You are on page 1of 6

Test of anti-islanding protections according to IEC

62116: an experimental feasibility assessment



F. Belloni, P. Groppelli, C. Chiappa, R. Chiumeo, C. Gandolfi
Ricerca sul Sistema Energetico RSE s.p.a. via Rubattino 54, 20134, Milano, Italy
Email: belloni@rse-web.it, Piero.Groppelli@rse-web.it, Claudio.Chiappa@rse-web.it, Riccardo.Chiumeo@rse-web.it,
Chiara.Gandolfi@rse-web.it

Abstract - One of the main issues concerning the Inverter
based Distributed Generators (DGs) is the possibility that
inverters could feed parts of the public grid, even when the grid
is disconnected from the main power system. Such a condition,
that is called unwanted island, is potentially critical both for
safety and for consumers operations. In order to avoid
unwanted islands, its mandatory equipping the generating plant
with an Interface Protection (IP) which has to detect the
islanding condition and, in this case, to disconnect the generator
from the public grid. Several methods for identifying island
condition have been proposed, both passive and active, each one
characterized by its pros and cons. The standard IEC 62116 was
promulgated with the aim of regulating a test procedure to
evaluate the IP effectiveness of PhotoVoltaic (PV) inverters
independently from the island detection method implemented.
The paper discusses the test procedure proposed by the standard
IEC 62116 and presents some experimental activities developed
by RSE spa, aimed at the assessment of test feasibility and
effectiveness.
Index Terms - Anti-islanding protections, Distributed
Generators, Photovoltaic Inverters
I. INTRODUCTION
The growing installed power from Distributed Generators
(DG) is changing the traditional paradigm in which the
primary substation is the only power source of a distribution
grid. The presence of a significant number of power sources
along the entire distribution system gives rise to phenomena
and issues which do not occur in a more conventional passive
distribution scheme. In this scenario one of the main issues is
the possibility that inverters could feed parts of the public
grid, even when the grid is disconnected from the main power
system, due to a protection trip or to maintenance on the
utility side. Such a condition is called unwanted island and it
is potentially critical both for safety, since a part the
distribution grid remains energized even if disconnected from
the rest of the system, and for consumers operations, since
the island voltage level and frequency are not related with
those of the mains grid, but only related to the power balance
between DG and loads [1, 2].
In order to avoid unwanted islands, its mandatory
equipping each grid-connected generating unit with an
Interface Protection (IP) whose purpose is to detect the
occurrence of a loss of mains and, in this case, to disconnect
the generator from the public grid [3]. Several methods for
islanding detection have been proposed and developed. They
can be classified in three main categories:
passive methods [4], which are based only on the
monitor of the local measurements at the DGs Point of
Common Coupling PCC (such as voltage, frequency,
phase and/or their rate of variation); in this case the
island condition is detected when parameters run out of
the required thresholds;
active methods [2, 57], which intentionally introduce a
perturbance in the grid in order to amplify the quantities
deviation from nominal values;
communication-based methods: recent standards have
introduced the possibility of a forced trip of the IP, that is
driven by an external signal under the control of
distribution system operators [3].
The most traditional passive methods based protections,
also used in the Italian electric system, are based on voltage
and frequency measurements:
over-voltage;
under-voltage;
over-frequency;
under-frequency.
Even though passive methods are easier and less expensive
to implement with respect to the active ones, they show some
difficulties about the proper setting of the admissible
parameter ranges. Narrow ranges may lead to recurring IP
unwanted trips and unnecessary DG disconnections, while
wide ranges may fail the island detection. The main drawback
of this methods is the so called Non-Detection Zone (NDZ).
In other words, passive methods are unable to detect islanding
if the mismatch between the DG generated power and loads
consumption in the island is small [4, 6].
Such limitation is partially overcome by active island
detection methods, which introduce perturbations in the
current injected purposely by the inverter to unbalance the
power generation and absorption. Even though more effective
than passive methods, also active detection methods have
some limitations and drawbacks:
NDZ still exists for some specific load impedances;
unwanted trips can be caused by other disturbance
signals coming from the grid;
the perturbing signals injected into the grid may be
disturbing for other grid connected systems.
In order to assess the effectiveness of islanding prevention
measures, independently from their implementation and from
the adopted island detection method, the International
Electrotechnical Commission (IEC) proposed the IEC 62116
standard [8], where a test procedure for the IP of utility
interconnected photovoltaic generators has been indicated.
978-1-4799-3254-2/13/$31.00 IEEE
This work describes the RSE experimental and theoretical
activities for the feasibility assessment of the IEC 62116 test
procedures. The paper is organized as follows: Section II
describes the IEC 62116 test procedure and in Section III the
relation between power unbalance and island
voltage/frequency is deduced; Section IV reports the testing
rig and Section V show the activities performed on three
different PV inverters; finally, some remarks of the test
procedure feasibility and effectiveness are given in the
conclusions.

II. IEC 62116 TEST DESCRIPTION
The main goal of the standard IEC62116 is to provide a test
procedure to evaluate the reliability of islanding prevention
measure that is implemented by the inverters used to connect
PV plant to the grid.
The experimental arrangement for the IEC 62116 tests is
shown in Figure 1 [8].


Figure 1: Experimental arrangement for the IEC 62116 test procedure.

The test procedure can be applied both to single phase and
three phase inverters. In particular this standard defines that a
suited behavioural model of a portion of a distribution grid
during island operation can be obtained neglecting line
transversal parameters, and considering the parasitic
capacitances to ground and the loads as lumped parameters
element, that are equivalent to a parallel Resistive-Inductive-
Capacitive (RLC) impedance. The prescribed test consists in
connecting a PV inverter in parallel to a power grid (or a
power grid simulator) and to a local adjustable RLC load with
given resonance characteristics. The nominal load conditions
are:
inv load
P
R
V
P = =
2
0
(1)
C L
Q Q = (2)
( ) 50 2 1
0
= = LC f Hz (3)
1 / = = L C R Q
f
(4)
Where V
0
is the nominal grid voltage, Q
L
is the inductive
reactive power, Q
C
the capacitive one, f
0
is the nominal
resonance frequency and Q
f
is the quality factor.
When the system is at the steady state, the test begins by the
opening of the switch S1 (Figure 1). The test consists in
measuring the time, the so called run-up time, for which the
Equipment Under Test (EUT) continues to supply the load
before the IP operation. The test must be repeated for
different active and reactive power conditions:
the balance condition (resonance), corresponding to the
(1) (4), between generation and load;
the controlled unbalance conditions, in accordance with
the set of (P, Q) percentage unbalances that are
reported in TABLE I
1
.

TABLE I
PERCENT POWER UNBALANCE CONDITIONS FOR IEC 62116 TESTS.
-10, +10 -5, +10 0, +10 +5, +10 +10, +10
-10, +5 -5, +5 0, +5 +5, +5 +10, +5
-10, 0 -5, 0 0, 0 +5, 0 +10, 0
-10, -5 -5, -5 0, -5 +5, -5 +10, -5
-10, -10 -5, -10 0, -10 +5, -10 +10, -10

Its worthwhile to note that the central line is related to
variations of only active power, while the central vertical
column is the related to variations of only reactive power.
The IP passes the test if both the following results for the
recorded run-up time are satisfied:
it has to be less than 2 s in each transient;
in each unbalanced condition of the white area of
TABLE I, it has to be less than that in the balanced one
(cell 0,0); if not, the tests must be extended to the grey-
shaded area.

III. RELATIONS BETWEEN POWER UNBALANCE AND ISLAND
VOLTAGE AND FREQUENCY
When the load is supplied by both the inverter and the grid
(switch S1 and S2 of Figure 1 are closed), voltage and
frequency are kept almost constant at their nominal values, V
0

and f
0
. In this case the grid supplies an amount of active and
reactive power equal to the difference between the active
powers that are supplied by the inverter and absorbed by the
load [9].
During the island condition (switch S1 open), voltage and
frequency must change values to V
1
and f
1
in order to keep the
power balance between load and generation. If the inverter
supplies power at unitary power factor the follow general
equations can be considered:
R R
V
P
inv
+
=
2
1
(5)
( )( ) ( ) C C L L f + + = 2 1
1
(6)
where R+R, L+L and C+C are the actual values of the
load representative of the test power unbalance (TABLE I).
Combining (3) and (6) and defining f
min
and f
max
the trip
values of under-frequency relay and over-frequency relay
respectively, its possible to obtain the following relations:



1
In particular, the active power unbalances are generated modifying the
resistive load, while reactive power unbalances are induced acting either on
the inductive component of the load or on the capacitive one.
1
) )( (
0
0 1

+ +
=

C C L L
LC
f
f f
(7)
0
0 max
0
0 min
1
) )( ( f
f f
C C L L
LC
f
f f

+ +

(8)

Considering only small variations of the parameters
(LC~0), the relation (8) can be written as:
1 1
2
min
0
2
max
0

f
f
L
L
C
C
f
f
(9)
Assuming Q the unbalance between the reactive power
inverter generated and the load absorbed one, after the loss of
mains, the following relation comes from (4), (5), (6) :

L
L
C
C
Q
P
Q
f
(10)
Combining (9) and (10) a relation between the NDZ of the
frequency protections and the reactive power unbalance can
be obtained:

2
min
0
2
max
0
1 1
f
f
Q
P
Q
f
f
Q
f f
(11)
A similar relation linking the NDZ of voltage protections
and active power unbalance, P, can be obtained from (5):
1 1
2
min
0
2
max
0

V
V
P
P
V
V
(12)
where V
min
and V
max
are the lower and upper limits of the
NDZ of the voltage relay. A typical NDZ for a passive island
detection method based on the measurements of voltage and
frequency is reported in Figure 2 [7]. If the active/reactive
power unbalance between generation and load are within the
NDZ, the loss of mains is not detected by the IP and the
island remains supplied by the inverter.


Figure 2: Non Detecting Zone for an Interface Protection with passive island
recognition.

According to relation (11) frequency variations are mainly
related to reactive power unbalances, while active power
unbalances are responsible for voltage variations (12).
IV. THE EXPERIMENTAL ARRANGEMENT
The system represented in Figure 1 has been implemented
at the RSE laboratory in Piacenza (Italy). It consists of:
a DC source with PV array VI characteristic emulation
feature (P=20 kW, V
max
=500 V);
a single/three phase connection to the public LV grid
(alternatively, a 12 kVA grid simulator is available);
a data acquisition system which monitors AC and DC
measurements and allows on-line calculations of active,
reactive power and root-mean-square (RMS) values; the
Supervision system offers some visual tool to easily
check the power balance between generator and load; an
example of the Man to Machine visual Interface (MMI)
of the data acquisition system is shown in Figure 3;
an adjustable parallel RLC load (single or three phase)
with overall power capability up to 18 kW (resistive) and
18 kVAr (inductive and capacitive).

A power absorption from 2 kW/2 kVAr to 18 kW/18
kVAr can be set through series/parallel connections of load
elements, and small power variations (up to about 5 % of the
nominal value) can be produced for every load component. In
implementing each load component, particular care was given
in minimizing the parasitic effects. Some details of the
resistive load actual implementation are shown in Figure 4.
Manually operated switches and differential relays, for safety
purposes, complete the experimental set-up. An overview of
the laboratory environment is shown in Figure 5.

Figure 3: Topological diagram of the experimental set-up (taken from the
data acquisition system visual interface).

Figure 4: Resistive component of the load. The wirewound resistor is
implemented so that inductive parasitic effects are minimized.
V. EXPERIMENTAL RESULTS
The IP of three different Low Voltage (LV) single and
three phase PV inverters, rated 3 kW, 6 kW and 10 kW
respectively, are tested according to the IEC 62116 standard.
All of them adopt the passive island recognition criterion
based on measurements of frequency and voltage with the
same setting, reported in TABLE II.

TABLE II
INTERFACE PROTECTIONS SETTINGS
Protection relay Threshold value Intentional Delay
Inv I Inv II e III
Over-voltage 110% VN 50 ms 200 ms
Under-Voltage 90% VN 50 ms 100 ms
Over-frequency 50.3 Hz 50 ms 100 ms
Under-Frequency 49.7 Hz 50 ms 100 ms

In the follow the experimental results are presented. The
nominal values of the LV grid are V
0
=230V and f
0
=50Hz.

A. Single phase 3 kW PV inverter
The actual balance condition corresponds to:
PV array power: P
N
=2.8 kW;
AC side inverter active power: 2.7 kW;
load active power: 2650 W;
capacitive/inductive reactive power: 2.7 kVAr.

With reference to the Figure 6, the loss of mains occurs at
t=2 s and is fully corresponding to the zeroing of the from-
inverter-to-grid current (white curve).
According to the previous calculated load values, the island
voltage and frequency would be V
1
=232.16 V and f
1
=50.02
Hz well, within the NDZ of the IP. Indeed, the load remains
supplied by the inverter for some minutes after the loss of
mains, as shown in Figure 6, where the island voltage
waveform is shown in green, the inverter currents are
reported in blue and the grid current in white.


Figure 6: Voltage and current waveforms recorded by the data acquisition
system during a test in balanced power conditions for a 3 kW inverter.

The periodic voltage drop of the green waveform of Figure
6 is due to the Maximum Power Point Tracking (MPPT)
algorithm implemented within the inverter control. The
voltage remains above 90% V
N
even during these phenomena
and no under-voltage protection trip occurs. The IP fails the
test also for some unbalanced conditions with small excess of
capacitive reactive power.

B. Single phase 6 kW PV inverter
In this case, the actual balance condition at the rated
inverter power is:
PV array power: P
N
=5.6 kW;
AC side inverter active power: 5.5 kW;
load active power: 5.5 kW,
capacitive/inductive reactive power: 5.5 kVAr.

Even though the balance condition is expected to give rise
to an island, according to theoretical calculations (voltage and
frequency within the NDZ of the IP), an island condition for
more than few hundreds milliseconds was never recorded. An
example of recorded test result is shown in Figure 7. In this
case the IP trips in 70 ms. Many comparable results were
obtained for unbalanced conditions and for reduced PV array
power (66% P
N
and 33% P
N
).


Figure 7:Voltage and current waveforms recorded by the data acquisition
system during a test in balance power conditions for a 6 kW inverter.

To explain the experimental results, a simulation model of
the experimental set-up is developed in the ATPDraw
environment. In general the modelling activities are mainly
focused on:
confirming relations among power unbalances and island
voltage and frequency;
forecasting the behavior of the inverter, grid and IP in
different conditions;
explaining experimental results (even though the
simulation model represents a three phase inverter and
the actual inverter is single phase, the overall behavior
remains the same).

The grid frequency simulated is reported in Figure 8 for the
power balanced case (i.e. equations (1)(4) are satisfied). The
loss of mains occurs at t=1.0 s, (the frequency deviation
visible at t=1.0 s is due to some transient voltage perturbation
occurring at the loss of mains). In Figure 9 the current (red
curve) and voltage (green curve) simulated in case of power
balance during island are presented.

Figure 8: Island frequency simulated with balance between generation and
load.

Figure 9:Voltage and current waveforms simulated.

In the model, a perturb and observe MPPT algorithm is
embedded within the PV inverter control, in order to take into
account the power variations introduced by the MPPT itself.
This algorithm is set to have an active power variation 300
ms after the loss of mains. Simulation results are presented in
Figure 10 for a power variation of -5% P
N
and in Figure 11
for -10% P
N
. In both cases, the loss of mains occurs at t=1.0 s
and the MPPT reduces the active power at 1.3 s, but only in
case of a 10% active power reduction the islanding condition
has been detected.

0,8 1,0 1,2 1,4 1,6 1,8 [s]
-400
-300
-200
-100
0
100
200
300
400
[V]

Figure 10: Voltage drop due to a -5% PN inverter power variation.
0,8 1,0 1,2 1,4 1,6 1,8 [s]
-400
-300
-200
-100
0
100
200
300
400
[V]

Figure 11: Voltage drop due to a -10% PN inverter power variation.

Then, a reasonable explanation of the unexpected IP
operation in balance condition (Figure 7) may be found in the
active power variations introduced by the MPPT algorithm.
This can cause enough large voltage drops to be detected by
the under-voltage relay.

C. Three phase 10 kW inverter
The tested three phase inverter offers some advanced
features such as the Fault Ride Through (FRT) and the
control of reactive power injection/absorption for grid voltage
support. The inverter is tested in each condition, according to
the TABLE I, also at reduced PV array power (66% P
N
and
33% P
N
). In every test case, the run-up time is less than 200
ms; the Figure 12 shows the voltage and currents measured in
the balance condition.

1.8 1.85 1.9 1.95 2 2.05 2.1 2.15 2.2 2.25 2.3
-400
-200
0
200
400
V
o
lta
g
e
[
V
]
1.8 1.85 1.9 1.95 2 2.05 2.1 2.15 2.2 2.25 2.3
-20
-10
0
10
20
C
u
rr
e
n
t [
A
]
1.8 1.85 1.9 1.95 2 2.05 2.1 2.15 2.2 2.25 2.3
-2
-1
0
1
2
Time [s]
C
u
r
r
e
n
t [
A
]

Figure 12: Experimental results for the test of island detection for a 10 kW
inverter. Voltages are shown in the upper graph, inverter currents in the
central one and from-inverter-to-grid currents in the lower one. Loss of mains
occurs when the grid currents drop to zero.

VI. CONCLUSIONS
The main goal of the standard IEC 62116 is to provide a
test procedure to evaluate the reliability of islanding
prevention measure that must be implemented by the grid
connected PV Inverters.
The paper discusses the test procedure proposed by the
standard IEC 62116 and shows the experimental activities
oriented to its feasibility assessment made by RSE in 2012.
The experimental results allow to claim the main IEC 62116
merit: the focus on the specific balance condition (Q
L
=Q
C
,
P
LOAD
=P
INV
) as the most critical condition for island
detection. Besides, thanks to the actual RSE implementation,
it is possible to evaluate the most critical parts related to the
test laboratory developments that are:
the implementation of the adjustable RLC load, which
must have low parasitic parameters and high thermal
stability and whose power should be adjustable ;
the difficulty to achieve a stable steady state with the
desired match between power Inverter generation and
load absorption before the loss of mains.
At the same time, the reading of the experimental results
points out some issues concerning this standard:
possible interactions among more inverters, are not
considered;
the RLC load is not representative of all the possible grid
loads, such as nonlinear or distorting loads.
Finally, the experimental activities, made on three PV
inverters (single and three phase), show the possibility of
supporting island condition, also in presence of unbalanced
power exchanges between generation and loads. Maximum
Power Point Tracking algorithm plays an important role in
creating power unbalances which is useful for the island
detection. The phenomenon has been further investigated and
confirmed also by simulations.
ACKNOWLEDGEMENTS
This work has been financed by the Research Fund for the
Italian Electrical System under the Contract Agreement
between RSE S.p.A. and the Ministry of Economic
Development - General Directorate for Nuclear Energy,
Renewable Energy and Energy Efficiency in compliance with
the Decree of March 8, 2006.

REFERENCES
[1] S. P. Chowdhurya, S. Chowdhurya, P.A. Crossleyb, Islanding
protection of active distribution networks with renewable distributed
generators: A comprehensive survey, Electric Power System Research
Journal, Volume 79, Isuue 6, 2009, Page(s): 984 992.
[2] D. Menniti, A. Pinnarelli, N. Sorrentino, G. Brusco, Islanding
detection active methods: implementation of simulation models using
ATP-EMTP, 14th International Conference on Harmonics and Quality
of Power (ICHQP), Bergamo, Italy, 26-29 Sept. 2010, pp. 1-8.
[3] CEI 0-21 Regola tecnica di riferimento per la connessione di Utenti
attivi e passivi alle reti BT delle imprese distributrici di energia
elettrica, Italian Standard, download: www.ceiweb.it, in Italian
lenguage.
[4] Zhihong Ye; Kolwalkar, A.; Yu Zhang; Pengwei Du; Reigh Walling;
"Evaluation of anti-islanding schemes based on nondetection zone
concept", IEEE Transactions on Power Electronics, Volume: 19, Issue:
5, 2004, Page(s): 1171 1176.
[5] T. Thacker, F. Wang, R. Burgos, D. Boroyevich, Islanding detection
using a coordinate transformation based phase-locked loop, Power
Electronics Specialists Conference 2007. PESC 2007, 17-21 June 2007,
Orlando (USA), Page(s): 1-6.
[6] M.G.M. Abdolrasol, S. Mekhilef, Robust hybrid anti-islanding method
for inverter based distributed generation, IEEE Region 10 Conference,
TENCON 2010, 21-24 Nov. 2010, Fukuoka (Japan), Page(s): 13-18.
[7] X. Zhu, C. Du, G. Shen, M. Chen, D. Xu, Analysis of the Non-
detection Zone with Passive Islanding Detection Methods for Current
Control DG System, Twenty-Fourth Annual IEEE Applied Power
Electronics Conference and Exposition 2009. APEC 2009, 15-19 Feb.
2009, Washington DC (USA), Page(s): 358 - 363.
[8] IEC 62116 Test procedure of islanding prevention measure for utility -
interconnected photovoltaic inverters, Ed.1.0, 09 2008.
[9] Zengqiang Mi, Fei Wang, "Power Equations and Non-Detection Zone
of Passive Islanding Detection and Protection Method for Grid
Connected Photovoltaic Generation System", Pacific-Asia Conference
on Circuits, Communications and Systems, 2009. PACCS '09, 2009,
Page(s): 360 363.



Figure 5: Overview of the test laboratory.

You might also like