they have
undesirable impact on system’s power quality [13]. When
these two methods come together, it is not only possible to
receive benefit from their advantages; it is possible to
overcome their undesirable features also. In the following
section, SFS method, ROCOF method, and the proposed
hybrid method are described.
A. Sandia Frequency Shift method (SFS)
SFS method is an efficient active anti-islanding method
introduced by Sandia Laboratory. Among the active
methods, SFS is known because of the fast speed of
response as well as its small NDZ [23]. SFS method
operates based on injecting a small disturbance into the
frequency of voltage at PCC, which can make a significant
change in frequency when islanding occurs. The operation
of SFS method can be expressed by equation (1) [24]:
))((
20ffkcf isinv  

(1)
where δinv is the inverter phase angle, fis is the frequency
of islanded system, fis the frequency of system in grid
connected mode, cf0is the chopping fraction when there is
no frequency error.Additionally, the positive feedback gain
is represented by k. SFS can be modelled by phase angle
transformation in control block diagram as shown in Fig. 3
[25]. As can be seen, the reference values of d-q current
components Id-ref and Iq-ref, are transformed through the
phase angle transformation matrix to new d-q components
Id-ref*and Iq-ref*.Then the new current components are used
for regulating inverter output current.
Fig. 3. SFS block diagram model in control interface scheme.

0885-8977 (c) 2015 IEEE. Translations and content mining are permitted for academic research only. Personal use is also permitted, but republication/redistribution requires IEEE
permission. See http://www.ieee.org/publications_standards/publications/rights/index.html for more information.
This article has been accepted for publication in a future issue of this journal, but has not been fully edited. Content may change prior to final publication. Citation information: DOI
10.1109/TPWRD.2015.2406577, IEEE Transactions on Power Delivery
In order to study the steady state condition in islanded
mode of operation, the inverter phase angle (δinv) should be
equated with the load phase angle. This condition can be
applied as an equal constraint, which can be represented as
follows [23]:
)]([tan 0
01 f
f
f
f
Qis
is
fload  

(2)

load

inv

→(3)
0
]tan[ 2
0
2
0 is
f
invis ff
Q
f
f
(4)
where
f
Q
is the load quality factor and
0
f
is the load’s
resonance frequency.
As can be seen, equation (4) states that NDZ of SFS
depends on k, cf0, Qfand f0[23].As it is demonstrated in
[23] and [25], in order to eliminate NDZ, the answer point
of equation (4) must be an unstable point. This constraint
results in the following equation:
is
inv
is
load df
d
df
d 
(5)
Moreover, using equations (1) and (2) it can be
concluded that:
k
df
d
is
inv 2


(6)
2
0
02
0
2
0
1
1

















f
f
f
f
Q
f
f
f
Q
df
d
is
is
f
is
f
is
load

(7)
Using equations (5), (6) and (7), the following unequal
constraint for kis resulted:

































0
0
2
0
2
0
1
1
2
f
f
f
f
Q
f
f
f
Q
kis
is
f
is
f

(8)
Under different loading conditions, different values for
Qfand f0are resulted. It, according to equation (8), leads to
different values for k, which is required in order to
guarantee the frequency deviations outside the relay
settings [23], [26].
B. Rate Of Change Of Frequency Method (ROCOF)
ROCOF relay is a passive islanding detection method
based on computing the rate of change of frequency. When
DG is islanded, the frequency of voltage at PCC is
confronted with a high df/dt rate of change. In order to
identify such a condition, df/dt is measured over a few
cycles, usually between 2 and 50 cycles by ROCOF relay,
then the required signal for islanding detection is computed
and sent to a low-pass filter. The application of low-pass
filter is aimed at eliminating high-frequency transients
made by power system components. The flow chart
diagram, which describes the operation of ROCOF method,
is shown in Fig. 4. As it can be seen, the output signal of
the low-pass filter is compared to the ROCOF relay setting
(β). In the case that the rate of change of frequency is
bigger than β, a trip signal will be sent to the DG’s circuit
breaker. ROCOF relay setting considering the system
nominal frequency equal to 60 Hz,is in the range of 0.1
(Hz/s) - 1.2 (Hz/s)[10], [26], [27].
Fig. 4. Block diagram of ROCOF method in islanding detection.
To avoid ROCOF relay’s malfunction during long
electrical motors start-up or short circuit fault conditions,
there is a block which compares the terminal voltage to a
constant minimum adjustable voltage setting. In other
words, when terminal voltage is reduced to a level less than
the minimum adjustable voltage setting (Vmin), the ROCOF
relay’s trip signal will be inactive[10], [26].
C. The Proposed Hybrid Method
In the proposed hybrid method, SFS is used as the active
method, and ROCOF as the passive method. By merging
the SFS and ROCOF methods, the islanding detection
performance can be improved. In the proposed hybrid
technique, SFS method gets activated only when islanded
condition is suspected by ROCOF relay. The block diagram
of the proposed method is shown in Fig. 5. System
frequency is estimated using PLL. When ROCOF relay
detects any variation in df/dt, a trip signal will be sent to a
multiple switch which activates SFS signal; whereas, in the
case of no disturbance in the system, ROCOF relay will not
send any signal to the multiple switch, which results in the
predefined Id-ref and Iq-ref current references get switched on.
Therefore, SFS will not be implemented until the next
disturbance occurs. The major problem with ROCOF
method is that the detection of islanding is difficult when
the load and generation capacities inside the islanded
system are close to each other. Furthermore, special care
should be taken while setting the thresholds for ROCOF
relay. If the threshold is set too low, it may lead to nuisance
trips of DG; and if the threshold is set too high,

0885-8977 (c) 2015 IEEE. Translations and content mining are permitted for academic research only. Personal use is also permitted, but republication/redistribution requires IEEE
permission. See http://www.ieee.org/publications_standards/publications/rights/index.html for more information.
This article has been accepted for publication in a future issue of this journal, but has not been fully edited. Content may change prior to final publication. Citation information: DOI
10.1109/TPWRD.2015.2406577, IEEE Transactions on Power Delivery
refd
refq
1V
V
tan



refq
2
refq
2VV  
refd
Vrefq
V
1
tan
2refd
V
2refq
V














invinv
invinv   cossin sincos
inv

sT1 1
a

Fig. 5. Block diagram of the proposed hybrid method.
islanding may not be detected. In the proposed hybrid
method, ROCOF relay calculates df/dt, then if islanding is
suspected, an activation signal will be sent to SFS; as a
result, SFS plays a major role in islanding detection.
A key point about merging ROCOF and SFS is keeping
the ROCOF threshold level as low as possible, which
makes it possible to detect islanding occurrence even if the
DG and load capacities are closely matched. In this study,
regarding the fact that the maximum normal frequency
deviation of under study system is about 1Hz/s; thereupon,
the ROCOF relay threshold (β) is set to be 1.2Hz/s. This is
due to the fact that DG size is small in comparison with
grid and also DG is an inverter-based type. In inverter-
based DGs, switching devices are used so that having
maximum frequency deviation around 1 Hz/s is normal and
unavoidable. The other point is related to the fact that
ROCOF relay’s trip signal will be inactive in case of large
motors start up, helping the proposed hybrid method not to
confuse motor start up condition with islanding occurrence.
Vmin is also set to be 0.85 in this study. Additionally,
regarding the fact that SFS is not continuously operating in
the proposed method, the power quality of system will
significantly improve. On the other hand, the performance
of SFS method depends on the value of its positive
feedback gain (k). The higher the value of k, the faster the
speed of islanding detection and the smaller the non-
detection zone [16]. However, higher kvalues may cause
false trips and reduce the power quality. Using the
proposed method, due to the fact that SFS is not
continuously active; as a result, it is possible to choose
higher values for kwithout any concern about false
protection trips and power quality problem. In this study, k
value is set to be 0.08. In section IV, the simulation studies
and results will demonstrate the stated achievements.
IV. SIMULATION STUDIES AND THE RESULTS
The system depicted in Fig. 1, with the parameters
values in Table I, is simulated using Matlab/Simulink.
TABLE I
PARAMETERS OF THE SIMULATED SYSTEM
DG rated output power
120 kVA
DG output active power
100 kW
Switching frequency
8 kHz
Inverter’s input DC
voltage
900 V
PCC voltage (L-L)
480 V
Nominal frequency (
f
)
60 Hz
Grid resistance (
s
R
)
0.012 Ω
Grid inductance (
s
L
)
0.3056 mH
Filter inductance (
f
L
)
800 µH
Filter resistance (
f
R
)
0.004 Ω
DG inverter controller
parameters
Idcontroller constants
Kp= 0.2,
Ki= 100
Iqcontroller constants
Kp= 0.2,
Ki= 100
The performance of the proposed hybrid method is
studied under different conditions including: different
loading conditions, grid connected load switching, motor
and capacitor switching and three-phase fault occurrence,
operation in weak grid, various load quality factors, and
multiple DGs operational conditions. The UL1741
[24]/IEEE1547 [1] combination of requirements are used to
evaluate the operation of the proposed method.
As mentioned earlier, the performance of SFS method
depends on its positive feedback gain (k) value. To find the
optimal kvalues versus different loading conditions, ICA,
which is a fast and reliable optimization technique, is used
[20]. R,L, and Cparameters for different loading
conditions are given in Table II, and the optimization
results are shown in Table III:
TABLE II
OPTIMUM kSFS VALUES FOR DIFFERENT LOADING
CONDITIONS
Qf
kSFSOpt
1
0.02133
1.8
0.04267
2.5
0.05332
3
0.06402
4
0.08034
5
0.09716
Using high values for kSFS,frequency can deviate from
the normal range for different loading conditions, which is
a positive point in islanding protection. However, it is
important to note that high values for kmay cause DG

0885-8977 (c) 2015 IEEE. Translations and content mining are permitted for academic research only. Personal use is also permitted, but republication/redistribution requires IEEE
permission. See http://www.ieee.org/publications_standards/publications/rights/index.html for more information.
This article has been accepted for publication in a future issue of this journal, but has not been fully edited. Content may change prior to final publication. Citation information: DOI
10.1109/TPWRD.2015.2406577, IEEE Transactions on Power Delivery
become unstable even if DG operates under a grid
connected mode [7]. Therefore, it is important to choose a
safe value for kwhich guarantees stable operation of DG in
grid connected mode as well as a fast and efficient
islanding detection. Due to the fact that SFS is not
continuously active in the proposed hybrid method; as a
result, higher values for kcan be chosen. In this paper, kis
set at 0.08.
A. System Evaluation under UL1741 Test Conditions
The proposed hybrid method is evaluated under grid-
connected and islanded DG operational conditions based on
the test condition presented in [21]. According to
UL1741Std., the load active power is adjusted so that the
inverter power is 25%, 50%, 100%, and 125% of inverter
and DG’s rated power. The reactive power is also adjusted
to be between -5% and +5% of the rated active power in
1% steps [21]. In order to evaluate the operation of the
proposed islanding detection method within the system for
study under islanded condition, five samples of the
simulation results are presented in this section, which are as
follows:
Case I: the load is adjusted at 100% of the rated active
power, and 0% reactive power (unity power factor).
Case II: the load is adjusted at 50% of the rated active
power, and 0% reactive power (unity power factor).
Case III: the load is adjusted at 125% of the rated active
power, and 0% reactive power (unity power factor).
Case IV: the load is adjusted at 100% of the rated active
power, and -1% reactive power (lag power factor).
Case V: the load is adjusted at 100% of the rated active
power, and +1% reactive power (lead power factor).
R, L, C parameters of the load, for the above cases, are
presented in Table IV [28].
TABLE IV
Load R, L, C PARAMETERS FOR DIFFERENT CASES OF TEST
CONDITION
P%
Q%
R(Ω)
L(H)
C(µF)
100
100
2.304
0.00345
2037
50
100
4.603
0.00345
2037
125
100
1.841
0.00345
2037
100
99
2.304
0.003488
2037
100
101
2.304
0.003419
2037
In this section, the performance of the system equiped
with the proposed hybrid method is investigated under
islanded condition. Simulation results for the five defined
cases are demonstrated in Fig. 6. To evaluate the
effectiveness of the proposed hybrid method, simulation
results of system without hybrid method for the defined
cases is represented in Fig. 7. In Fig. 6, as can clearly be
seen, during the normal operational condition of DG,
frequency is fixed at the nominal value, and DG has a
stable operation. However, when islanding occurs at t=2s,
DG fails to keep its stable operation in all cases, and
frequency deviates from under and over frequency
thresholds. It can be seen when there is active power
Fig. 6. PCC frequency for the system equipped with the proposed hybrid
method.
Fig. 7. PCC frequency for the system without the proposed hybrid method.
mismatch between load and DG (cases II and III),
frequency deviates faster than the balanced condition in
case I. The same result is evident when there is reactive
power mismatch between load and DG (cases IV and V); in
these cases, frequency deviates faster in comparison with
the balanced condition (case I). In Fig. 7, it can clearly be
seen that, for all cases, the frequency does not deviate from
the normal range in the system without hybrid method; as a
result, islanding cannot be detected. Regarding the fact that
for all cases of UL1741 test condition, using the proposed
hybrid method, the frequency deviates from the normal
range, it can be concluded that the proposed hybrid method
provides an acceptable performance.
B. Hybrid Method Operation under Load Switching in the
Grid Connected Mode
In this section, the proposed hybrid method is tested
under load switching conditions in the grid connected mode
of operation. The purpose of the test is to ensure that the
proposed technique is not sensitive to load switching under
grid connected mode of DG operation. In other words, in
order to prevent from a false operation, the proposed
method should be able to distinguish between load
switching conditions and islanding conditions. In the test
procedure, an additional load is brought in at t= 1.5s and
switched off at t=2.5s. Three different cases are considered
in order of describing the additional load:
Case I: the load with 100kVA power with 0.8 lag PF is
added to the base load at t=1.5s.
Case II: the load with 100kVA power with 0.8 lead PF is
added to the base load at t=1.5s.
Case III: the load with 100kVA power with unity PF is
added to the base load at t=1.5s.
The base load is adjusted at 100kW active power with
unity PF. The frequency-time curve, for the three above
cases, is shown in Fig. 8. As it can be seen in Fig. 8, during
load switching conditions in grid connected mode of
operation, frequency maintains the same level, though the
additional load is switched in and off. In other words, the
proposed method is not sensitive to load switching and
does not give a false trip.
TABLE III
LOAD R, L, C PARAMETERS FOR DIFFRENT QfVALUES
Qf
R(Ω)
L(H)
C(F)
1
2.304
0.00607
0.00115
1.8
2.304
0.00304
0.00231
2.5
2.304
0.00244
0.00288
3
2.304
0.00203
0.00346
4
2.304
0.00152
0.00461
5
2.304
0.00122
0.00575

0885-8977 (c) 2015 IEEE. Translations and content mining are permitted for academic research only. Personal use is also permitted, but republication/redistribution requires IEEE
permission. See http://www.ieee.org/publications_standards/publications/rights/index.html for more information.
This article has been accepted for publication in a future issue of this journal, but has not been fully edited. Content may change prior to final publication. Citation information: DOI
10.1109/TPWRD.2015.2406577, IEEE Transactions on Power Delivery
Fig.8. PCC frequency: (a) case I (b) case II (c) case III.
C. Operation under Motor and Capacitor Switching
In this section, the proposed hybrid method is tested
under motor and capacitor switching conditions in the grid
connected mode of operation. In similarity to section B, the
purpose of the test in here is ensuring that the proposed
technique is not sensitive to the events such as motor and
capacitor switching, and does not operate falsely. To test
operation under motor switching condition, an induction
motor load (150HP, 480v, 60Hz, 1785rpm and 15N.m) is
switched in at t=1.5s and is Switched off at t=2.5s. The
simulation result including system frequency is shown in
Fig. 9. In order to examine the performance of proposed
method under capacitor switching condition,
a150kVAR capacitor bank is switched in at t=1.5s and
switched off at t=2.5s. The simulation result is depicted in
Fig.10.
Fig. 9. PCC frequency in case of motor switching.
Fig. 10. PCC frequency in case of capacitor switching.
D. Hybrid Method Operation under Fault Occurrence
Fault occurrence is an abnormal condition in system
operation, in which voltage and frequency deviate from
their normal values. Small fault impedances result in large
fault currents which cause voltage and frequency deviate
from the OV/UV and OF/UF relays’boundaries. However,
when the impedance of a fault is large, the voltage and
frequency do not deviate from the setting of voltage and
frequency relays (non-detection zone). In this case, anti-
islanding detection may confuse the case with islanding
condition, and trip falsely. To test the performance of
proposed method during high impedance faults, a three-
phase fault with the fault impedance equal to 1.5Ωhas
been switched in at t=1.5s and cleared after 0.2 s. Fig. 11
illustrates the simulation
result.
Fig. 11. PCC frequency in case of fault occurrence.
E. Hybrid Method Operation in a Weak Grid
Another condition, needed to be considered for
examining the effectiveness of proposed method, is
operating under a weak grid case. A power system is
considered weak when the line impedance or X/R ratio is
relatively large. In this condition, power system has a low
short-circuit level and small stability margin. Using
positive feedback anti-islanding methods can cause a
stability problem since intentional disturbance is injected to
the system by these methods [29]. In a weak grid case,
frequency deviates more than in a strong grid. If this
frequency deviation exceeds the ROCOF relay threshold,
the SFS method will be initiated and might cause a stability
problem due to the reason explained above. In order to test
the proposed method, firstly, a constant value for X/R ratio
equal to 5is considered, and system impedance (Z) is
changing from 0.1 to 0.4 Ω. The results are displayed in
Fig. 12. In the other test, the constant value of 0.2 is
considered for Z, and X/R ratio is changed from 5to 20.
The results are depicted in Fig. 13.
Fig.12. PCC frequency in case of weak grid (different Z).
Fig.13. PCC frequency in case of weak grid (different X/R).
It can clearly be seen in Fig. 12 and Fig. 13 that the
proposed method performs normally in case of different
line impedances and X/R ratios, and does not have any
negative influence on the system stability.
F. Different Load Quality Factors
There are different standards and test conditions each
considering a different load quality factor in order to test
the operation of islanding protection methods. For instance,
in IEEE Std. 929, the use of
5.2Qf
is offered; however,
IEEE 1547 and UL1741 test conditions suggest the use of
1Qf
and
8.1Qf
respectively. In the United States, a
quality factor equal to 2.5, and, in the United Kingdom, a
quality factor higher than 0.5 is recommended in order to
evaluate the performance of islanding detection methods
[6]. In this section, in order to show that the proposed
hybrid method is efficient under all test requirements
defined by different standards, the system under study is
simulated for a wide range of
f
Q
changes varying from

0885-8977 (c) 2015 IEEE. Translations and content mining are permitted for academic research only. Personal use is also permitted, but republication/redistribution requires IEEE
permission. See http://www.ieee.org/publications_standards/publications/rights/index.html for more information.
This article has been accepted for publication in a future issue of this journal, but has not been fully edited. Content may change prior to final publication. Citation information: DOI
10.1109/TPWRD.2015.2406577, IEEE Transactions on Power Delivery
0.5 to 5[6]. The positive feedback gain is set on 0.08.The
load R, L,Cparameters for different
f
Q
values, is
presented in Table V. Simulation results are depicted in
Fig. 14.
TABLE V
LOAD R, L, C PARAMETERS FOR DIFFERENT
f
Q
VALUES
Qf
R(Ω)
L(H)
C(F)
0.5
2.304
0.0122
0.000575
1
2.304
0.00607
0.00115
1.77
2.304
0.00345
0.00237
2.5
2.304
0.00244
0.00288
3
2.304
0.00203
0.00346
5
2.304
0.00122
0.00575
Fig. 14. Frequency deviation of the hybrid method for various
f
Q
values.
As it can be seen in Fig. 14, using the proposed hybrid
method, during a small period of time, the frequency
deviates from under/over frequency relay thresholds for all
f
Q
values. In other words, for a broad range of
f
Q
changes, considered by different standards and test
requirements, islanded condition can be detected by the
proposed hybrid method within a short period of time after
islanding occurrence. To evaluate the efficiency of the
proposed method in comparison with SFS method, all
simulations are repeated for the system equipped with SFS
(k=0.05).As it was mentioned before increasing value of k
in SFS method risk the system stability and have negative
impact on power quality.while increasing k in the poposed
method has a relatively low negative effect on power
quality of the distribution system. This is because the SFS
part is not injecting the perturbations to the grid all the
time.The results are demonstrated in Fig. 15. It can clearly
be seen that by the use of the proposed hybrid method, the
frequency deviations, for different
f
Q
values, is much
faster than the case where SFS method is used.
Additionally, when SFS method is used, the frequency does
not deviate from under/over frequency relay thresholds for
the loads with high
f
Q
values (for example
5Q f
); as a
result, the islanded condition may not be detected.
However, using the proposed hybrid method, the frequency
is deviating from relay thresholds, for the full range of
f
Q
changes defined by different standards, in a fraction of
a second after islanding occurs.Therefore, it can be
concluded that the proposed hybrid technique has a smaller
NDZ and a faster speed of response in comparison with
SFS method.
Fig. 15. Frequency deviation for different
f
Q
values, using SFS method.
G. Multiple DG Units
In this section, the operation of the proposed hybrid
method is evaluated under the presence of multiple DG
units. As depicted in Fig. 16, two inverter-based DG units
each one rated at an output active power equal to 100kW,
both equipped by the proposed hybrid method, are
connected to the grid at PCC.
Fig. 16. Single-line diagram of the system including multiple DG unit.
In the simulation, islanding occurs at t=2s and
simulation is carried out for the following three cases:
Case I: the load is adjusted to receive 200kW from DG unit
I and DG unit II, with the quality factor
885.0Qf
and the
resonance frequency
Hz5.60f0
.
Case II: the load is adjusted to receive 200kW from DG
unit I and DG unit II, with the quality factor
885.0Qf
and
the resonance frequency
Hz60f0
.
Case III: the load is adjusted to receive 200kW from DG
unit I and DG unit II, with the quality factor
885.0Qf
and
the resonance frequency
Hz3.59f0
.
The load R,L, and Cparameters for the above cases are
presented in Table VI. The Simulation results are depicted
in Fig. 17.
TABLE VI
LOAD R, L, C PARAMETERS FOR THE THREE CASES
Resonance frequency(Hz)
R(Ω)
L(H)
C(F)
60.5
2.304
0.00399
0.002037
60
2.304
0.00345
0.002037
59.3
2.304
0.003526
0.002037
Fig. 17. PCC frequency deviation, in multiple DGs case, under different
loading conditions.
As it can be seen, in all the three cases, when islanded
condition occurs, the PCC frequency deviates from
under/over frequency relay thresholds within a short period
of time, and islanding occurrence is rapidly identified.
Therefore, the proposed hybrid method can easily be
implemented on a multiple DGs system using inverter
based DGs. In the above analysis, it is assumed that both
DGs are connected to the same location and equipped with
the proposed hybrid method with the same parameters.
Other situations for distributed generation units such as
different SFS setting parameters, Different locations, as
well as different control schemes can be considered. As
mentioned earlier, in the proposed method, frequency

0885-8977 (c) 2015 IEEE. Translations and content mining are permitted for academic research only. Personal use is also permitted, but republication/redistribution requires IEEE
permission. See http://www.ieee.org/publications_standards/publications/rights/index.html for more information.
This article has been accepted for publication in a future issue of this journal, but has not been fully edited. Content may change prior to final publication. Citation information: DOI
10.1109/TPWRD.2015.2406577, IEEE Transactions on Power Delivery
oscillation detected by ROCOF relay leads to SFS method
activation. Therefore, multiple DGs operational case is like
when it is intended to consider multiple DGs case for SFS
method. Comprehensive studies focussed on multiple DGs
operation for distributed generations equipped with SFS
method have been carried out in [31] and [30].
V. CONCLUSIONS
The effectiveness of a new hybrid islanding detection
technique based on the combination of Sandia Frequency
Shift method and Rate of Change of Frequency relay is
investigated in this paper. The performance of SFS method
depends on its positive feedback gain (k). The optimum
value of kfor different load conditions using Imperialist
Competitive Algorithm is determined. To benefit from
positive aspects of SFS and ROCOF, these methods are
merged as a proposed hybrid method. The resulted method
has been studied under different conditions of test
requirements. The simulation studies concluded the
following results on the efficiency of the proposed hybrid
method:
I. The proposed hybrid islanding detection technique is
accurate, and can detect islanding occurrence fast,
though there is a small mismatch between DG and
load active/reactive powers.
II. The proposed technique decreases NDZ and improves
the speed of response in comparison with SFS.
III. This method improves the steady state power quality
of the system, because active method is not
continuously operating; as a result, disturbance is not
continuously injected to the system.
IV. This method can discriminate between the load
switching conditions and the islanded condition; as a
result, preventing false trips in the case of load
switching occurrence.
V. This method operates properly when it is
implemented on a multi DG system.
VI. REFERENCES
[1] “IEEE Standard for Interconnecting Distributed Resources with
Electric Power Systems”, IEEE Std. 1547, Jul.2003.
[2] N. Kan’an, L. Farouk, H. Zeineldin, and W. Woon, “Effect of DG
Location on Multi-parameter Passive Islanding Detection Methods”,
in Proc.IEEE Power Energy Soc. Gen. Meet., Jul. 2010.
[3] A. H. K. Alaboudy and H. H. Zeineldin, “Islanding Detection for
Inverter-based DG Coupled with Frequency-dependent Static
Loads”, IEEE Trans. Power Del., vol. 26, Apr. 2011.
[4] L. A. C. Lopes, and H. Sun, “Performance Assessment of Active
Frequency Drifting Islanding Detection Methods”, IEEE Trans.
Energy Conv. , Mar. 2006.
[5] M. Ropp and W. Bower, “Evaluation of Islanding Detection
Methods for Photovoltaic Utility Interactive Power Systems”,
International Energy Agency implementing agreement on
photovoltaic power systems, Tech. Rep. IEAPVPST5-09, 2002.
[6] H. H. Zeineldin, E. F. EI-Saadany, and M. M. A. Salama, “Impact of
DG Interface Control on Islanding Detection and Non Detection
Zones”,IEEE Trans. Power Del., vol. 21, no. 3, pp. 1515–1523, Jul.
2006.
[7] H. H. Zeineldin, and J. L. Kirtley, “A Simple Technique for
Islanding Detection with Negligible Non Detection Zone”,IEEE
Trans. Power Del., vol. 24, no. 2, pp. 779–786, Apr. 2009.
[8] H. H. Zeineldin, and J. L. Kirtley, “Performance of the OVP/UVP
and OFP/UFP Method With Voltage and Frequency Dependent
Loads”, IEEE Trans. Power Del., vol. 24, no. 2, April 2009.
[9] B. Singam, and Y. Huil, “Assessing SMS and PJD Schemes of Anti
Islanding with Varying Quality Factor”, In Proc. IEEE Power and
Energy Conf., Nov. 2006.
[10] J.C.M. Vieira, W. Freitas, Z. Huang, W. Xu., A. Morelato,
“Formulas for Predicting the Dynamic Performance of ROCOF
Relays for Embedded Generation Applications”, IEE proceeding
generation and transmission and distribution, 2006.
[11] M. A. Redfern, O. Usta, and G. Fielding, “Protection Against Loss
of Utility Grid Supply for a Dispersed Storage and Generation Unit”,
IEEE Trans. Power Del., vol. 8, no. 3, pp. 948–954, Jul. 1993.
[12] Sung-Il Jang, and Kwang-Ho Kim, “An Islanding Detection Method
for Distributed Generations Using Voltage Unbalance and Total
Harmonic Distortion of Current”, IEEE Trans. Power Delivery, vol.
19, no. 2, pp. 745-752, April 2004.
[13] R. S. Kunte, Wenzhong Gao, “Comparison and Review of Islanding
Detection Techniques for Distributed Energy Resources”, Power
Symposium, 2008.
[14] R.S. Kunte, and W. Gao, “Comparison and Review of Islanding
Detection Techniques for Distributed Energy Resources”, In 40th
North American Power Symposium, 2008.
[15] M. G. M. Abdolrasol, and S. Mekhilef, “Robust Hybrid Anti-
islanding Method for Inverter-based Distributed Generation”,
TENCON conference, 2011.
[16] V. Menon, and M. H. Nehrir, “A Hybrid Islanding Detection
Technique Using Voltage Unbalance and Frequency Set Point”,
IEEE Transactions on Power Systems, 2007.
[17] C. Wen-Yeau, “A Hybrid Islanding Detection Method for
Distributed Synchronous Generators”, in Power Electronics
Conference (IPEC), 2010.
[18] P. Mahat, C. Zhe, and B. Bak-Jensen, “A Hybrid Islanding Detection
Technique Using Average Rate of Voltage Change and Real Power
Shift”, IEEE Transactions on Power Delivery, 2009.
[19] H. H. Zeineldin, and M. M. A. Salama, “Impact of Load Frequency
Dependence on the NDZ and Performance of the SFS Islanding
Detection Method”, IEEE Trans. Ind. Electron., vol.58, no.1,
pp.13914, Jan.2011.
[20] E. Atashpaz-Gargari, and C. Lucas, “Imperialist Competitive
Algorithm: An Algorithm for Optimization Inspired by Imperialistic
Competition”, IEEE Congress on evolutionary computation, 2007.
[21] “The Standard for Inverters, Converters and Controllers for Use in
Independent Power Systems”, Underwriters Laboratories Inc (UL),
Jan.2011.
[22] H. Vahedi, R. Noroozian, A. Jalilvand, G. B. Gharehpetian, “Hybrid
SFS and Q-f Islanding Detection Method for Inverter-Based DG”,
PECON conference, 2010.
[23] H. Zeineldin, and S. Kennedy, “Instability Criterion to Eliminate the
Non-detection Zone of the Sandia Frequency Shift Method”, PSCE
conference, 2009.
[24] M.Ropp, D.Joshi, R.Reedy, A.Shaffer, “New Results for
Power Line Carrier-based Islanding Detection & an Updated
Strengths & Weaknesses Discussion”, PVSC IEEE conference,2011.
[25] X. Wang, W. Freitas, W. Xu, V. Dinavahi, “Impact of DG Interface
Controls on the Sandia Frequency Shift Anti-islanding Method”,
IEEE Transaction on Energy conv., Sept 2007.
[26] W. Freitas, W. Xu, C. M. Affonso, and Z. Huang, “Comparative
analysis between ROCOF and vector surge relays for distributed
generation applications”, IEEE Trans. Power Del., vol. 20, no. 2, pt.
2, pp.1315–1324, Apr. 2005.
[27] R. Bugdal, A. Dysko, G.M. Burt, J.R. McDonald. "Performance
Analysis of the ROCOF and Vector Shift Methods Using a Dynamic
Protection Modelling Approach", 15th International Conference on
Power System Protection, pp. 139-144, 2006.
[28] H. H. Zeineldin, “A (Q-f) Droop Curve for Facilitating Islanding
Detection of Inverter-Based Distributed Generation,” IEEE Trans.
Power Electron.,vol. 24, no. 3, pp. 665–673, March 2009.
X. Wang, W. Freitas, “Impact of Positive Feedback Anti-Islanding
Methods on Small-Signal Stability of Inverter-Based Distributed
Generation”, IEEE Trans. Energy Convers., 23(3): 923–931, 2008.
[29] X. Wang, W. Freitas, and W. Xu, “Dynamic Non-Detection Zones of
Positive Feedback Anti-Islanding Methods for Inverter-Based
Distributed Generators,” IEEE Trans. Power Del., vol. 26, no. 2, pp.
1145–1155, April 2011.
[30] H. H. Zeineldin and S. Conti, “Sandia Frequency Shift Parameter
Selection for Multi-Inverter Systems to Eliminate Non-Detection
Zone”, Inst.Eng. Technol. Renew. Power Gen., vol. 5, no. 2, pp.
175–183, March 2011.

0885-8977 (c) 2015 IEEE. Translations and content mining are permitted for academic research only. Personal use is also permitted, but republication/redistribution requires IEEE
permission. See http://www.ieee.org/publications_standards/publications/rights/index.html for more information.
This article has been accepted for publication in a future issue of this journal, but has not been fully edited. Content may change prior to final publication. Citation information: DOI
10.1109/TPWRD.2015.2406577, IEEE Transactions on Power Delivery
Mahdiyeh khodaparastan received her B.S.c
and M.S.c Degree in Electrical Power
Engineering from Amirkabir University of
Technology,Tehran Iran in 2010 ,2012
respectively.
Her research interest include power system
Protection. Distributed generator, Distribution
system control and operation.
Hesan Vahedi received his M.S. degrees in
electrical engineering in 2010 from Zanjan
University (ZNU), Zanjan, Iran, graduated with
the First Class Honor. He is currently working
toward the Ph.D. degree in electrical
engineering from the Amirkabir University of
Technology (AUT), Tehran, Iran. His research
interests include islanding detection of DGs,
power electronic, applications of power
electronics in power systems, distributed
generation, power quality , control systems and artificial intelligence.
Farid Khazaeli received his B.Sc. degree in
Electrical Engineering from Amirkabir University
of Technology, Tehran, Iran. Now, he has
graduated with M.Sc. degree in Electrical
Engineering at Electrical Engineering department
of Sharif University Technology, Tehran, Iran. His
research interests include Modeling and control of
power system dynamics with particular interest in
control of grid connected renewable electrical
energy systems, electric drives and power
electronics.
Hashem Oraee (SM’98) received the First Class
Honours B.Eng. degree in electrical and electronic
engineering from University of Wales, Cardiff,
U.K., in 1980 and a Ph.D. degree in electrical
machines from University of Cambridge,
Cambridge, U.K., in 1984.He is currently a
Professor of Electrical Engineering at Sharif
University of Technology, Tehran, Iran. His
research interests include electrical energy
conversion and power quality. He is also active in
commercialization of brushless doubly fed induction
generators in wind generation industry