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.2014.2369479, IEEE Transactions on Power Delivery

Distance Protection of Lines Emanating from Full-Scale Converter-

Interfaced Renewable Energy Power Plants―Part I: Problem Statement
Ali Hooshyar, Maher A. Azzouz, and Ehab F. El-Saadany, Senior Member, IEEE
Abstract—The fault ride-through (FRT) requirement of modern converter similarities, this study is equally applicable to the
grid codes results in interactions between full-scale converter- lines emanating from voltage-source-converter-based HV dc
interfaced renewable energy power plants (CIREPPs) and the (HVDC) links, another rapidly expanding technology [9].
protection systems of high voltage (HV) transmission grids, As power plants have been traditionally composed of syn-
which normally involve distance elements either as the primary
chronous generators (SGs), modelled by Thevenin equivalent
or the back-up relay. Such interactions are influenced by the
CIREPPs’ exclusive fault behavior, and have been left largely circuits for fault studies [10], the conventional wisdom from
unnoticed in relaying literature. Part I of this paper develops a the protective relaying perspective has been that CIREPPs can
CIREPP model suitable for relaying studies, and highlights the also be represented similarly, with the only consideration be-
CIREPP properties by which the protection system is endangered ing a large source impedance to take account of the CIREPPs’
the most. Then, the operating scenarios leading to malfunction of modest fault current contribution. However, the CIREPP fault
a distance relay that is located at a CIREPP substation and pro- behavior is governed by the control system that determines the
tects the adjacent line are unveiled. These scenarios include in- switching pulses of the grid-side converter [11]. The objec-
zone short-circuits missed by the relay, and incorrect tripping for tives and settings of this control system can be quite different
out-of-zone faults, which would, in turn, neutralize FRT schemes
depending on the host system GC, the internal references, and
implemented inside CIREPPs. The findings of this study also
hold true for the ac lines emanating from voltage source convert- the adopted reference frame, among other factors. Thus, mod-
er-based HV dc connections. elling a CIREPP with all of its complications by a simple
source is not an accurate approach, especially during faults.
Index Terms—Converter-interfaced renewable energy power On this basis, section II of the paper proceeds with a brief
plant, line distance relay, photovoltaic farm, type IV wind farm. overview of the GC articles that will later be found to influ-
ence the CIREPP-relay interactions. Section III describes the
I. INTRODUCTION system under study. The analysis of the distance relay’s possi-
HE recent progress in renewable energy (RE) technolo- ble maloperation for zone one balanced faults is covered in
T gies has led to the erection of RE power plants (REPPs)
up to the order of several hundred megawatts. Unlike their
section IV, and is extended to unbalanced faults in section V.
Section VI studies relay failures during zone two faults. Sec-
predecessors, which appeared mainly in the form of dispersed tion VII shows the relay operation for a conventional source.
generation (DG) coupled with distribution systems, such large Concluding remarks are presented in section VIII.
REPPs are naturally part of high-voltage (HV) transmission
networks and hold non-negligible proportions of the genera- II. GC ARTICLES FOR FAULT CONDITIONS
tion. As a result, the fault ride-through (FRT) requirement has Besides FRT capability, modern GCs determine active and
become an essential part of modern grid codes (GCs) [1]. reactive power requirements that REPPs have to comply with.
As a result of the FRT schemes, the REPPs’ impact on pro- Apart from the power requirements that are to be met over
tective relaying of the systems with which they are integrated normal operation, European GCs (EU-GCs) impose regula-
can become challenging. In the case of RE-based DGs coupled tions specific for faulty conditions, and give higher priority to
with distribution systems, where overcurrent protection based reactive power support from REPPs [12]. Take, for instance,
on the assumption of a single path for the fault current is pre- the Spanish grid; within 150 ms after the beginning of a volt-
dominant, relay coordination might be affected by the addi- age dip at the point of interconnection (POI), the reactive cur-
tional current paths introduced by the DGs, or the protection rent generated by a WF must be situated inside the shaded area
system selectivity may be risked because of bi-directional of Fig. 1 [13], [14]. In Fig. 1, V + and I + are the positive se-
fault currents. Various publications on distribution system re- +
quence voltage and current, Vn is the nominal voltage, and IQ
laying offer effective remedies for these problems [2], [3].
is the positive sequence reactive current, given by
Overcurrent elements do not play a primary role in trans-
mission networks, where the backbone of the protection sys-  
IQ  I  Itot sin(  ) (1)
tem is formed by distance and, to some extent, line differential + + +
in which φ is the angle between V and I , and Itot is the total
relays. Therefore, the above publications on the protection of fault current. The lower the voltage level, the larger the reac-
systems with DG are of limited use for transmission grids. tive current generated by the WF must be. Meanwhile, during
New zone settings are proposed in [4] and [5] for a distance the 150 ms period, the WF is allowed to consume both active
relay that protects the line connected to a WF. These papers and reactive power at the POI as long as they are limited to
model the WF by a voltage source. The failure of a distance 10% and 60% of the rated power, respectively. A broadly sim-
relay in the case of type III WFs is reported in [6] and [7]. The ilar requirement is imposed by the German GC [15]. The GCs
reason and the solution for this problem are presented in [8]. that comply with the reactive current requirements of [12] are
This paper reveals the problems associated with distance referred to as EU-GCs in this paper.
protection of lines connected to full-scale voltage source con- On the issue of disturbance behavior, the current North
verter-interfaced REPPs (CIREPPs). CIREPPs mainly consist American GCs (NA-GCs) focus only on how REPPs must ride
of photovoltaic farms (PVFs) and type IV wind farms (WFs), through the fault, and do not enforce any specific regulation
which are both growing globally at exponential rates. Due to on the active or reactive components of REPP fault currents

0885-8977 (c) 2013 IEEE. Personal use is permitted, but republication/redistribution requires IEEE permission. See
http://www.ieee.org/publications_standards/publications/rights/index.html for more information.
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.2014.2369479, IEEE Transactions on Power Delivery

IQ+ Fault condition Normal operation Zs1 1 Line 15

5
Line 35 3 Zs2
1.0
0.9
Q Source 1 Source 3
generation DS25 DS52 Line 45 Zs4
Line 25
1.0 V +/Vn Reactor 4 Source 4
2 CIREPP
0 0.5 0.85 Q π model Inverter
consumption
LC
Fig. 1. WF reactive current requirement by the Spanish GC. LPF
Rch Is
[16], [17]. Thus, the control system of an REPP would natural-
ly attempt to satisfy the NA-GCs’ normal demand, which is Fig. 2. Single line diagram of the test system, including the CIREPP.
0.95 plus positive sequence power factor (PF), lead or lag. model representing the cable system is able to simulate a WF
HVDC links normally operate at high PF. To the best of our accurately [19], [20]. This process, designated as equivalenc-
knowledge, there is currently no GC in place that mandates re- ing, can be similarly carried out for PVFs as well. The equiva-
active power generation during faults for HVDC substations. lencing practice was used to model the CIREPP in this study.
Only one very recent draft of a GC for HVDC connections, The main transformer of large REPPs is commonly three-
prepared for public consultation by the European Network of winding, regularly configured as ygdYG or ddYG. On the grid
Transmission System Operators for Electricity (ENTSO‐E), side, both configurations result in the same behavior. For this
allows regional utilities to have specific reactive power de- study, the ddYG configuration is chosen. The transformer rat-
mands over fault conditions [18]. This draft does not impose ings are 150 MVA, 34.5/34.5/230 kV and X=0.1 pu.
any obligation in this regard, and has not gained final approval
C. CIREPP’s Control System and its GC Compliance
yet. Therefore, similar to REPPs conforming to the NA-GCs,
HVDC links should be expected to operate at high PFs after a The findings of this study are independent of the reference
fault, especially during the immediate narrow time frames frame in which the CIREPP’s control system is implemented.
when transmission system relays have to take action. The dq frame with proportional integral (PI) controllers, volt-
age feed-forward compensation and cross-coupling terms, de-
III. DESCRIPTION OF TEST SYSTEM AND CIREPP tailed in [21], is chosen, as it is widely deployed in the RE in-
dustry [22]. The phase-locked loop (PLL) of the control sys-
A. Test System tem is immunized against unbalanced conditions using the
Simulated using PSCAD software, the 230 kV, 60 Hz sys- method described in [23]. A braking chopper circuit is used to
tem depicted in Fig. 2 was used to study the protection of line suppress the dc link voltage during faults [24].
25 connected to the 100 MW CIREPP on bus 2. DS25 and The reference current generation system of the CIREPP is
DS52 are the distance relays located at buses 2 and 5, respec- similar to the ABB generic type IV wind turbine model, which
tively, and protect line 25. The positive- and zero-sequence can comply with the EU-GCs [25]. The d and q components of
impedances along with the angle of each source in the PSCAD the reference currents, denoted by id-ref and iq-ref, are passed
source model are through a saturation block to protect the switches by limiting
+ o
Source 1: ∡Vs1=-7.1°, Zs1 =12.66∡83° Ω, Zs1=11.1∡83° Ω, the short-circuit currents to 1.2 times the rated value. Over
+ + normal operation, the primary objective is to fix the dc link
Source 3: ∡Vs3=-2.0 , Zs3 =7.76∡86 Ω, Zs3 =5.0∡86° Ω,
° °
+ +
Source 4: ∡Vs4=-6.9°, Zs4 =12.7∡82° Ω, Zs4 =11.2∡82° Ω. voltage by controlling the active power. Once a voltage drop is
The reactor at bus 5 is rated 75 MVAR. When the CIREPP detected, priority is shifted to reactive current for EU-GCs.
generates its rated capacity at unity PF, the voltages at buses 2 Being determined for the positive sequence circuit, the reac-
and 5 are 230.6∡3.0° and 228.2∡-2.4° kV, respectively. The tive current requirements of EU-GCs are not phase selective,
positive- and zero-sequence impedances of the lines are and were found to potentially pose the risk of temporary over-
+ o voltages in healthy phases during unbalanced faults. There-
ZL =0.0357+j0.5077 Ω/km and ZL =0.3630+j1.3262 Ω/km. The
fore, besides the GC-generated look-up tables that provide re-
lengths of lines 15, 25, 35 and 45 are 155, 100, 110 and 120
active current versus voltage drop curves, iq-ref is also moni-
km, respectively.
tored to maintain each phase voltage within acceptable limits.
B. CIREPP Structure
The time spans associated with wind speed and solar inten- IV. ZONE ONE BALANCED FAULTS
sity variations are appreciably longer than the short fault inter- This section inspects the performance of DS25 for balanced
vals. In addition, the fast dynamics of the renewable source faults in zone one, which covers 90% of line 25. Several case
are essentially decoupled from the grid by the dc capacitor. studies will be presented, through which the problematic and
Hence, the renewable source and the converter that provides non-problematic conditions from the line protection perspec-
the dc voltage are represented by a controllable current source. tive and the underlying causes for relay maloperation will be
A large WF is composed of dozens of turbines connected to identified. The case studies, which are conceptually similar to
the farm substation through underground cable. Each unit in- those of [26], take both the EU-GCs and NA-GCs into ac-
cludes a back-to-back converter, a low-pass filter (LPF) and a count, and include different properties of fault, such as its re-
transformer that boosts the voltage up to the medium level. sistance and location within DS25’s zone one. The problem is
Modelling each turbine unit separately and the cable system described using a self-polarized impedance-based relay with a
with its fine details is computationally intensive. Furthermore, quadrilateral characteristic and a large 100 Ω resistive reach.
it has been shown that a single aggregated unit along with a π Furthermore, to validate the existence of the discussed prob-

0885-8977 (c) 2013 IEEE. Personal use is permitted, but republication/redistribution requires IEEE permission. See
http://www.ieee.org/publications_standards/publications/rights/index.html for more information.
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.2014.2369479, IEEE Transactions on Power Delivery

Current (pu) 1.25 1.25

Current (pu)
1 I+ I+ 1 I+ I+
0.75 P Q 0.75 P Q
0.5 0.5
0.25 0.25
0 0
1990 2000 2010 2020 2030 2040 2050 2060 2070 2080 2090 1990 2000 2010 2020 2030 2040 2050 2060 2070 2080 2090
Time (ms) Time (ms)
Fig. 3. The active and reactive current of the CIREPP for EU-GC compliance Fig. 5. The active and reactive current of the CIREPP for EU-GC compliance
in case 1. in case 2.
600

Impedance ()
80 AG element BC element R X
400 Time = 2099.5
60
Zone 2 X = 124.6653
X ()

200
40
Zone 1 0
20
 Line replica impedance Time = 2099.5
0 1990 2000 2010 2020 2030 2040 2050 2060 2070 2080 2090
R = 0.3259
Time (ms)
-20 0 20 40 60 80 100 120 (a)
R ()
150 AG element BC element
Fig. 4. The impedance calculated by DS25 for case 1.

X ()
100
lems in practical applications, the response of a memory-
50
polarized mho characteristic commonly deployed in commer-
cial relays is also scrutinized. 0
0 20 40 60 80 100 120
The voltages and currents are passed through a second-order R ()
Butterworth anti-aliasing filter with a 1 kHz cut-off frequency. (b)
The voltage is measured by full-cycle discrete Fourier trans- Fig. 6. The impedance calculated by DS25 for case 2, (a) Impedance of the
form (FCDFT), while the current phasor is computed by AG element versus time, (b) Impedance plane of the AG and BC elements.
FCDFT for GC compliance purposes, and by the one-cycle es to 0.29 pu compared to less than 0.1 pu in case 1, whereas
least error squares based digital filters discussed in [27], which the fault current equals the CIREPP’s maximum current limit
express the fault current’s decaying DC offset by its Taylor se- again. Since the voltage drop is above 50%, the control system
ries expansion, for relaying purposes. At the sampling rate of attempts to increase the reactive current up to the full capacity.
64 times per fundamental cycle, the impedance was updated In contrast to Fig. 3, where the reactive current variation was
for every 4 new pairs of voltage and current samples. virtually instantaneous, because it was determined by the line
+
impedance, the IQ shown in Fig. 5 undergoes a transient peri-
A. Case 1
od influenced by the CIREPP’s control system.
Case 1 includes a bolted balanced fault at 60% of line 25. When the initial transients disappear, the resistance and re-
The fault occurs at t=2 s, and the greater than 90% voltage actance seen by the AG element of DS25, plotted with respect
drop is accompanied by the converter maximum current, nei- to time in Fig. 6(a), are fixed at 0.33 Ω and 124.67 Ω, respec-
ther displayed because of space limitations. Before the fault, tively. The latter is in excess of the actual reactance to the
the CIREPP generates the nominal power at unity PF. During fault by more than fourfold. Fig. 6(b) illustrates the relay’s
the fault, the CIREPP attempts to meet the EU-GCs. The maloperation in the first 100 ms after the fault inception, in-
CIREPP GC compliance is illustrated in Fig. 3, which shows cluding the initial 20 ms, during which DS25 must take action.
the significant quick jump of the per-unit reactive current after As shown, the operating point lies far above zone one, making
+
the fault is initiated. In this figure, IP is the positive sequence an impedance-based DS25 fail to trip the line. Unlike case 1,
active current. The CIREPP’s rated peak current is used to the CIREPP does have full control over its current, as the volt-
per-unitize the quantities displayed in the paper. The imped- age at the fault location is not zero and the CIREPP is not
ances calculated by the phase A to ground (AG) and phase B completely isolated from the rest of the grid.
to phase C (BC) elements of DS25, depicted in Fig. 4, enter Contrary to conventional power plants whose voltages over
zone one in 14.5 ms and 13 ms, respectively. fault conditions are regulated by the SGs’ excitation system
Similar results are obtained for the NA-GC. In that case, the and are thus modeled by Thevenin equivalent circuits for fault
control system does not meet its objective, and the PF drops to studies [10], voltage-source converters in RE and HVDC ap-
about 0.09. The zero fault resistance, Rf, makes the voltage at plications are controlled through their currents, due to numer-
the fault location zero and causes the CIREPP to be connected ous advantages such as superior dynamic performance and
to an effectively dead circuit. Consequently, the difference be- limited overcurrents [28]. Consequently, a CIREPP behaves as
tween the voltage and current angles during the fault is dictat- a controlled current source that follows certain active and re-
ed by the mainly inductive line impedance, over which the active power commands given by the host system GC and/or
CIREPP has no control. In fact, GC compliance for all zone dc link voltage control loop. The generated P and Q are trans-
one bolted balanced faults is determined not by the CIREPP’s lated into certain resistance and reactance in the impedance
control system, but by the physics of line 25 impedance. plane, respectively. Take, for instance, the phase A active and
B. Case 2 reactive power curves of the CIREPP for case 2, displayed in
Fig. 7, which exhibit high correlation with the R and X curves
Unlike the above case, actual faults usually include some of Fig. 6(a). Since the control system of the CIREPP makes
level of fault resistance. Case 1 with the EU-GC is repeated, the current mainly reactive after the transients fade out in Fig.
while Rf is increased to 10 Ω. The fault voltage at the POI ris- 5, P and Q are fixed at about 0.03 MW and 11.54 MVAr, re-

0885-8977 (c) 2013 IEEE. Personal use is permitted, but republication/redistribution requires IEEE permission. See
http://www.ieee.org/publications_standards/publications/rights/index.html for more information.
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.2014.2369479, IEEE Transactions on Power Delivery

Power (MW / MVAr) 90 1

Active power Reactive power
30

 (degree)
0.98
Time = 2099.5 45
20 0.96
Q = 11.5415
 + PF

PF
10 0 0.94

+
0.92
0 -45
0.9
Time =2090
1990 2000 2010 2020 2030 2040 2050 2060 2070 2080 2099.5 1990 2000 2010 2020 2030 2040 2050 2060 2070 2080 2090
Time (ms) P = 0.030172 Time (ms)
Fig. 7. Phase A power for case 2. (a)

Power (MW / MVAr)

Time = 2099.5 Active power Reactive power
360 30
 2 = 275.0458
 (degree)

Time = 2099.5
270 20 P = 8.1645
Time = 2099.5 10
180
 1 = 273.5847 0
90 Trip zone  1
 2
Time = 2099.5
0 1990 2000 2010 2020 2030 2040 2050 2060 2070 2080 2090
Q = -1.7111
1990 2000 2010 2020 2030 2040 2050 2060 2070 2080 2090 Time (ms)
Time (ms) (b)
Fig. 8. Operation of memory-polarized mho AG element of DS25 for case 2. Fig. 9. Measurements at the CIREPP POI for case 3, (a) φ+ and the PF for the
NA-GC compliance, (b) Phase A power.
spectively, which correspond to R=0.33 and X=127.72 Ω, con-
50 AG element
sidering the 1.2 pu current. The tiny difference between these
0 BC element

X ()
values and the resistance and reactance pinpointed in Fig. 6(a)
-50
arise from the fact that the 1.2 pu limit is for the converter cur-
-100
rent, while the relay is at the POI.
For the same fault conditions, if the CIREPP’s current limit 0 20 40 60 80 100 120 140 160
R ()
was set to 1.1 pu, Q would decrease to 10.34 MVAr, resulting (a)
in X=135.66 Ω; i.e., the reactance jumps by about 8 Ω because Time = 2099.5
360  1 = 275.1753
of a minor change, not in the fault properties, but in the  (degree)
270
CIREPP’s control system. Similarly, increasing the CIREPP’s Time = 2099.5
current limit reduces the impedance seen by the relay. In other 180
 2 = 268.8531
words, the impedance calculated for this case should not be 90 Trip zone 1 2
expected to lie along the transmission line replica impedance 0
1990 2000 2010 2020 2030 2040 2050 2060 2070 2080 2090
or inside the correct zone of operation, as it is highly affected Time (ms)
by the CIREPP’s control system. (b)
Commercial distance relays utilize alternatives to impedance Fig. 10. Operation of DS25 for case 3, (a) Impedance of the AG and BC ele-
ments, (b) Angle of the memory-polarized mho AG element.
calculation to find the zone in which the fault has occurred.
For instance, several relays, such as [29] and [30], compare As a result, the CIREPP power for phase A in Fig. 9(b) is
the angle α between a polarizing and an operating quantity, mainly active, and the reactive component takes a small nega-
denoted by Vpol and Vop, respectively, to form a mho character- tive value. These power curves are directly reflected in the re-
istic, i.e, α=∡Vop-∡Vpol. Vpol is normally the memorized posi- sistance and reactance of the AG and BC elements of DS25,
tive sequence voltage. Vop is given by (2) for the AG element: depicted in Fig. 10(a). A large positive resistance is followed
Vop  AG  VA  Zr I A  K o I o   (2)
by a small negative reactance, making the operating point fall
in the fourth quadrant outside zone one, putting the dependa-
where VA and IA are phase A voltage and current, Zr is the bility of DS25 in jeopardy. Comparison of totally different
o impedances for the same fault condition in Figs. 6(b) and
reach impedance of the respective zone, K is the zero se-
o
quence compensation factor, and I is the zero sequence cur- 10(a) reveals the high dependence of DS25 on the CIREPP’s
rent. The trip signal of the relay’s n-th zone is asserted if the control system. Memory-polarized mho elements do not func-
corresponding angle, denoted by αn, is inside the [90°,270°] in- tion properly either. α1AG in Fig. 10(b) is about 5° above the
terval. α1 and α2 of the AG element of DS25 for case 2, de- tripping zone. Only α2AG is on the boundary of the [90°,270°]
picted in Fig. 8, approach the tripping zone during the transi- interval, which leads to unnecessary delayed tripping.
ent response time of the CIREPP’s control system, but they
maintain 3.5° and 5° distance from the trip boundaries after the V. ZONE ONE UNBALANCED FAULTS
transients disappear. Similar incorrect operation is observed Discussion of the relay maloperation for balanced faults in
for other mho elements, such as that of [31]. the last section was from the viewpoint of the CIREPP’s ac-
For a fault at zone one reach setting, the reactance and α1 tive and reactive power generation. This approach cannot
given by DS25 are 163 Ω and 279°, respectively, which are at properly explain the relay behavior in the event of unbalanced
even greater distances from their trip zones. short-circuits. On the other hand, it is instructive to examine
C. Case 3 the relay performance using the practices established for stud-
ying the impact of remote infeed and fault resistance on dis-
Case 1 with the NA-GC is re-simulated, while Rf is in- tance protection. The insight gained will be helpful in provid-
creased to 10 Ω. In contrast to case 1, where the GC require- ing a clue to overcome the problems later on.
ment was impossible to meet for the NA-GC, the PF in Fig.
9(a) indicates that the CIREPP restores the control over the A. Case 4
current and keeps the PF above 0.95 after the fault inception. A phase B to phase C to ground (BCG) fault with 10 Ω fault
resistance was simulated at 80% of line 25. The middle of the

0885-8977 (c) 2013 IEEE. Personal use is permitted, but republication/redistribution requires IEEE permission. See
http://www.ieee.org/publications_standards/publications/rights/index.html for more information.
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.2014.2369479, IEEE Transactions on Power Delivery

Current (pu) 4 Phase A Phase B Phase C VBL VBR

IBL Zf IBR
2 Rph Rg
0 KVL
-2 ICL Zf Rph ICR
VCL VCR
1990 2000 2010 2020 2030 2040 2050 2060 2070 2080 2090
Fig. 13. Circuit diagram of phases B and C for a BCG fault.
Time (ms)
| IBR – ICR |

Current (pu)
(a) Time = 2075.06
20
20 Phase A Phase B Phase C | IBL – ICL | Current = 25.2467
Current (pu)

10 10 Time = 2075.06
Current = 2.0693
0
0
-10 1990 2000 2010 2020 2030 2040 2050 2060 2070 2080 2090
Time (ms)
1990 2000 2010 2020 2030 2040 2050 2060 2070 2080 2090
(a)
Time (ms)
180  (I – ICR )  (I – ICL )

Angle (degree)
(b) BR BL
90
Fig. 11. Fault currents for case 4, (a) Local end, (b) Remote end. Time = 2075.06
60 0 Angle = -91.1506
40 -90
X ()

20 -180
0 1990 2000 2010 2020 2030 2040 2050 2060 2070 2080
Time =2090
2075.06
-20 Time (ms) Angle = -177.3709
-40 R = 11.6691 (b)
-10 X =0 -20.599910 20 30 40 50 60 70
Fig. 14. The terms in the numerator and denominator of MRf-BC for case 4, (a)
R ()
Magnitude, (b) Angle.
(a)
Time = 2075.06 tance to ZBC. A distance relay would, therefore, be in danger of
360  1 = 277.9704 overreach or underreach [26].
 (degree)

270
Time = 2075.06
Founded upon the voltage source modeling of the systems
180 behind the relays, a basic assumption in distance protection is
 2 = 275.8771
90 Trip zone 1 2 that the phase difference between the two end currents is simi-
0 lar to that between the two end pre-fault voltages [32]. As a
1990 2000 2010 2020 2030 2040 2050 2060 2070 2080 2090
Time (ms) result, the angles of the two end currents typically differ by
(b) only a few degrees, since a sizable phase difference between
Fig. 12. Operation of DS25 for case 4, (a) Impedance of the BC element, (b) the two end voltages may harbor the risk of instability. There-
Angle of the memory-polarized mho BC element.
fore, the fictitious reactance caused by the combined effect of
fault resistance was grounded. The fault currents of the
remote infeed and fault resistance is normally insignificant.
CIREPP when it follows the NA-GC are depicted in Fig.
In a conventional power plant with the excitation system for
11(a). Fig. 11(b) displays the currents at the remote end of the
the SGs, the phase angles of fault currents are determined
line, which are substantially larger. The impedance of the BC
based on the system impedance and the SGs’ voltage. For a
element of DS25 in Fig. 12(a) falls in the fourth quadrant out-
CIREPP, however, the converter current is governed by the
side zone one, thereby failing to trip the line. Both α1 and α2 of
control system and according to the host system GC or the in-
the BC element of the memory-polarized mho element remain
ternal references. Therefore, the above routine assumption re-
outside the trip zone in Fig. 12(b), as well.
garding the phase difference between the line end currents
For the circuit of Fig. 13, which demonstrates a BCG fault
does not hold true. The local currents in phases B and C of
with Rph and Rg as the phase-to-phase and phase-to-ground
Fig. 11(a) are IBL=0.893∡159.1° and ICL=2.515∡108.4° pu. For
resistances at the fault, Kirchhoff’s voltage law (KVL) is:
the remote end currents of Fig. 11(b), IBR and ICR equal
VBL  VCL  Z f I BL  Z f ICL  R ph  I BL  I BR  ICL  ICR  (3) 13.332∡165.4° and 13.117∡20.1° pu, respectively. Since the
where Zf is the positive sequence impedance between the relay zero-sequence component flowing through the ground of the
and the fault, VB and IB are phase B voltage and currents, VC CIREPP’s main transformer constitutes the bulk of the local
and IC are phase C voltage and currents, and the L and R sub- currents, IBL–ICL, which appears in the denominator of MRf-BC
scripts denote local and remote end quantities, respectively. and is plotted in Fig. 14(a), is much smaller than IBR–ICR, en-
Rearranging the terms in (3) and dividing the two sides over larging the magnitude of MRf-BC. More importantly, the phase
(IBL–ICL) yields the impedance of the BC element: difference between IBL–ICL and IBR–ICR in Fig. 14(b) exceeds
Z BC  Z f  R ph 1   I BR  ICR   I BL  ICL   (4) 86°. Thus, RphMRf-BC becomes 9.02-j60.87 Ω. Considering
Zf=2.86+j40.62 Ω, Rph=5 Ω and Rg=0 Ω, the ZBC given by (4)
M would be 11.88-j20.25 Ω, which agrees with the impedance
Rf BC
The primary effect of Rph MRf-BC is normally on the resistive shown in Fig. 12(a) and causes the relay to overreach.
part of ZBC. That is why a quadrilateral characteristic, whose B. Case 5
resistive reach can be set independently, is known to be more To demonstrate the heavy reliance of the above relay behav-
robust against the effect of fault resistance. Meanwhile, due to ior upon the CIREPP’s control system, consider the same fault
load flow in the system before the fault inception, the phase condition when the CIREPP meets the EU-GC. The local and
angles of the local and remote end currents are not identical, remote currents are depicted in Fig. 15. The resistance and re-
making MRf-BC a complex number, and adding a fictitious reac- actance of the BC element of DS25 are displayed with respect

0885-8977 (c) 2013 IEEE. Personal use is permitted, but republication/redistribution requires IEEE permission. See
http://www.ieee.org/publications_standards/publications/rights/index.html for more information.
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.2014.2369479, IEEE Transactions on Power Delivery

Current (pu) 4 Phase A Phase B Phase C | IBR – ICR |

Current (pu)
20 Time = 2099.37
2 | IBL – ICL | Current = 25.6193
0 10 Time = 2099.37
Current = 2.0797
-2
0
1990 2000 2010 2020 2030 2040 2050 2060 2070 2080 2090 1990 2000 2010 2020 2030 2040 2050 2060 2070 2080 2090
Time (ms) Time (ms)
(a) (a)
180

Angle (degree)
20 Phase A Phase B Phase C  (I – ICR )
Current (pu)

BR
90 Time = 2099.37
10  (I
BL
– ICL ) Angle = 154.2627
0
0
-90
-10
-180
1990 2000 2010 2020 2030 2040 2050 2060 2070 2080 2090 1990 2000 2010 2020 2030 2040 2050 2060 2070 Time
2080= 2099.37
2090
Time (ms) Time (ms) Angle = -176.629
(b) (b)
Fig. 15. Fault currents for case 5, (a) Local end, (b) Remote end. Fig. 17. The terms in the numerator and denominator of MRf-BC for case 5, (a)
Magnitude, (b) Angle.
Impedance ()

R Time = 2099.37
80 X = 70.9779 R = 65.5617
X 100 X = 64.9043 Case 6 Case 7
70

X ()
60 50
Time = 2099.37
50 R = 62.3652 0
1990 2000 2010 2020 2030 2040 2050 2060 2070 2080 2090
Time (ms) R = 10.4275
0 X =20 40 60 80 100 120
(a) -19.865
R ()
100
Fig. 18. Impedance of the BC element of DS25 for cases 6 and 7.
X ()

50 R = 62.3652 the remote infeed current. As a result, MRf-BC will have a large
X = 70.9779
imaginary part, and the measured reactance becomes very dif-
0
ferent from the actual reactance to the fault. Relay malopera-
0 20 40 60 80 100 120
R () tion for both GC categories confirms that the malfunction of
(b) DS25 does not stem from GC requirements.
Fig. 16. Performance of the BC element of an impedance-based DS25 for
case 5, (a) Impedance versus time, (b) Impedance plane. C. Other Fault Types
to time in Fig. 16(a). Similar to Figs. 6 and 10(a), the two The problems associated with line to line to ground (LLG)
curves undergo notable fluctuations caused by the CIREPP faults in relation to the combined effect of remote infeed and
transient response. Inside the impedance plane of Fig. 16(b), fault resistance are valid for line to line (LL) faults as well.
the operating point initially falls inside zone one for less than The positive and negative sequence components are basically
5 ms, and then moves through zone two, eventually lying at similar for LLG and LL short-circuits. The main difference is
Z=62.37+j70.98 Ω, outside the first two zones. The period the absence of the zero sequence component in LL fault cur-
during which the impedance passes through the first two zones rents. However, since the phase currents are subtracted in the
will be much shorter or may not even exist if the power con- numerator and denominator of MRf-BC, the zero-sequence cur-
trol loop is faster and/or CIREPP does not operate at unity PF rent does not impact MRf-BC and the consequent relay malfunc-
and generates reactive power prior to the fault–not an uncom- tion. Take, for instance, cases 4 and 5 without the ground at
mon scenario for the EU-GCs. the fault, designated as cases 6 and 7, respectively. The result-
IBL–ICL and IBR–ICR in Fig. 17 are 2.080∡154.3° and ant ZBC is depicted in Fig. 18. The impedances pinpointed in
25.619∡-176.6° pu, respectively. Therefore, diametrically op- this figure are quite close to those previously shown for cases
posed to the impedance of case 4, where RphMRf-BC had a large 4 and 5, and are far outside zone one of DS25.
negative imaginary part, the fictitious reactance added to ZBC In contrast to balanced, LLG and LL faults, for which the
is 58.81+j29.96 Ω, leading to considerable underreach. The calculated reactance may not be even remotely close to the ac-
marked difference between ZBC for cases 4 and 5 happens de- tual fault reactance, a distance relay performs more successful-
spite the identical pre-fault voltage angles at the two line ends ly for single line to ground (SLG) faults. That being said, the
and exactly the same fault condition, which should result in common countermeasures against underreach and overreach
identical underreach or overreach according to well- of a ground distance element due to remote infeed does not
established distance protection principles. yield successful results if a CIREPP is connected to the line,
Unlike a conventional power plant, the fault current angle requiring special precautions elaborated in Part II of the paper.
for a CIREPP is regulated by the control system of the con-
verter. The reference angle for the current is typically given by VI. ZONE TWO FAULTS
host system GC. Meanwhile, if the fault current characteristics In the event of short-circuits on neighboring lines, DS25 is
are not dictated by the GC, or if the CIREPP does not follow a in danger of malfunction even if the fault is bolted. While the
GC, the current angle is determined by the internal references relay suffered from lack of dependability in the last two sec-
of the converter’s current control loop. No matter how the ref- tions, the faults on adjacent lines may jeopardize the security
erence angle is specified, or which GC the CIREPP complies of the protection system and, in turn, neutralize the GCs’ FRT
with, this angle can be significantly different from the angle of

0885-8977 (c) 2013 IEEE. Personal use is permitted, but republication/redistribution requires IEEE permission. See
http://www.ieee.org/publications_standards/publications/rights/index.html for more information.
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.2014.2369479, IEEE Transactions on Power Delivery

Phase A Phase B Phase C 20 | IBif – ICif |

Current (pu)

Current (pu)
2 Time = 2080
| IBL – ICL | Current = 21.6136
0 10 Time = 2080
Current = 2.0717
-2
0
1990 2000 2010 2020 2030 2040 2050 2060 2070 2080 2090 1990 2000 2010 2020 2030 2040 2050 2060 2070 2080 2090
Time (ms) Time (ms)
(a) (a)
180

Angle (degree)
20 Phase A Phase B Phase C
Current (pu)

Time = 2080
10 90  (I – ICif ) Angle = 174.8061
Bif
Time = 2080
0 0  (I
BL
– ICL ) Angle = -79.9209
-10 -90
1990 2000 2010 2020 2030 2040 2050 2060 2070 2080 2090 1990 2000 2010 2020 2030 2040 2050 2060 2070 2080 2090
Time (ms) Time (ms)
(b) (b)
Fig. 19. Fault currents for case 8, (a) Local currents, (b) Infeed currents. Fig. 22. The numerator and denominator of Mif-BC for case 8, (a) Magnitude,
(b) Angle.
80
60 Z BC  Z1  Z 2  Z 2 I Bif  I Cif    I BL  ICL 
X ()

(6)
40
20 R = 54.072
M
0 X = 38.274 if BC
0 20 40 60 80 100 120 140 Z2Mif-BC is the impedance added to ZBC by the intermediate
R ()
Fig. 20. The impedance of the BC element of DS25 for case 8. infeed. As discussed earlier, the difference between the angles
of the local and infeed currents in a conventional power sys-
VBL
IBL
Z1 Z2 tem is considered to be almost equal to the difference between
KVL the voltage angles before the fault, which is normally small.
ICL
VCL
Z1 Z2 Therefore, the imaginary part of Mif-BC is insignificant, and the
angle of Z2 Mif-BC is very close to that of the fault impedance.
IBi f ICi f
That is why a distance relay has been traditionally known to
Fig. 21. Circuit diagram for a BCG fault with intermediate infeed. underreach in the presence of intermediate infeed.
articles. The relay may also fail to provide the anticipated Since the angles of CIREPP-generated currents are not re-
back-up protection. lated to pre-fault voltages and are determined based on the GC
requirements or internal references, the above analysis does
A. Case 8
not hold true in the presence of a CIREPP. Even for the same
The currents recorded by DS25 for a bolted zone two BCG GC, the angle of a CIREPP fault current may be quite unpre-
fault on line 45, only 10 km away from bus 5 when the dictable. Take, for instance, the Spanish GC where the operat-
CIREPP complies with the NA-GC are plotted in Fig. 19(a). ing point of a CIREPP can be anywhere inside the shaded area
While DS25 operation must be delayed to let line 45’s relays of Fig. 1, or the German GC where the slope of the reactive
clear the fault, the impedance seen by the BC element of DS25 current curve versus the voltage drop can vary from 0 to 10
in Fig. 20 illustrates that zone one of the relay is asserted soon [15], or the NA-GC that allows any PF higher than 0.95. On
after the onset of the fault; thus, the line is tripped immediate- top of that, the GC reactive current requirements always per-
ly. Any FRT scheme implemented in the CIREPP is subse- mit a transient response time, which is longer than operational
quently negated by the incorrect relay operation. times of distance relays. For example, the operating point of a
Since the fault is bolted, the relay malfunction is not caused Spanish WF is allowed to be outside the shaded area of Fig. 1
by the currents flowing from the remote end of line 45. The for 150 ms after the voltage drop, resulting in erratic current
other currents present in the fault loop are the intermediate in- angles during distance relay operation time spans. No matter if
feeds, i.e., the currents flowing from lines 15, 35 and the reac- the CIREPP complies with the GC immediately after the fault,
tor. Fig. 19(b) shows the sum of the infeed currents, which are or a response time is allowed, the angles of the CIREPP cur-
appreciably larger than the local currents, due to the CIREPPs’ rents are regulated by the CIREPP’s control system, and are,
modest short-circuit capacity. Meanwhile, intermediate infeed thus, considerably different from the angles of the infeed cur-
has been known to cause a distance relay to underreach, rents, causing relay maloperation.
whereas the relay has overreached. To investigate this issue, After the transients fade out, the phase B and C infeed cur-
let us consider the circuit diagram of the fault in Fig. 21. The rents in Fig. 19 are 11.578∡158.8° and 10.960∡11.7° pu, re-
KVL for the BC loop can be expressed as spectively. The DS25 currents are IBL=0.484∡-163.5° and
VBL  VCL  Z1  I BL  ICL   Z 2 I BL  I Bif  I CL  I Cif   (5) ICL=2.074∡113.5° pu. Hence, as shown in Fig. 22, the numera-
tor and denominator of Mif-BC are 21.614∡174.8° and 2.072∡-
where the if subscript denotes infeed quantities, Z1 is the im-
79.9° pu, respectively, and exhibit a significant phase differ-
pedance between the relay and infeed location, and Z2 is the
ence. Therefore, Z2Mif-BC lags Z2 by 105.3°, which makes the
impedance between the infeed and the fault. Dividing (5) over
reactance seen by DS25 smaller than the actual reactance to
IBL–ICL yields:
the fault by about 18 Ω, and induces substantial overreach.
Similar incorrect operation is observed for mho elements.

0885-8977 (c) 2013 IEEE. Personal use is permitted, but republication/redistribution requires IEEE permission. See
http://www.ieee.org/publications_standards/publications/rights/index.html for more information.
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.2014.2369479, IEEE Transactions on Power Delivery

Phase A Phase B Phase C

Current (pu)
80
2
60

X ()
0 40
20 R = 15.2552
-2 0 X = 39.6445

1990 2000 2010 2020 2030 2040 2050 2060 2070 2080 2090 -20 -10 0 10 20 30 40 50 60 70
Time (ms) R ()
(a) Fig. 26. The impedance of the BC element of DS25 for the fault of cases 4
20 Phase A Phase B Phase C and 5 when the CIREPP is replaced by a conventional source.
Current (pu)

Time = 2099.0883
| IBR – ICR |

Current (pu)
10 30 Current = 24.6573
| IBL – ICL |
0 20
-10 10 Time = 2099.0883
1990 2000 2010 2020 2030 2040 2050 2060 2070 2080 2090 Current = 16.7962
0
Time (ms) 1990 2000 2010 2020 2030 2040 2050 2060 2070 2080 2090
(b) Time (ms)
Fig. 23. Fault currents for case 9, (a) Local currents, (b) Infeed currents. (a)

Angle (degree)
20 | IBif – ICif |  (I – ICR )  (I – ICL )
Current (pu)

Time = 2080 -45 BR BL

| IBL – ICL | Current = 21.6383 -90 Time = 2099.0883
10 Time = 2080 -135 Angle = -163.5037
Current = 2.0829
-180
0 Time = 2099.0883
1990 2000 2010 2020 2030 2040 2050 2060 2070 2080 2090 1990 2000 2010 2020 2030 2040 2050 2060 2070Angle
2080 2090
= -171.9162
Time (ms) Time (ms)
(a) (b)
180  (I – ICif ) Fig. 27. The terms in the numerator and denominator of MRf-BC for the fault of
Angle (degree)

Bif
90 Time = 2080 Fig. 26, (a) Magnitude, (b) Angle.
 (I
BL
– ICL ) Angle = 175.2209
0
Time = 2080
location is farther from the infeed point or if there is a fault re-
-90 Angle = -167.4172 sistance. The above cases have discussed LLG faults, but the
-180 same problems also apply to LL and balanced faults.
1990 2000 2010 2020 2030 2040 2050 2060 2070 2080 2090
Time (ms) Similar to the previously discussed case of remote infeed,
(b) maloperation of DS25 in the presence of intermediate infeed
Fig. 24. The numerator and denominator of Mif-BC for case 9, (a) Magnitude, arises from the fault behavior intrinsic to CIREPPs, and is not
(b) Angle. caused by particular GC requirements.
150
100 VII. DS25 OPERATION FOR A CONVENTIONAL SOURCE
X ()

R = 23.177
50 X = 105.2676 To further highlight the effect of a CIREPP on line protec-
0
tion by DS25, the CIREPP was replaced by a conventional
0 10 20 30 40 50 60 70 80 90 source whose impedance and phase angle are 12.7∡88.9° Ω
R () and 4.5°. The active and reactive powers of this source are
Fig. 25. The impedance of the BC element of DS25 for case 9. similar to those of the CIREPP. As shown in Fig. 26, the im-
pedance calculated by DS25 for the zone one fault of cases 4
B. Case 9 and 5 correctly enters zone one in less than one cycle. Unlike
The fault of case 8 is simulated when the CIREPP conforms Figs. 12(a) and 16(b), which exhibited huge overreach and un-
to the EU-GC. The fault currents are shown in Fig. 23. At the derreach, a less than 1 Ω overreach is observed. If the fault is
relay side, the phase B and C currents are 1.827∡130.4° and on the zone boundary, such small overreach can easily be ad-
2.031∡65.3° pu, respectively, making the denominator of Mif- dressed by tilting the reactance element of the relay by a few
°
BC in Fig. 24 equal to 2.083∡-167.4 pu. For the infeed quanti- degrees. The correct operation of DS25 in Fig. 26 stems from
ties, IBif=11.699∡160.0 and ICif=10.799∡11.8° pu, leading to
°
the relatively close angles of the two line end fault currents.
21.638∡175.2° pu for the numerator of Mif-BC in Fig. 24. Plug- The magnitudes and angles of the terms in the numerator and
ging these values into (6), the reactance calculated by the relay the denominator of MRf-BC are plotted in Fig. 27. In contrast to
becomes 105.08 Ω, which is consistent with the impedance of Figs. 14(b) and 17(b), the current magnitudes and angles in
DS25 in Fig. 25, and is almost twice the actual fault reactance. Fig. 27 are very close. Plugging the values pinpointed in Fig.
In contrast to case 8, where the intermediate infeed caused 27 into (4) yields the impedance shown in Fig. 26.
overreach, DS25 underreaches for this case. The excessive The operation of DS25 was tested for the zone two fault of
magnitude of underreach arises from the limited non-zero se- cases 8 and 9 as well. The impedance in Fig. 28 is almost at
quence components of the CIREPP currents and the removal the edge of zone two. Nevertheless, this is the normal un-
of the zero-sequence component due to the subtraction of the derreach expected due to the intermediate infeed. In contrast
current phasors in (6). Apart from the huge underreach, the to Figs. 22 and 24, the numerator and denominator of Mif-BC,
impedance shown in Fig. 25 includes a significant resistive displayed in Fig. 29, are almost aligned, as a result of which
part as well, even though the fault is bolted. the impedance in Fig. 28 exhibits the normal zone two un-
The above problems exist in spite of the fact that the infeed derreach, and is also situated on the line replica impedance.
currents flow through less than 10% of the distance between Moreover, the comparable magnitudes of the local and infeed
the fault and DS25. The problems are exacerbated if the fault currents keep the amount of underreach small.

0885-8977 (c) 2013 IEEE. Personal use is permitted, but republication/redistribution requires IEEE permission. See
http://www.ieee.org/publications_standards/publications/rights/index.html for more information.
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.2014.2369479, IEEE Transactions on Power Delivery

100 [5] M. Khoddam and H. K. Karegar, “Effect of wind turbines equipped with
doubly-fed induction generators on distance protection,” Intl. Conf. on
X ()

50
Advanced Power System Automation and Protection, Oct. 2011, Beijing.
R = 6.1828 [6] S. D. Rijcke, P. S. Perez, and J. Driesen “Impact of wind turbines
X = 64.2607
equipped with doubly-fed induction generators on distance relaying,”
0
IEEE Power and Energy Soc. Gen. Meeting, Jul. 2010, Minneapolis.
-20 -10 0 10 20 30 40 50 60 70 [7] L. Qin-hao, Z. Yong-jun, Y. Wei-peng, and L. Feng, “A study on influ-
R ()
ence of wind power on positive sequence voltage polarized impedance
Fig. 28. The impedance of the BC element of DS25 for the fault of cases 8 relay,” IEEE Power Eng. and Automation Conf., Sep. 2012, Wuhan.
and 9 when the CIREPP is replaced by a conventional source.
Time = 2099.0883
[8] A. Hooshyar, M. A. Azzouz, and E. F. El-Saadany, “Distance protection
30 | IBif – ICif | of lines connected to induction generator-based wind farms during bal-
Current (pu)

Current = 20.1145
20 | IBL – ICL | anced faults,” IEEE Trans. Sustain. Energy, vol. 5, no. 4, pp. 1193–
1203, Oct. 2014.
10 [9] N. Flourentzou, V. G. Agelidis, and G. D. Demetriades, “VSC-based
Time = 2099.0883
Current = 11.8015 HVDC power transmission systems: An overview,” IEEE Trans. Power
0
1990 2000 2010 2020 2030 2040 2050 2060 2070 2080 2090 Electron., vol. 24, no. 3, pp. 592–602, Mar. 2009.
Time (ms) [10] P. M. Anderson, “Sequence impedance of machines,” in Analysis of
(a) Faulted Power Systems, IEEE reissued edn., New York, NY: Wiley-
180  (I – ICif ) IEEE Press, 1995, pp. 183–230.
Angle (degree)

90
Bif
Time = 2099.0883 [11] J. Rocabert, A. Luna, F. Blaabjerg, and P. Rodriguez, “Control of power
 (I
BL
– ICL ) Angle = 175.2105 converters in AC microgrids,” IEEE Trans. Power Electron., vol. 27, no.
0
Time = 2099.0883 11, pp. 4734–4749, Nov. 2012.
-90 Angle = -175.0282 [12] “ENTSO-E network code for requirements for grid connection applica-
-180 ble to all generators,” Brussels, Belgium, Mar. 2013. [Online]. Availa-
1990 2000 2010 2020 2030 2040 2050 2060 2070 2080 2090 ble:
Time (ms) http://networkcodes.entsoe.eu/wp-
(b) content/uploads/2013/08/130308_Final_Version_NC_RfG1.pdf
Fig. 29. The numerator and denominator of Mif-BC for the fault of Fig. 28, (a) [13] M. Tsili and S. Papathanassiou, “A review of grid code technical re-
Magnitude, (b) Angle. quirements for wind farms,” IET Renew. Power Gen., vol. 3, no. 3, pp.
308–332, Sep. 2009.
VIII. CONCLUSION [14] “Procedure for verification validation and certification of the require-
ments of the PO 12.3 on the response of wind farms and photovoltaic
For non-bolted balanced faults in zone one of DS25, the plants in the event of voltage dips,” Version 9, May 2011. [Online].
CIREPP keeps control over the current. The reactive current Available: www.aeeolica.org
for EU-GCs translates into impedances close to the X axis. [15] “Ordinance on system services by wind energy plants (System service
ordinance – SDLWindV),” Jun. 2010. [Online]. Available:
The CIREPP’s large PF for the NA-GCs makes the impedance www.erneuerbare-energien.de
lie along the R axis. In both cases, the relay impedance can be [16] “Interconnection for wind energy,” Docket No. RM05-4-001, Order No.
far away from the actual fault impedance and is highly influ- 661-A, United States of America Federal Energy Regulatory Commis-
enced by the parameters of the CIREPP’s control system, such sion, Dec. 2005. [Online]. Available: www.ferc.gov
[17] M. P. LeBlanc et al “Canadian grid code for wind development - Re-
as the current limit setting. view and recommendations,” Canadian Wind Energy Association, Oct.
The common assumption regarding the phase difference be- 2005. [Online]. Available: www.nrcan.gc.ca
tween the line end currents during a zone one fault is not valid [18] “ENTSO-E draft network code on high voltage direct current connec-
for a line that emanates from a CIREPP. Consequently, it is tions and DC-connected power park modules,” Draft for public consulta-
tion, ENTSO-E, Nov. 2013. [Online]. Available: www.entsoe.eu
highly likely that the underreach or overreach of DS25 due to [19] E. Muljadi et al “Method of equivalencing for a large wind power plant
the remote infeed during LL or LLG faults will become very with multiple turbine representation,” IEEE Power and Energy Soc.
large. The relay may also underreach for a case where it is General Meeting, Pittsburgh, PA, Jul. 2008.
normally expected to overreach. [20] J. Brochu, C. Larose, and R. Gagnon, “Validation of single- and multi-
ple-machine equivalents for modeling wind power plants,” IEEE Trans.
The intermediate infeed has been known to cause distance Energy Convers., vol. 26, no. 2, pp. 532–541, Jun. 2011.
relays to underreach. However, the angles of the local and in- [21] A. Timbus et al “Evaluation of current controllers for distributed power
feed currents may be widely apart if a CIREPP is behind the generation systems,” IEEE Trans. Power Electron., vol. 24, no. 3, pp.
relay. Depending on the CIREPP’s control strategy, the relay 654–664, Mar. 2009.
[22] S. Li, T. A. Haskew, “Converter control of variable-speed wind tur-
may overreach during zone two faults, which negates any FRT bines,” U.S. Patent 8 577 508 B2, Nov. 5, 2013.
scheme, or experience a huge underreach and fail to provide [23] P. Rodriguez et al “New positive-sequence voltage detector for grid syn-
back-up operation. chronization of power converters under faulty grid conditions,” Proc.
37th IEEE Power Electron. Specialists Conf., Jun. 2006, pp. 1–7.
[24] J. F. Conroy and R. Watson, “Low-voltage ride-through of a full con-
REFERENCES verter wind turbine with permanent magnet generator,” IET Renewable
[1] A. Morales et al “Advanced grid requirements for the integration of Power Gen., vol. 1, no. 3, pp. 182–189, Sep. 2007.
wind farms into the Spanish transmission system,” IET Renewable Pow- [25] M. Asmine et al., Ad Hoc Task Force on Wind Generation Model Vali-
er Gen., vol. 2, no. 1, pp. 47–59, Mar. 2008. dation, “Model validation for wind turbine generator models,” IEEE
[2] T. Seegers et al., “Impact of distributed resources on distribution relay Trans. Power Syst., vol. 26, no. 3, pp. 1769–1782, Aug. 2011.
protection,” IEEE Power System Relaying Committee, Aug. 2004. [26] S. H. Horowitz and A. G. Phadke, “Nonpilot distance protection of
[Online]. Available: www.pes-psrc.org transmission lines,” in Power System Relaying, 3rd ed., New York, NY:
[3] M. Bollen and F. Hassan, “Protection,” in Integration of Distributed Wiley, 2008, pp. 101–131.
Generation in the Power System, 1st. ed., Hoboken, NJ: Wiley-IEEE [27] M. S. Sachdev et al., “Understanding microprocessor-based technology
Press, 2011, pp. 299–366. applied to relaying.” IEEE Power System Relaying Committee, Jan.
[4] A. K. Pradhan and G. Joos, “Adaptive distance relay setting for lines 2009. [Online]. Available: www.pes-psrc.org
connecting wind farms,” IEEE Trans. Energy Convers., vol. 22, no. 1, [28] M. P. Kazmierkowski and L. Malesani, “Current control techniques for
pp. 206–213, Mar. 2007. three-phase voltage-source PWM converters: A survey,” IEEE Trans.
Ind. Electron., vol. 45, no. 5, pp. 691–703, Oct. 1998.

0885-8977 (c) 2013 IEEE. Personal use is permitted, but republication/redistribution requires IEEE permission. See
http://www.ieee.org/publications_standards/publications/rights/index.html for more information.
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.2014.2369479, IEEE Transactions on Power Delivery

[29] “Line distance protection REL670 application manual,” ABB, Vasteras,

Sweden, Dec. 2012. [Online]. Available:
http://new.abb.com/substation-automation/products/protection-
control/line-distance-protection/rel670
[30] “D90Plus line distance protection system,” GE, Markham, Canada, Mar.
2012. [Online]. Available:
www.gedigitalenergy.com/products/manuals/d90plus/gek-113240b.pdf
[31] E. O. Schweitzer III, “Computationally-efficient distance relay for pow-
er transmission lines,” U.S. Patent 5 325 061, Jun. 28, 1994.
[32] G. Ziegler, “Mode of operation,” in Numerical Distance Protection:
Principles and Applications, 4th ed., Erlangen, Germany: Publicis Pub-
lishing, 2011, pp. 20–208.

Ali Hooshyar received the B.Sc. and M.Sc. degrees in

electrical engineering from Isfahan University of Tech-
nology, Isfahan, Iran, and the University of Tehran, Teh-
ran, Iran, in 2006 and 2009, respectively. Currently, he is
pursuing the PhD degree in electrical engineering at the
University of Waterloo, Waterloo, ON, Canada.
His research interests include protection and control of
renewable energy systems and smart grids.

Maher Abdelkhalek Azzouz (S’10) was born in Giza,

Egypt, on 1986. He received the B.Sc. and M.Sc. degrees
(with honors) in electrical engineering from Cairo Univer-
sity, Giza, Egypt, in 2008 and 2011, respectively.
Currently, he is pursuing the Ph.D. degree at the De-
partment of Electrical and Computer Engineering, Univer-
sity of Waterloo, Waterloo, ON, Canada. His research in-
terests include dynamics and control of power converters,
distributed and renewable generation, and control of smart
distribution systems.

E. F. El-Saadany (SM’05) was born in Cairo, Egypt,

in 1964. He received the B.Sc. and M.Sc. degrees in elec-
trical engineering from Ain Shams University, Cairo,
Egypt, in 1986 and 1990, respectively, and the Ph.D. de-
gree in electrical engineering from the University of Wa-
terloo, Waterloo, ON, Canada, in 1998.
Currently, he is a Professor in the Department of Elec-
trical and Computer Engineering, University of Waterloo.
His research interests are smart grids operation and con-
trol, power quality, distributed generation, power elec-
tronics, digital signal processing applications to power systems, and mecha-
tronics.

0885-8977 (c) 2013 IEEE. Personal use is permitted, but republication/redistribution requires IEEE permission. See
http://www.ieee.org/publications_standards/publications/rights/index.html for more information.