See discussions, stats, and author profiles for this publication at: https://www.researchgate.

net/publication/305417550

Analytical Approach to Estimate Feeder Accommodation Limits Based on

Protection Criteria

Article  in  IEEE Access · January 2016

DOI: 10.1109/ACCESS.2016.2589545

CITATIONS READS
17 4,189

4 authors:

Harsha Vardhana Padullaparti Pisitpol Chirapongsananurak

National Renewable Energy Laboratory Chulalongkorn University
37 PUBLICATIONS   254 CITATIONS    21 PUBLICATIONS   77 CITATIONS   

SEE PROFILE SEE PROFILE

Miguel Hernandez Surya Santoso

Electric Power Research Institute University of Texas at Austin
34 PUBLICATIONS   251 CITATIONS    293 PUBLICATIONS   6,682 CITATIONS   

SEE PROFILE SEE PROFILE

Some of the authors of this publication are also working on these related projects:

PV smart inverter control tuning View project

Photovoltaic Hosting Capacity View project

All content following this page was uploaded by Surya Santoso on 31 January 2017.

The user has requested enhancement of the downloaded file.

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/ACCESS.2016.2589545, IEEE Access
Analytical Approach to Estimate Feeder
Accommodation Limits based on Protection Criteria
Harsha V. Padullaparti, Student Member, IEEE, Pisitpol Chirapongsananurak, Student Member, IEEE,
Miguel Hernandez, Student Member, IEEE, and Surya Santoso, Fellow, IEEE
Abstract—Widespread deployment of large-scale photovoltaics
(PV) and energy storage systems (ESS) in distribution net-
works necessitates the development of methods to assess their
possible system impacts. One of the primary concerns related
to the integration of these systems is short-circuit overcurrent
protection problems. Present techniques use simulations of full
and detailed distribution circuit models with large number of
scenarios to estimate the PV and ESS sizes a distribution feeder
can accommodate without causing adverse impacts. As this
process requires considerable time and effort, this paper develops
a practical and simplified analytical approach to conservatively
estimate a utility distribution feeder’s accommodation limits
without causing protection problems. Sympathetic tripping and
relay insensitivity problems are considered in this work under
both symmetrical and unsymmetrical fault conditions. Using
the analysis presented, the factors influencing these protection
problems are determined to provide insights into relay settings.
The feeder accommodation limits obtained using the proposed
analytical approach are compared with those obtained using
simulations of an actual detailed distribution circuit model.
The findings show that the proposed approach is accurate in
estimating the feeder’s accommodation limits for integrating
large-scale PV and ESS.
Index Terms—Photovoltaic systems, energy storage, power
distribution faults, short circuit currents
I. I NTRODUCTION
T HE installed capacity of photovoltaic (PV) systems has
grown rapidly in the past decade in the U.S. This growth
can be attributed primarily to the decline in the PV prices
and the incentives from the government [1]. Being highly
scalable, the PV installations can be found in various sizes
ranging from few kilowatts to several hundreds of megawatts.
The PV systems of smaller ratings such as those found on
residential and commercial rooftops can be considered as
small-scale PV systems. According to Energy Information
Administration (EIA), the PV systems of ratings above 1 MW
can be considered as utility-scale or large-scale PV systems
[2]. Besides the small-scale residential PV systems, large-scale
PV installations are also in commercial operation now in the
distribution feeders [3]. Integration of distributed generation
(DG) such as PV, in general, brings numerous technical
benefits such as reduced line losses, improved voltage profiles,
and enhanced system reliability [4]. However, it also poses
challenges to existing system protection schemes and utility
voltage regulation practices [5].
Before integrating DG, system studies need to be performed
to assess associated impacts on the existing overcurrent protec-
tion schemes [6]. The impact is highly dependent on the type
of DG being integrated. Influence of rotating machine-based
2169-3536 (c) 2016 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.
DG (RBDG) such as synchronous machine-based DG and
induction machine-based DG on feeder overcurrent protection
schemes is reported in [7], [8]. As opposed to the RBDG,
inverter-interfaced distributed generation (IIDG) such as PV
systems do not make much fault current contribution. While
the fault current contribution of PV systems can vary from
1.1 to 2.5 times the inverter rated current depending on the
type of inverter, the rule of thumb in the industry for system
studies is 2 times the inverter rated current [9], [10]. Such
low PV fault currents may not interfere with the overcurrent
protection schemes in case of small-scale PV systems with
low penetration level. However, large-scale PV installations
can inject high fault currents into the system and can lead
to protection problems [11], [12]. Some of the operational
issues associated with PV can be mitigated by installing and
operating ESS along with PV [13].
Energy storage systems (ESS) have many applications in
both transmission and distribution systems. In a distribution
system, the benefits of ESS include peak shaving, load level-
ing, time-shifting of renewable energy, and voltage stabiliza-
tion [14]. Although the ESS have various benefits, their wide-
spread deployment is being limited by few barriers. The key
barrier among them as on date is the cost. Nevertheless, there
are numerous large-scale ESS projects in commercial opera-
tion now world-wide [15]. Further, the deployment of ESS is
expected to grow dramatically in near future [16]. Similar to
PV systems, the fault current contributions from ESS can also
have adverse impacts on the distribution feeder overcurrent
protection schemes [17]. Considering the widespread existence
and expected growth of large-scale ESS, there is a need to
study the influence of ESS on feeder overcurrent protection
schemes.
The amount of PV a distribution feeder can accommodate
without violating the nominal system operating conditions
is the feeder’s PV hosting capacity or feeder’s PV accom-
modation limit [18]. In case of ESS, it can be referred to
as ESS accommodation limit of the circuit. Approaches for
determination of a feeder’s accommodation limits with relay
insensitivity and sympathetic tripping as impact criteria using
circuit simulations are reported in [19], [20]. Keeping in view
of the new grid regulations and worst case system conditions,
these studies are performed considering ‘low voltage ride
through’ (LVRT) requirement of the PV systems. With the
growing concerns to integrate higher amounts of PV into distri-
bution systems along with possible adverse impact mitigating
measures such as installing ESS, suitable methods are required
for estimating the sizes of large-scale PV and ESS a feeder
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/ACCESS.2016.2589545, IEEE Access
can accommodate without creating protection problems. In
this work, an analytical approach is presented to determine
a distribution feeder’s accommodation limits of integrating
large-scale PV and ESS based on relay insensitivity and
sympathetic tripping as impact criteria considering LVRT. The
analysis presented in this work is validated against simulation
studies on a distribution circuit model in OpenDSS developed
using actual circuit data from a utility. The results match
within 7% of theoretical predictions, thus validate the analysis.
This paper is organized into six sections. The problem
statement and the motivation of this work are detailed in
Section II. An overview of distribution feeder overcurrent
protection is provided in Section III. The relay insensitivity
and sympathetic tripping problems are also described in this
section. In Section IV, the analysis of relay insensitivity and
sympathetic tripping protection problems in the presence of
large-scale PV and ESS is presented and key parameters in-
fluencing these problems are discussed. Section V presents the
simulation results using a detailed distribution circuit model
that validate the analysis. The conclusions and discussion are
provided in Section VI.
II. P ROBLEM S TATMENT AND M OTIVATION
Protection studies, similar to other impact studies, for
deployment of PV systems require development of detailed
distribution system models using distribution system sim-
ulation tools [21]. Then the developed model is validated
using suitable methods such as comparing power flow results
with actual field measurements. Once the validated model
is available, simulation studies are performed to assess the
impact of PV integration on protection schemes with large
number of scenarios. However, this process requires consider-
able time and effort [22]. Furthermore, in this particular case
of determining feeder accommodation limits of integrating PV
using circuit simulations, the approach presented in [19] is
useful to study only small test systems with few nodes. Such a
simulation based approach when applied to actual distribution
feeders having thousands of buses poses several complications
of scalability. Some of the reported scalability challenges are
power flow solution convergence problems, complexity of
analysis and visualization of results considering large number
of fuses, huge time requirement to study all possible combina-
tions of PV sizes, and fault locations [20]. These complexities
can be handled to some extent by using network reduction
technique [23] and some reasonable approximations such as
modeling only few main protection devices such as reclosers,
simulating only few representative scenarios of selected PV
locations [20]. However, the accommodation limits obtained
as outcome pertains only to that particular circuit and is valid
only for a particular set of protection device settings. As such,
the conclusions may not necessarily be applicable to other
circuits or even to the same circuit with different protection
device settings as the results are highly dependent on the
circuit characteristics.
In the absence of insights into the relationship among
various system parameters, the utility planners are usually
forced to apply conservative rules of thumb penetration levels
2169-3536 (c) 2016 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.
to accept PV interconnection requests. For PV integration, the
penetration level is limited to 15% of the maximum feeder
load on a line section to avoid any adverse impacts [19].
Alternatively, system studies need to be performed with all
possible combinations of fault types, fault locations, PV size
and locations which is time consuming posing scalability
challenges and resulting in circuit specific outcomes. In this
context, the main contribution of this work is development of
a scalable analysis-based method to conservatively estimate
large-scale PV and ESS accommodation limits of a distribution
feeder without causing relay insensitivity and sympathetic
tripping problems. The objective is to simplify the estimation
process with reasonable approximations under conservative
settings. The other contribution is providing insights into
overcurrent relay settings when large-scale PV and ESS are
present. The analysis for assessing the impact of ESS installed
either individually or in conjunction with PV on the protection
relay response and its application to assess accommodation
limits of a utility distribution feeder is not reported in litera-
ture.
In recent years, there are several changes to the regulations
for grid interconnection of DG. While previous regulations
mandated disconnection of DG during abnormal system con-
ditions, many new regulations require the DG to remain
connected during the faults, referred to as ‘low voltage ride
through’ (LVRT) or ‘fault ride through’ (FRT). Similar to DG,
as part of grid support functions, the ESS also offers LVRT
functionality [24]. Since protection studies are performed con-
sidering the worst case situation, it is reasonable to consider
that the large-scale PV and ESS remain connected in the
system during the fault conditions and assess the impact of
their steady state current injection into the distribution network
[20]. Thus, in this work, analysis is provided considering
LVRT requirements of PV and ESS.
III. OVERVIEW OF D ISTRIBUTION F EEDER OVERCURRENT
P ROTECTION AND I MPACTS OF PV AND ESS
Majority of the conventional distribution systems are radial
in nature as the radial systems are least expensive, simpler in
planning, design and operation compared to other alternatives
[25]. In radial systems, the primary substation is the only
source feeding the distribution load and there exists only
a single path between each customer and the substation.
Typically, overcurrent protection is the main feeder protec-
tion in radial systems. Overcurrent relays, circuit breakers,
reclosers, sectionalizers and fuses are the main protection
devices in overcurrent protection schemes. The role of these
protection devices and the overcurrent protection coordination
are briefly explained in this section. Then relay insensitivity
and sympathetic tripping problems that may arise due to the
presence of large-scale PV and ESS are described.
A. Overcurrent Protection Devices
1) Circuit Breakers: Circuit breakers are used to protect the
distribution networks by interrupting the fault currents during
faulty circuit conditions. Overcurrent relays control the circuit
breaker operations. While isolating the faulted line section,
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/ACCESS.2016.2589545, IEEE Access
due to the interruption of inductive current, electric arc forms
between the circuit breaker contacts. To quench the electric
arc, circuit breakers use different types of arc-extinguishing
media such as air, oil, vacuum or SF6 gas.
2) Overcurrent relays: Overcurrent relays are the fault
sensing devices that detect faults in the distribution system
and open the circuit breaker. They can function as phase
overcurrent or ground overcurrent relays depending on the
current being monitored. Phase overcurrent relays detect the
fault currents flowing in the lines. When the line currents are
above the phase overcurrent relay pickup setting, the relay
trips the circuit breaker. The trip time depends upon the time-
dial setting of the relay. For detecting the unbalanced faults,
ground overcurrent relays are used. These relays provide more
sensitive protection for phase to ground faults. The ground
overcurrent relays respond to the residual or unbalance current
which is the sum of the currents in all the three phases in
a three-phase distribution system. When the residual current
seen by the ground overcurrent relay exceeds the relay pickup
setting, it opens the circuit breaker.
3) Reclosers: A recloser trips the circuit whenever a fault is
detected and automatically recloses after a predetermined time
in the anticipation of the fault getting cleared. This reclosing
action is performed a preset number of times after which the
recloser will stay open and isolates the fault if the fault still
persists in the circuit. As majority of the faults in the overhead
distribution systems are temporary in nature and last for only
a few cycles, the reclosers avoid unnecessary service outage
for temporary faults. Reclosers can detect both phase and
ground overcurrent conditions and therefore have the ability
to protect the distribution network in case of both balanced
and unbalanced permanent faults.
4) Sectionalizers: Sectionalizers are used to isolate the
faulted sections of a distribution circuit after an upstream
protection device has interrupted the fault current. Since
sectionalizers do not have fault current interrupting capability,
they must be used in conjunction with backup protection
devices such as circuit breaker or recloser and are installed
downstream from the backup protection device. Under fault
conditions, the sectionalizer counts the number of upstream
recloser operations. In case of temporary faults, the fault
gets cleared before the recloser finishes the preset number of
reclose operations. Then the circuit recloses successfully and
the sectionalizer counter resets to its normal position. Thus,
the service to the customers is continued with few momentary
interruptions during recloser operations. In case of permanent
faults, the recloser remains open after the preset number of
recloser operations, isolating the faulty section of line.
5) Fuses: Fuses consist of a fusible metallic conductor link
that passes current under normal circuit conditions. During
faults, the fuse link melts due to overheat produced by the
fault current and opens the fuse contacts, thus interrupts the
fault current. The time of operation of a fuse depends upon
the fault current and the time current characteristic (TCC) of
the fuse. The TCC has two curves: (i) the minimum melt
curve and (ii) the total clearing curve as shown in Fig. 1.
The minimum melting curve is a plot of the minimum time
required against the current required to melt the fuse link. The
2169-3536 (c) 2016 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.
total clearing curve is a plot of the longest melting time versus
the current required to melt the fuse link. When a fuse melts,
the fault current and its associated time would lie between the
two curves.
1000
100
10
Time [seconds]
Total Clearing
1
0.1
Minimum Melting
0.01
10 100 1000 10000 100000
Current [Amperes]
Fig. 1. Typical TCC curve of a fuse.
B. Overcurrent Protection Coordination
The process of selecting the appropriate settings for the
overcurrent relays such that whenever there is a fault, the
overcurrent relay operations ensure that the fault is isolated
as fast as possible while causing minimum service outage
to the customers is known as protection coordination. The
coordination arrangement is done such that whenever there
is a fault, all the protective devices sense the fault but the
device that is closest to the fault (main protective device)
operates first. If the main protective device fails to operate,
the next protective device upstream (backup protective device)
would operate after some time delay to isolate the fault. Thus,
upstream protective devices provide backup protection for the
downstream protective devices.
Following distribution system data is required for choosing
proper overcurrent relay settings [26]:
i) One-line diagram of the distribution system with locations
of protective devices and their time current characteristics
ii) Maximum expected load current through the protective
devices
iii) Maximum and minimum fault currents or short-circuit
levels at each protective device
The details of expected load currents and fault currents
through the protective devices are obtained from load flow
study and fault studies respectively. All the protective devices
need to be well coordinated with each other. In what follows,
the typical protective device coordinations that need to be
taken care of are briefed.
1) Fuse-Fuse Coordination: The coordination among the
fuses connected in series can be achieved by appropriate
selection of TCC curves of the fuses. When fuses are in series,
the fuse that is nearest to the fault can be referred to as main
fuse or protecting fuse. The next fuse upstream towards the
source of supply is backup fuse or protected fuse. For proper
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/ACCESS.2016.2589545, IEEE Access
Recloser slow curve multiplied by
safety factor
C’
C
Fuse TCC curves
A Fig. 3 shows the typical arrangement of circuit breakers,
Time, S
Isc max at recloser
Current, A
(a)
Recloser fast curve
multiplied by safety factor
A’
A
A S – Sectionalizer
Time, S
Fuse TCC curves
Isc max at recloser
Current, A
(b)
Fig. 2. Fuse-recloser coordination (a) when fuse is at the source side (b)
when fuse is at the load side.
coordination, the total clearing time of the main fuse should
not exceed 75% of the minimum melting time of the backup
fuse for a given current level [26]. This ensures that the main
fuse melts and isolates faulty line section before the backup
fuse is damaged. The 25% margin accounts for the variables
such as load current, ambient temperature and partial melting
of the fuse due to fault currents of short duration.
2) Recloser-Fuse Coordination: The coordination between
recloser and fuse depends upon their relative locations and
which device is intended to act as the backup device to the
other. When the fuse is at the source side and upstream from
the recloser, the fuse is the backup device to the recloser.
Then to prevent melting of the fuse, all the recloser operations
should be faster than the fuse. To account for cumulative
heating effect on the fuse due to multiple recloser operations,
the recloser slow curve is multiplied by a factor to make it
slower as shown in Fig. 2(a). Still, this slower curve also
should lie below the minimum melting curve of the fuse
making the recloser operations faster before the fuse gets
affected. When the fuse is at the load side and downstream
from the recloser, the fuse acts as the main protective device. In
this case, both the minimum melting and total clearing curves
of the fuse should lie in between the recloser fast and slow
curves as shown in Fig. 2(b).
3) Recloser-Recloser Coordination: When there are more
than one recloser present in a distribution system, coordination
is required among them. In such situations, utilities prefer to
2169-3536 (c) 2016 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.
accomplish the recloser-recloser coordination by selecting dif-
ferent types of reclosers. If same type of reclosers are chosen,
the coordination can be done by selection of different current
ratings or different operating sequences of the reclosers.
4) Relay Coordination with Reclosers and Sectionalizers:
reclosers, and sectionalizers in a distribution typical distribu-
tion system. The feeder breakers are usually equipped with
overcurrent protection relays. The operations of the relays
controlling the circuit breaker need to be time delayed to allow
the fault clearance by the downstream reclosers. In case of
permanent faults downstream of the recloser, the sectionalizers
isolate the faulted section at a current zero after preset number
of recloser operations. This operation of the sectionalizer
allows the service to resume to healthy feeder section after
recloser closes the circuit.
CB – Circuit Breaker
R – Recloser
CB F – Fuse
Substation
CB
S F
CB R
S F
F
Fig. 3. Typical arrangement of protective devices in a distribution system.
C. Impacts of PV and ESS on Distribution System Protection
Distribution system overcurrent protection schemes assume
single source of supply in the system and unidirectional current
flow from the source to the customer loads. The presence of
PV systems, ESS and their corresponding current injections
effectively creates multiple sources and bi-directional current
flows in the system. This interferes with the protection relaying
and causes protection issues [6]. Following impacts of PV
and ESS on the distribution system overcurrent protection are
reported in literature:
1) Reduction of Reach: In distribution systems, overcurrent
relays control the utility breaker and reclosers. The overcurrent
relays are set to detect faults within a certain distance along
the feeder, referred to as ‘reach’ of the relay. Reach of an
overcurrent relay is determined by its pickup current setting.
Usually, the pickup current setting is chosen above the maxi-
mum load current and below the minimum fault current in the
feeder so that the relay does not respond to the load currents
and be able to detect and respond to the fault currents [27]. The
fault current contributions from DG reduce the fault current
seen by the upstream protection devices. As a result, reach of
the upstream protection devices is reduced. For a sufficiently
high level of fault current injections by DG, as happens in the
case of large-scale PV, upstream protection devices may fail to
detect the fault. This phenomenon is called relay insensitivity.
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/ACCESS.2016.2589545, IEEE Access
Reduction of relay reach is illustrated in Fig. 4 where a large-
scale PV system is planned to be connected to a bus in the
middle of the feeder. The fault current contribution by the PV
for a fault downstream of the PV reduces the reach of the
relay upstream.
Vsg
Fault
R sympathetic tripping problems.
IR
PV
Fig. 4. Illustration of reduction of relay reach.
Fault analysis to calculate reduction of relay reach consider-
ing RBDG is presented in [19], [28]. Unlike the RBDG, which
acts as a voltage behind a reactance during the faults, IIDG
acts as a current source due to the current controllers present
in the IIDG controlling their power output [12]. Therefore, the
fault analysis of RBDG is not applicable to IIDG. Considering
this fact, analysis for the impact of IIDG on relay reach with
IIDG acting as a current source is reported in [29]–[31]. A
real-world example of a potential relay insensitivity problem in
a distribution feeder due to the integration of a large-scale PV
system is reported in [3], indicating the necessity of studying
relay insensitivity problem.
2) Sympathetic Tripping: Sympathetic tripping occurs
when the fault current from the PV causes relays on the healthy
feeder to pick up for faults on the adjacent feeders and results
in unnecessary outage of the healthy feeder [5], [6]. This is
illustrated in Fig. 5 where a large-scale PV system is connected
in Feeder 1 and a fault occurs on adjacent Feeder 2. During
the fault, if the PV system remains in the circuit as per LVRT
requirements, it feeds the fault back through the relay ‘R’ at
the substation. When the large-scale PV is of considerable size,
the fault current from the PV can go above the pickup setting
of the overcurrent relay at the substation causing unwanted
tripping of the Feeder 1 [32]–[34].
PV
Vsg Feeder 1
IfPV
R
Ifgrid
Fault
Feeder 2
Fig. 5. Illustration of sympathetic tripping.
While sympathetic tripping may not be likely a problem
in case of small-scale PV with low penetration levels, it is
a potential problem with high penetration of PV or large-
scale PV [35], [36]. The IIDG impact study results using an
2169-3536 (c) 2016 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.
actual utility distribution circuit model and protection device
settings reported in [20] indicates that sympathetic tripping
is one of the major issues utility planners face due to the
integration of IIDG. Solutions for preventing the protection
problems under study are reported in [37]–[39]. However,
implementing these over many distribution networks can be
a costly decision [20]. Therefore, there is a strong motivation
to perform comprehensive analysis into relay insensitivity and
3) Protection Coordination Issues: The bi-directional cur-
rent flows due to fault current injections from PV and ESS in-
terfere with the coordination of overcurrent protective devices
and could lead to misoperations. The exact impact typically
depends upon the size of the PV, relative locations of PV, ESS
and protective devices, protection settings and the distribution
system feeder impedance.
4) Reverse Power Flow: At higher PV penetration levels
and minimum loading condition in the feeder, the power flow
direction reverses at the substation when the PV generation
is more than the power consumption in the feeder. The
reverse power flow could possibly operate the reverse power
relays installed at the head of the feeder causing outage of
the feeder significantly affecting the system reliability [40].
The power reversal could also cause incorrect operation of
voltage regulators and capacitor controls as these controls
are not designed for reverse power flow. This increases the
number of operations and related equipment wear in addition
to power quality problems to the customers. The operational
problems due to reverse power flow include over voltages on
the distribution feeder and increased short-circuit levels.
5) Fuse Saving and Fuse Tripping: Majority of the faults
in the distribution systems are temporary in nature. They
occur momentarily for various reasons and are cleared by
themselves. Isolation of such faults by blowing the fuses cre-
ates unnecessary long duration service outages. To minimize
the service outages for temporary faults, utilities use ‘fuse
saving practice for coordination of breaker or recloser with
the downstream fuses. In this practice, the breaker or recloser
operates on fast curve before the fuse gets affected as was
shown in Fig. 2(b). This saves the fuse in case of temporary
faults. For permanent faults, the fuse blows when the breaker
or recloser operates on slow curve after preset number of fast
operations. The faster curve of breaker or recloser can also
be replaced with instantaneous overcurrent elements. In that
case, the instantaneous overcurrent element trips the breaker
or recloser before the fuse melts and is blocked temporarily.
If the fault is permanent, the fuse blows as the breaker or
recloser operation is delayed based on their slow curve. The
fuse saving practice leads to increased number of momentary
service interruptions as the breaker or recloser trips the entire
line downstream for all faults. As such, the utilities follow fuse
tripping practice wherein the fuse is allowed to blow for all
faults. The fuse blowing practice produces fewer momentary
interruptions but more sustained interruptions.
The DG installations could affect the timing coordination
between breaker/recloser and the fuse due to the additional
fault current contribution from the DG. As a result, the fuse
could blow first or both the fuse and breaker/recloser could
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/ACCESS.2016.2589545, IEEE Access
operate at the same time [41].
IV. A NALYSIS OF R ELAY R EACH AND S YMPATHETIC
T RIPPING WITH PV AND ESS
In this section, the impact of large-scale PV and ESS on the
overcurrent relay reach and sympathetic tripping is analyzed.
Battery energy storage system is assumed for ESS.
A. Reduction of Relay Reach
Consider the distribution circuit shown in Fig. 6 where
a large-scale PV system is planned to be connected to a
bus in the middle of the feeder. The upstream transmission
system of the substation including the substation transformer
is represented by a Thevenin equivalent circuit with impedance
Zeq . Phase overcurrent relay ‘R’ is installed at the substation.
Since the faults farther from the substation cause minimum
fault currents which are more challenging to detect, a remote
end fault with resistance Rf is assumed for conservative
study. The impedances of the feeder sections upstream and
downstream of the PV location are represented by ZLine1 and
ZLine2 , respectively.
Vsg
Zeq
R ZLine1
IR,no PV
PV
Fig. 6. Distribution system with fault.
1) Analysis for symmetrical faults: Consider the fault in
Fig. 6 is a three-phase fault. Fig. 7(a) shows the equivalent
circuit of the system when the PV is not connected. Using
Kirchoff’s voltage law (KVL), the fault current seen by the
relay is given by:
Vsg
IR,noP V =
(Zeq + ZLine1 + ZLine2 + Rf )
where Vsg and IR,noP V are the line-to-ground pre-fault volt-
age at the substation bus and fault current through the relay
without PV, repectively.
When the large-scale PV system is connected injecting a
fault current IPf V during the three-phase fault, the equivalent
circuit of the distribution system is shown in Fig. 7(b).
Then the fault current through the relay with the large-scale
PV connected can be obtained by applying superposition in
Fig. 7(b) as
(ZLine2 + Rf )
IR,P V = IR,noP V −
(Zeq + ZLine1 + ZLine2 + Rf ) P V
Equation (2) shows that the fault current through the relay
is reduced due to the fault current contribution from the PV
during a three-phase fault. This current seen by the relay is
also influenced by the available short-circuit capacity at the
2169-3536 (c) 2016 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.
ZLine1 ZLine2
IR,no PV
Zeq
Rf
Vsg
(a)
ZLine1 ZLine2
IR,PV
Zeq
IfPV Rf
Vsg
(b)
Fault
Fig. 7. Equivalent circuit of the distribution system a) without PV and b)
ZLine2
with PV.
Rf
substation. The location of PV alters the fault current distri-
bution. During faults, the PV systems behave like constant
power sources and the PV inverters try to push the same
power even at the low terminal voltages by injecting higher
currents [31]. As such, a PV system located close to the fault
can inject a higher current due to its lower terminal voltage
compared to the PV system located far away from the fault
which has higher terminal voltage during the same fault for
a given PV power output. Conservatively, this current can be
considered as 2 times the inverter rated current [31]. When
the PV fault current contribution is fixed at this value, the
(1)
PV can be placed close to the substation with the fault at
the end of the feeder as per (2) so that ZLine2 is maximum
causing more reduction in the fault current seen by the relay
at the substation. In such a scenario, when the three-phase
fault resistance is sufficiently high, the fault current seen
by the relay goes below its phase overcurrent pickup value,
making the relay insensitive to the fault. Under this situation,
any additional fault current contribution from the PV would
further reduce the fault current seen by the relay and make it
insensitive to the faults with lesser fault resistance. In other
words, the large-scale PV reduces the reach of the overcurrent
If relay.
Considering the bus voltage phasor where the large-scale
(2)
PV/ESS is connected as reference and the steady-state fault
current contribution from the large-scale PV is twice the PV
rated current in phase with the bus voltage [9], [10], [20], the
fault current through the phase overcurrent relay from (2) can
be expressed in terms of the PV size as (3). Taking the PV/ESS
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/ACCESS.2016.2589545, IEEE Access

bus voltage as reference significantly reduces the complexity ZLine1 ZLine2

of analysis without introducing much error in the estimate. R
Although the analysis is done in this paper by assuming PV Zeq I R0 ,noPV
fault current contribution is 2 times its rated current (rule of
thumb for system studies), the analysis can be extended along Vsg
the same lines for other levels of PV fault current contributions
depending upon PV inverter technology.
ZLine1 ZLine2
 
2 × PV rating R
IR,P V = IR,noP V − Z  √ (3)
3 × PV bus voltage I R0 ,noPV 3Rf
Zeq
(ZLine2 + Rf )
where Z  =
(Zeq + ZLine1 + ZLine2 + Rf )
When the ESS is also present in the system, we need to Z0Line1 Z0Line2
R
consider its fault current contribution also [42] for assessing
the reduction in the upstream relay current. Since the ESS I R0 ,noPV
Z0eq
also uses power electronic interface similar to PV, its fault
current contribution may also be approximated as 2 times its
rated current for conservative analysis. Furthermore, the ESS
acts as an active load when it is in charging mode and as (a)
a generator when it is in discharging mode. For conservative
protection studies, the ESS need to be acting as a generator
ZLine1 ZLine2
to contribute fault current. Therefore, it is considered as R
operating in discharge mode with 100% discharge rate to
ensure maximum fault current contribution. This additional Zeq I R1, PV
fault current contribution from ESS results in further reduction f
I PV 1
in the relay reach. Then (2) can be modified as: Vsg

IR,P V +ESS = IR,noP V − Z (IPf V + f
IESS )
f
where IESS is the fault current contribution from the ESS. ZLine1 ZLine2

It is observed from Z  that the maximum reach reduction R

occurs when ZLine2 is maximum i.e., when the PV or ESS is I R 2, PV 3Rf
installed close to the substation and the fault is located at the Zeq f
I PV 2
remote end. Under this condition, if the grid impedance Zeq
is neglected as it is typically much smaller than the feeder
impedance, then Z  = 1. This implies that the reduction in Z0Line1 Z0Line2
phase overcurrent relay current is exactly equal to the fault R
current injection from the PV or ESS.
I R 0, PV
2) Analysis for unsymmetrical faults: When the fault at the Z0eq I PVf 0
remote end of the feeder is single-line to ground (SLG), the
analysis can be done using sequence networks. The sequence
network diagrams of the distribution system without PV and (b)
with PV during the fault are shown in Fig. 8(a) and Fig.
8(b), respectively. The relay ‘R’ at the substation is considered Fig. 8. Distribution system with SLG fault a) without PV and b) with PV.
as ground overcurrent relay for unsymmetrical fault analysis.
Noting that the positive and negative sequence impedances
of distribution lines are equal, corresponding impedances of 0
where ZT S = 2(ZLine1 + ZLine2 ) + (ZLine1 0
+ ZLine2 )+
feeder sections upstream and downstream of the PV location 0
(2Zeq + Zeq ) + 3Rf
are denoted as ZLine1 and ZLine2 respectively in the figure.
Zero sequence impedances of the feeder sections are denoted The fault current through the relay at the substation is three
0
by ZLine1 0
and ZLine2 . times the zero sequence current
When the large-scale PV is not connected, from Fig. 8(a), 3Vsg
IRN,noP V = (4)
the positive, negative and zero sequence currents through the ZT S
relay are all equal and are given by: where ‘N ’ denotes the neutral.
0 Vsg When the PV is connected, the sequence components of the
IR,noP V =
ZT S PV current injections during the SLG fault condition can be

2169-3536 (c) 2016 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/ACCESS.2016.2589545, IEEE Access

3Vsg IP V 0
IRN,P V = − [2ZLine2 + ZLine2 + 3Rf
ZT S ZT S
f
I PVc I PVc − 3(ZLine1 + Zeq )]

IRN,P V = IRN,noP V − Z  × IP V (8)

I PVa I f
2 I PVa 1
PVa
where Z  = 0
[2ZLine2 +ZLine2 +3Rf −3(ZLine1 +Zeq )].
ZT S
f
I PVb I PVb

In terms of PV rating, (8) may be expressed as

 
PV rating
Fig. 9. Fault current injections by PV systems for maximum unbalance during IRN,P V = IRN,noP V − Z  √ (9)
the SLG fault. 3 × PV bus voltage

Considering maximum PV unbalance, the maximum reduc-

computed as: tion in ground fault current seen by the relay due to the
[IPf V ]012 = [A]−1 [IPf V ]abc presence of PV is given by (8). It indicates that the ground fault
⎡ ⎤ current reduction depends on PV rated current (PV size), fault
1 1 1 resistance, PV location and available short-circuit capacity at
where [A] = ⎣1 a2 a ⎦ and a = 1∠120◦ . the substation. For the worst case scenario, PV is installed
1 a a2 close to the substation and SLG fault occurs at feeder’s remote
For SLG faults, it is expected that the ground overcurrent end resulting in maximum ground fault current reduction.
relay senses the ground fault current which is three times the
zero sequence current and trips. The zero sequence current From the parameters of Z  , it is observed that the maximum
through the relay can be computed by applying superposition reach reduction occurs when ZLine2 is maximum i.e., when
in Fig. 8(b) for each source Vsg , IPf V 1 , IPf V 2 , IPf V 0 and then the PV is installed close to the substation and the fault is at
adding the zero sequence currents resulting by each of these remote end. Under this condition, if Zeq is neglected, then
sources. The zero sequence current through the relay is: Z  = 1. This implies that the reduction in ground overcurrent
0 1  relay current is exactly equal to the rated PV current which is
IR,P V = Vsg + (Zeq + ZLine1 ) IPf V 1 less than PV fault current contribution. As such, the reduction
ZT S
in ground fault current seen by feeder head-end relay is less
+ (Zeq + ZLine1 ) IPf V 2 − (2ZLine1 + 2ZLine2 (5)

than the phase overcurrent relay current reduction in case of
0
+ZLine2 + 2Zeq + 3Rf IPf V 0 three-phase faults.

For the worst case scenario, the maximum unbalance is When the ESS is also present, analysis can be done along
considered in the large-scale PV fault current injections i.e., it the same lines and the ground fault current seen by the
is assumed that the PV is injecting 2 times the rated current upstream relay can be obtained from (8) as
(IP V ) in the faulted phase and rated currents in the healthy
phases as shown in Fig. 9. IRN,P V +ESS = IRN,noP V − Z  (IP V + IESS ) (10)
The sequence components of fault currents injected by the
PV for maximum unbalance is given by: For conservative estimation, (10) considers that both ESS
⎡ ⎤ ⎡ ⎤ and PV fault currents are in phase with each other. Then, the
2IP V a IP V ∠0◦
1
[IPf V ]012 = [A]−1 ⎣ IP V b ⎦ = ⎣4IP V ∠0◦ ⎦ (6) PV rating in (9) includes both PV and ESS ratings.
3
IP V c IP V ∠0◦
Sometimes the phase overcurrent relays alone are used for
From (5) and (6), the zero sequence current through the protection against all types of faults [43]. As such, we are
relay for maximum unbalance is given by: also interested in determining the reduction of fault current
Vsg IP V 0
seen by the phase overcurrent relay installed at the substation
IR0,P V = − [2ZLine2 + ZLine2 + 3Rf due to the ground fault current injections by the PV and ESS
ZT S 3ZT S (7)
− 3(ZLine1 + Zeq )] during remote end SLG fault. The fault current seen by phase
overcurrent relay in the presence of large-scale PV can be
The ground fault current seen by the relay is three times computed by applying superposition in Fig. 8(b) for each
the zero sequence current: source Vsg , IPf V 1 , IPf V 2 , IPf V 0 and by adding the resulting
currents to determine the current flowing through the relay
IRN,P V = 3IR0,P V in each sequence network IR1,P V , IR2,P V , IR0,P V . Then the

2169-3536 (c) 2016 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/ACCESS.2016.2589545, IEEE Access

fault current through the relay is

IR,P V = IR1,P V + IR2,P V + IR0,P V PV
PV

1  0 0
= IR,noP V − (−ZLine1 + 2ZLine2 + ZLine1 + ZLine2 Vsg Feeder 1
IfPV
ZT S Zeq
0 R
−Zeq + Zeq + 3Rf )(IPf V 1 + IPf V 2 ) + (2ZLine1 + 2ZLine2

Ifgrid
0 0 0
−2ZLine1 + ZLine2 + 2Zeq − 2Zeq + 3Rf )IPf V 0 Fault

(11)
Feeder 2
Rf
When the maximum unbalance is considered, from (6) and
(11) we can determine the maximum reduction in this fault
current as:
(a)
IR,P V = IR,noP V − Z  × IP V (12)
Zeq
1
where Z  = 0
[4ZLine2 ++ZLine1 0
2ZLine2 0
+Zeq +6Rf − R
ZT S
(ZLine1 + Zeq )]
Ifau lt,3φ IR=-IfPV
Ifgrid
The above equation indicates that the ground fault current
contribution of the large-scale PV during the SLG fault Rf IfPV
downstream causes reduction in the fault current seen by the Vsg
upstream phase overcurrent relay also. The above analysis can
be extended to include the impact of ESS also as
IR,P V +ESS = IR,noP V − Z  (IP V + IESS ) (13) (b)
In (13), fault current injection of ESS is considered as Fig. 10. Distribution system with three-phase fault a) without PV and b) with
in phase with the PV fault current injection for maximum PV.
reduction. In terms of the PV and ESS ratings, (13) can be
expressed as
  study. Considering the PV and ESS fault current contributions
 (PV + ESS) rating  are in phase for conservative estimation, the fault current seen
 
|IR,P V +ESS | = IR,noP V − Z × √
3 × PV bus voltage  by relay becomes |IR | = |IPf V + IESS
(14)
B. Sympathetic Tripping
To analyze the sympathetic tripping, consider the system
shown in Fig. 10(a) where a large-scale PV system is installed
in Feeder 1 and a three-phase fault is assumed on adjacent
Feeder 2. Its equivalent circuit is shown in Fig. 10(b). During
the fault, the PV system feeds the fault back through the relay
‘R’ at the substation.
1) Analysis for symmetrical faults: From Fig. 10(b), the
fault current through the relay in Feeder 1 is
IR = −IPf V
f
| and the PV rating in
(15) includes ESS rating also.
2) Analysis for unsymmetrical faults: In case of SLG faults,
the ground overcurrent relay of the feeder shall detect the
ground fault current and trip. The ground fault current or
residual current through the relay is the sum of the fault
currents being injected by the PV system in all the three phases
and can be expressed as:
IRN,P V = −(IPf V a + IPf V b + IPf V c ) = −3IPf V 0 (16)
From (6) and (16), the current seen by the ground overcur-
rent relay under maximum unbalance condition is |IRN,P V | =
|IP V |. In terms of PV rating, this can be expressed as
 
 PV rating 
It indicates that the current through the relay solely depends |IRN,P V | = |IP V | =  √  (17)
on the fault current from the large-scale PV system and flows 3 × PV bus voltage 
in the reverse direction. Therefore, as the PV size increases, When the ESS is also present, the total ground fault current
its fault current contribution also increases and eventually trips through the relay would be vectorial sum of both the ground
the relay when the fault current contribution from the PV goes fault currents from the PV and the ESS which is given by
above the phase overcurrent relay pickup current setting.
IRN,P V +ESS = − (IPf V a + IPf V b + IPf V c )
Considering that IPf V is twice the IP V for the conservative
f f f
study, the magnitude of fault current flowing through the relay − (IESSa + IESSb + IESSc ) (18)
can be expressed in terms of the PV rating as: =− 3(IPf V 0 + f
IESS0 )
 
 2 × PV rating 
|IR | = |IPf V | =  √  where IPf V 0 and IESS0
f
(15) are the zero sequence current injec-
3 × PV bus voltage  tions from PV and ESS respectively.
If the ESS is also present in the system, we need to consider For conservative estimation, the ESS fault current injection
its fault current contribution also for the sympathetic tripping would be in phase with PV fault current. Then the PV rating

2169-3536 (c) 2016 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/ACCESS.2016.2589545, IEEE Access
TABLE I
S EQUENCE IMPEDANCES OF DIFFERENT S ECTIONS OF THE UTILITY
DISTRIBUTION FEEDER
Feeder Section Sequence Impedance (Ω)
Zeq = 0.1016∠72.4◦
Substation
0 = 0.312∠74.1◦
Zeq
ZLine1 = 1.389∠55.4◦
Upstream of PV & ESS
0
ZLine1 = 2.193∠55.56◦
ZLine2 = 0.538∠59.45◦
Downstream of PV & ESS
0
ZLine2 = 1.678∠60.97◦
in (17) gives combined PV and ESS rating.
V. VALIDATION OF R ELAY R EACH AND S YMPATHETIC
T RIPPING USING U TILITY D ISTRIBUTION F EEDER M ODEL
The analysis presented in the previous section is validated
through the simulation studies on a utility distribution feeder
model in OpenDSS. The generator and storage objects avail-
able in the OpenDSS model library are used for simulating PV
and ESS respectively. A common approach to model inverters
for a fault study is to assume inverters behave like synchronous
generators during faults. The transient reactance Xd then
can be adjusted to obtain desired fault current output like a
PV inverter [44]. The generator model (model 7) available
in OpenDSS is useful for this purpose to approximate the
nominal behavior of the inverter based DG systems [45].
For simulating PV fault current behavior two times its rated
current, the ‘xdp’ parameter of the generator model (model 7)
is adjusted to 0.5.
A. System Description
For simulation studies, a detailed distribution circuit model
has been developed in OpenDSS using the data obtained
from a utility. The data comprises of transformer, line data,
load profiles, capacitors, regulator data etc. The system has
a 24 MVA transformer at the substation. The three-phase
and single-phase short-circuit MVA of transmission system
upstream of the substation transformer are 562 MVA and 93
MVA respectively. A 1.6 mi long 12.47 kV level primary
distribution feeder is supplied by this transformer. Maximum
loading in the feeder is 15.77 MW and the corresponding load
current is 730 A. Minimum loading is 6 MW. This detailed
distribution model is used for simulations.
Fig. 11 shows the single-line diagram of the distribu-
tion system used for theoretical calculations. The sequence
impedances of the system are obtained from the short-circuit
study which are given in the Table I.
Following are the circuit conditions considered for this
study:
a) The breaker installed at the feeder head is controlled by
both phase and ground overcurrent relays. The pickup current
setting for the phase overcurrent relay should be above the
maximum cold load pickup current in the feeder and below
the minimum fault current. Typically, a pickup value of 1.5 to
2169-3536 (c) 2016 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.
10
Vsg Bus 1 Bus 2 Bus 3
Zeq, Z0eq ZLine 1, Z0Line 1 ZLine 2, Z0Line 2
Fault
R
IR,PV+ESS Rf
IfPV IfESS
PV
PV ESS
PV
Fig. 11. Single-line diagram of utility distribution feeder with fault.
3 times the maximum steady-state load current is used [46].
In this study, pickup current of 1500 A is used for the phase
overcurrent relay which is 2.05 p.u. of maximum steady-state
load current in the circuit (730 A). For ground overcurrent
relays, pickup current need to be above the maximum expected
unbalance current in the circuit. For most feeders, the ground
relays are set with a pickup range from 25% to 50% of the
phase relay pickup [46]. In this study, 750 A is chosen as
pickup current setting for the ground overcurrent relay which
is 0.5 p.u. of the phase relay pickup setting.
b) The PV and ESS are connected to a bus approximately 0.83
mi away from the substation (Bus 2) and the fault is simulated
at the remote bus (Bus 3) as shown in Fig. 11.
c) Although loading condition does not have much impact
on short-circuit studies, more current from PV fault current
injection is available when the load is minimum. Therefore,
minimum loading condition is considered for conservative
study.
B. Reduction of Relay Reach
The utility distribution feeder model is simulated initially
with a bolted three-phase fault at bus 3 without the large-scale
PV and ESS in the system. The fault resistance is increased
and the phase-A current through the relay at the substation
is monitored. Next, a large-scale PV system is connected to
bus 2 and the study is repeated. The fault current profiles seen
by the phase overcurrent relay for varied fault resistance as
the PV size is increased from 1 MW to 8 MW are shown in
Fig. 12. It is observed that when the PV is not connected, the
relay at the substation is insensitive to the three-phase faults
with Rf > 3.8 Ω. In other words, the phase overcurrent relay
cannot detect the three-phase remote end faults having fault
resistance higher than 3.8 Ω. The relay is blind to the fact that
a fault is present in the system. As the PV size is increased, the
corresponding PV fault current contribution IPf V is increased
resulting in relay insensitivity at lower fault resistances. For
the large-scale PV size of 8 MW, the phase overcurrent relay is
insensitive to the remote end three-phase fault with resistance
above 2.5 Ω. This means, if the relay at the substation is set
to detect a 3.8 Ω fault resistance considering no PV in the
system, the relay fails to perform its intended operation due
to the presence of PV.
The simulation results can be compared with theoretical
relay currents predicted using (2) to validate the analysis.
The pre-fault voltage of bus 3 is obtained from the load
flow program as Vsg = 7279∠ − 33.7◦ V. Using the system
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/ACCESS.2016.2589545, IEEE Access

11

Phase-A fault current seen by relay (A)

4000

3000
No PV
No PV
PV = 1 MW PV = 1 MW &
PV = 1 MW
2000 ESS = 3 MW
Ip = 1500 A
Ip = 1500 A
1000 IfPV = 63 A
Rf = 2.5 Ω Rf = 3.8 Ω Rf = 3.1 Ω Rf = 3.6 Ω IfESS = 179.6 A
PV = 8 MW
0
0 2 4 6 8 10
Fault resistance Rf (Ω)

Fig. 12. Fault current seen by phase overcurrent relay with large-scale PV
and varied remote end three-phase fault resistance using circuit simulations.
Fig. 14. Fault current seen by phase overcurrent relay installed at the
substation for various PV ratings and three-phase fault resistance.

in the previous case. When the 1 MW PV is connected, the

No PV
PV = 1 MW
Ip = 1500 A
Rf = 2.5 Ω Rf = 3.5 Ω
PV = 8 MW
Fig. 13. Predicted current seen by relay for various PV ratings and Rf using
(19).
parameters given in Table I, (2) can be written in terms of the
PV rating as
7279∠ − 33.7◦
IR,P V =
(2.024∠57.32◦ + Rf )
  (19)
(0.538∠59.45◦ + Rf ) 2 × PV rating
− √
(2.024∠57.32◦ + Rf ) 3 × 12.47 kV
Fig. 13 shows the predicted fault currents seen by phase
overcurrent relay at the substation using (19) for various
PV ratings and Rf . The theoretical predictions shown in
Fig. 13 closely match with the results obtained using circuit
simulations shown in the Fig. 12, confirming the validity of the
analysis. In particular, the PV size to cause relay insensitivity
for a remote end three-phase fault resistance of Rf = 2.5 Ω
is 8 MW in both the figures.
The impact of ESS on the fault current seen by the upstream
phase overcurrent relay is studied by connecting a three-phase
ESS of rating 3 MW to bus 2 along with the large-scale PV
of rating 1 MW. The variation of fault currents seen by the
phase overcurrent relay when there is no PV, when the PV is
connected to bus 2, and when both PV and ESS are connected
to bus 2 is shown in Fig. 14. It is observed that when the
large-scale PV is not connected, the phase overcurrent relay
is insensitive to the three-phase faults with Rf > 3.8 Ω as
relay is insensitive to the three-phase faults with Rf > 3.6 Ω
due to the fault current injection of IPf V = 63 A by the
PV. And when the 3 MW ESS is also connected, the relay
insensitivity occurs at further lower value of Rf > 3.1 Ω due
f
to the additional fault current contribution of IESS = 179.6 A
by the ESS. The results show that the effect of ESS is similar
to that of the large-scale PV for protection studies. As such,
the ESS of higher ratings can reduce the overcurrent seen by
the upstream phase overcurrent relays and can interfere with
the feeder overcurrent protection.
The impact of ground fault current contribution from the
large-scale PV on the fault current seen by the ground overcur-
rent relay can be predicted using (9) for the system parameters
in Table I. The predicted ground overcurrent results when the
large-scale PV ratings are 1 MW and 8 MW are shown in
Fig. 15. For ease of comparison, the ground overcurrent results
obtained from the SLG fault simulation using the detailed
distribution system model are also shown on the same figure.
From both theoretical predictions and simulation results, it is
observed that the reduction in the ground fault current seen
by the upstream relay increases as the large-scale PV size
increases. Note that (9) gives most conservative predictions
by considering maximum unbalance in the PV fault current
injection. It is observed that the error between the predicted
and simulated fault current results in the region of interest
where the relay insenstivity of ground overcurrent relay occurs
that is at Rf = 7 Ω is very less which validates the analysis.
When the ESS is also connected, its fault current contribu-
tion causes more reduction in the fault currents seen by the
phase and ground overcurrent relays. Fig. 16 shows the fault
current seen by phase overcurrent relay at the substation when
there is no PV, when 1 MW PV connected to bus 2 (three
single-phase PV rated 333 kW, unity power factor each), and
when both 1 MW PV and 3 MW ESS (three single-phase ESS
rated 1 MW each). These results are obtained from distribution
circuit simulations. The results confirm that both large-scale
PV and ESS cause reduction in the fault currents seen by
phase overcurrent relays located upstream for the SLG fault

2169-3536 (c) 2016 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/ACCESS.2016.2589545, IEEE Access

12

Ground fault current seen by relay (A) 3000

Predicted
No PV
2500 Simulated
8 MW PV Error
2000
1500
1000 Ipn = 750 A Ip = 1500 A

500 Rf = 7 Ω
0 0 1 2 3 4 5 6 7 8 9 10
Fault resistance Rf (Ω)

Fig. 15. Comparison of predicted ground fault current seen by relay using
(8) and simulated values without PV and with 8 MW PV for varied Rf .
Fig. 17. Locational impact of large-scale PV systems.
Phase-A fault current seen by relay (A)

PV
3000
Vsg
Zeq IfPV
2500 No PV R
IfESS
2000 1 MW PV, No ESS
ESS
PV
1500 I = 1500 A
p

1000
1 MW PV, 3 MW ESS
500 Fault
0
0 1 2 3 4 5 6 7 8 9 10 Adjacent feeder
Fault resistance Rf (Ω)

Fig. 18. Utility distribution system with adjacent feeder.

Fig. 16. Simulation results of impact of PV and ESS on the fault current
seen by the phase overcurrent relay.
from 1 to 24, that is total size of large-scale PV systems
is varied from 500 kW to 12000 kW. Twenty scenarios are
downstream of the PV and ESS. Reduction in the ground
considered for each 500 kW step increment of PV size, with
fault current is observed in case of SLG fault. However, the
the locations of the PVs varied stochastically among the four
reduction in the ground fault current due to the presence of
chosen PV locations [18] in each scenario. A three-phase
PV and ESS is less as compared to the reduction in the phase
fault with resistance Rf = 2.5 Ω is simulated at the end of
overcurrent due to the reasons discussed in Section II.
the feeder at bus 3 and the fault current at the substation is
The analysis provided in Section IV considers a single large-
recorded for each scenario. The results are shown in Fig. 17 as
scale PV connected at middle of the feeder. The large-scale PV
a scatter plot wherein each point correspond to one scenario.
systems behave like current sources, particularly during fault
It is observed that for a three-phase fault with resistance
conditions. They inject fault currents of magnitude determined
Rf = 2.5 Ω, the feeder can host about 8 MW of large-scale PV
by their inverters into the system. As such, the locations
with relay insensitivity as impact criteria. As can be seen in the
of the large-scale PV systems generally do not have much
figure, the impact of different locations for a given step size is
influence on the relay insensitivity. Therefore, a set of few
less, therefore analysis for the conservative PV location (that
large-scale PV systems can be approximated as a single large-
is close to substation) would provide an approximate estimate
scale PV system of size totaling their individual ratings located
of the feeder accommodation limit.
anywhere on the primary feeder. As mentioned earlier, for
conservative protection study, the single equivalent large-scale
PV can be placed close to the substation and fault can be C. Sympathetic Tripping with PV and ESS
considered at the end of the feeder. Fig. 18 shows the single-line diagram of the utility distri-
To demonstrate the influence of PV locations, four locations bution feeder including an adjacent feeder. In the OpenDSS,
along the feeder are chosen for PV deployment. The locations the adjacent feeder is modeled as an aggregated load equal to
are at substation, mid-feeder and at two feeder ends. The the yearly demand of the main feeder which is 4726 kW.
PV systems are of three-phase and rated 500 kW each. The For sympathetic tripping simulation study, initially a large-
number of PV systems installed in the system is increased scale PV system is connected to the bus 1 and a bolted three-

2169-3536 (c) 2016 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/ACCESS.2016.2589545, IEEE Access
Phase-A fault current seen by relay (A)
2000
Ip = 1500 A
1500
IfPVa =Ifa,noESS
1000
500
00 2 4 6 8 10 12 14 16
Fig. 19. Fault current seen by phase overcurrent relay without and with ESS
as the PV size is increased.
phase fault is simulated at a bus on adjacent feeder. Both the
buses are closest to the substation on the respective feeders.
Fig. 19 shows the fault current through the phase overcurrent
relay as the large-scale PV size is increased from 0 MW
to 20 MW. The figure also shows the effect of ESS on the
relay fault current. It is observed that, when the ESS is not
connected, the fault current through the phase overcurrent
relay is increased as the PV size is increased and reached
the pickup value of 1500 A at 16.2 MW of PV size. The fault
current in the phase-A of the feeder at this point is noted as
f
Ia,noESS = IPf V a = 1500.2∠153.2◦ A. The large-scale PV
size for causing sympathetic tripping can also be estimated
from (15), i.e., 16.2 MW.
When the ESS is also connected, it would also inject fault
current more than its rated current during the fault conditions.
The effect of the fault current contribution from the ESS is
studied by connecting an ESS of size 3 MW to bus 1 along
with the large-scale PV. The simulation results are shown in
Fig. 19. When the ESS is also connected to the system along
with PV, it is observed that the sympathetic tripping occurs at
14.4 MW of large-scale PV rating which is less compared to
the case without ESS. This is due to the additional fault current
contribution from the ESS. At this point, total 17.4 MW of PV
and ESS present in the system. The fault current contributions
from the large-scale PV and the ESS in the phase-A are noted
as: IPf V a = 1333.4∠155.9◦ A and IESSa
f
= 321.03∠92.2◦ A
respectively. The fault current in the phase-A of the feeder is
Iaf = IPf V a + IESSa
f
= 1503.6∠ − 35.1◦ A. We can make a
conservative estimate of the combined PV and ESS rating for
the possibility of sympathetic tripping using (15) also which
is 16.2 MW. Note that the conservative estimate of the PV
and ESS rating for the likelihood of sympathetic tripping is
slightly lower than the simulated result. This is due to the fact
that the fault current injections from the PV and the ESS in
the simulation are out of phase.
The sympathetic tripping study of the utility feeder for
unsymmetrical faults is done by simulating an SLG fault on the
adjacent feeder as shown in Fig.18. The ground fault current
seen by the ground overcurrent relay is shown in Fig. 20
as the large-scale PV size is increased from 0 to 20 MW.
2169-3536 (c) 2016 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.
13
Ground fault current seen by relay (A)
14.4 MW 1000
14.8 MW
f 800 Ipn = 750 A
I a,ESS
16.2 MW IRN,PV+ESS
16.2 MW
600
IRN,PV
400
IfESSa IfESSn
200
18 20 00 2 4 6 8 10 12 14 16 18 20
PV rating (MW) PV rating (MW)
Fig. 20. Fault current seen by ground overcurrent relay as the PV size is
increased.
It is observed that as the PV size is increasing, its ground
fault current contribution is also increasing, going above the
ground overcurrent pickup value of 750 A at a PV size of
16.2 MW. At this point, the feeder currents in all the three
phases are Iaf = 1500.3∠ − 24.5◦ A, Ibf = 848.49∠40.4◦ A
and Icf = 889.76∠ − 88.7◦ A and the residual current seen
by the relay is IRN,P V = Iaf + Ibf + Icf = 753.83∠ − 22◦ A.
The large-scale PV rating to cause sympathetic tripping can
be estimated from (17), i.e., 16.2 MW.
Fig. 20 also shows the simulation results when three single-
phase ESS of 1 MW rating each are also connected to bus
1. It can be observed that the sympathetic tripping now
occurs at a large-scale PV size of 14.8 MW which is lower
than the previous case due to the additional fault current
contribution from the ESS. When the ground fault current
through the feeder goes above the pickup value, the ground
fault current contributions from the PV and the ESS are
IPf V n = 718∠ − 16.8◦ A and IESSn f
= 226.1∠ − 111.46◦
A respectively. The residual current seen by the relay is
IRN,P V +ESS = Iaf + Ibf + Icf = 751∠ − 39◦ A. The PV
and ESS combined size in the simulation for the sympathetic
tripping is 17.8 MW. The PV and ESS size conservative
estimate from (17) is 16.2 MW. The conservative estimate
is slightly lower than the simulated result because the ground
fault current injections by the PV and ESS are out of phase.
However, the deviation of estimated PV and ESS size from
the simulated size in this case is less than 7%.
VI. C ONCLUSIONS AND D ISCUSSION
In this paper, comprehensive analysis of relay insensitivity
and sympathetic tripping problems in a distribution system is
provided with an aim to conservatively estimate the large-scale
PV and ESS accommodation limits of a utility distribution
feeder. Based on the analysis, the factors influencing the im-
pact of large-scale PV and ESS on the overcurrent protection
schemes are discussed. The proposed analysis-based approach
is scalable and thus can be applied to large distribution circuits
also. The analytical results are validated by simulation studies
using a detailed distribution circuit model of a utility in
OpenDSS.
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/ACCESS.2016.2589545, IEEE Access
Normally, in addition to the breaker at the substation,
the distribution circuits can have few reclosers placed along
the feeder. In such cases, the expected fault current can
be computed at each recloser using the proposed method.
Once we have the expected fault currents with large-scale
PV and ESS, the possibility of relay/recloser insensitivity can
be checked quickly by comparing with the pickup current
setting. Based on the expected fault currents and pickup
current settings of the breaker and the reclosers, the possibility
of miscoordination problems can also be studied. Note that
protection device miscoordination is not in the scope of this
study.
In case of sympathetic tripping, the reclosers play a limited
role as faults beyond the reclosers do not create sympathetic
tripping. When the faults are beyond the reclosers, usually
the reclosers perform reclosing operations when the fault is
detected in their protected zone creating momentary outages
or nuisance tripping. Relay in the faulted feeder clears the
fault before these reclosing operations finish and sympathetic
tripping happen. Sympathetic tripping is a severe problem
compared to nuisance tripping or momentary outages as it
results in permanent outage of a complete feeder.
ACKNOWLEDGMENT
The work was supported in part by Electric Power Research
Institute, the National Science Foundation Industry/University
Cooperative Research Center on Next Generation Photo-
voltaics (IIP−1134849 and IIP−1540028), and the Colombian
Department of Science, Technology and Innovation (Colcien-
cias) in cooperation with Universidad de los Andes.
R EFERENCES
[1] D. Feldman, G. L. Barbose, R. Margolis, and T. James, “Photovoltaic
(PV) pricing trends: Historical, recent, and near-term projections,”
National Renewable Energy Laboratory, Tech. Rep., November 2012.
[2] B. Tsuchida, S. Sergici, B. Mudge, W. Gorman, P. Fox-Penner,
and J. Schoene, “Comparative generation costs of utility-scale and
residential-scale PV in Xcel Energy Colorado’s service area,” Tech. Rep.,
2015.
[3] J. Bank, B. Mather, J. Keller, and M. Coddington, “High penetration
photovoltaic case study report,” National Renewable Energy Laboratory,
Tech. Rep., November 2013.
[4] P. Chiradeja and R. Ramakumar, “An approach to quantify the tech-
nical benefits of distributed generation,” IEEE Transactions on Energy
Conversion, vol. 19, no. 4, pp. 764–773, Dec 2004.
[5] R. A. R. Walling, R. Saint, S. Member, R. C. Dugan, J. Burke, and L. A.
Kojovic, “Summary of distributed resources impact on power delivery
systems,” vol. 23, no. 3, pp. 1636–1644, 2008.
[6] “IEEE guide for conducting distribution impact studies for distributed
resource interconnection,” IEEE Std 1547.7-2013, pp. 1–137, Feb 2014.
[7] H. Yazdanpanahi, W. Xu, and Y. W. Li, “A novel fault current control
scheme to reduce synchronous DG’s impact on protection coordination,”
IEEE Trans. on Power Delivery, vol. 29, no. 2, pp. 542–551, April 2014.
[8] H. Yazdanpanahi and W. Xu, “Contribution of induction-machine
distributed generators to fault current and assessing their impact on
overcurrent protection,” in 26th Annual IEEE Canadian Conference on
Electrical and Computer Engineering, May 2013, pp. 1–4.
[9] S. Bhattacharya, T. Saha, and M. J. Hossain, “Fault current contribution
from photovoltaic systems in residential power networks,” in IEEE
Australasian Universities Power Engg Conf., 2013, pp. 1–6.
[10] J. Keller and B. Kroposki, “Understanding fault characteristics of
inverter-based distributed energy resources,” National Renewable Energy
Laboratory, Tech. Rep., January 2010.
2169-3536 (c) 2016 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.
14
[11] T. Sidhu and D. Bejmert, “Short-circuit current contribution from large
scale PV power plant in the context of distribution power system protec-
tion performance,” in IET Conference on Renewable Power Generation,
Sept 2011, pp. 1–6.
[12] M. E. Baran and I. El-Markaby, “Fault analysis on distribution feeders
with distributed generators,” IEEE Transactions on Power Systems,
vol. 20, no. 4, pp. 1757–1764, Nov 2005.
[13] S. Essakiappan et al., “A utility scale battery energy storage system
for intermittency mitigation in multilevel medium voltage photovoltaic
system,” in Energy Conversion Congress and Exposition (ECCE), 2015
IEEE, Sept 2015, pp. 54–61.
[14] A. Barnes, J. Balda, A. Escobar-Mejia, and S. Geurin, “Placement of
energy storage coordinated with smart PV inverters,” in IEEE Innovative
Smart Grid Technologies, Jan 2012, pp. 1–7.
[15] G. Huff, “DOE global energy storage database.” Sandia National Labora-
tories (SNL-NM), Albuquerque, NM (United States), Tech. Rep., 2015.
[16] I. Gyuk et al., “Grid energy storage,” U.S. Department of Energy, Tech.
Rep., 2013.
[17] W. S. Moon, J. N. Won, J. S. Huh, and J. C. Kim, “A study on
the application of a superconducting fault current limiter for energy
storage protection in a power distribution system,” IEEE Transactions
on Applied Superconductivity, vol. 23, no. 3, p. 5603404, June 2013.
[18] A. Dubey, S. Santoso, and A. Maitra, “Understanding photovoltaic
hosting capacity of distribution circuits,” in IEEE Power Energy Society
General Meeting, July 2015, pp. 1–5.
[19] D. V. Tu, S. Chaitusaney, and A. Yokoyama, “Maximum-allowable
distributed generation considering fault ride-through requirement and
reach reduction of utility relay,” IEEE Transactions on Power Delivery,
vol. 29, no. 2, pp. 534–541, April 2014.
[20] J. Seuss, M. J. Reno, R. J. Broderick, and S. Grijalva, “Maximum PV
size limited by the impact to distribution protection,” in IEEE 42nd
Photovoltaic Specialist Conference, 2015, pp. 1–6.
[21] B. Mather et al., “Southern California Edison high-penetration photo-
voltaic project–year 1,” Tech. Rep., 2011.
[22] A. Hoke, R. B. J. Hambrick, and B. Kroposki, “Maximum photovoltaic
penetration levels on typical distribution feeders,” Tech. Rep., 2012.
[Online]. Available: http://www.nrel.gov/docs/fy12osti/55094.pdf
[23] M. J. Reno, K. Coogan, R. Broderick, and S. Grijalva, “Reduction of
distribution feeders for simplified PV impact studies,” in IEEE 39th
Photovoltaic Specialists Conference, June 2013, pp. 2337–2342.
[24] V. Scaini, “Whitepaper: Grid support stability for reliable, renewable
power,” Eaton Corporation, 2012.
[25] H. L. Willis, Power distribution planning reference book. CRC press,
2004.
[26] T. Gonen, Electric Power Distribution Engineering. CRC press, 2014.
[27] “Distribution system feeder overcurrent protection.” [Online]. Available:
http://www.geindustrial.com/publibrary
[28] M. H. Bollen and F. Hassan, Integration of distributed generation in the
power system. John wiley & sons, 2011, vol. 80.
[29] R. Jafari, M. Naderi, G. Gharehpetian, and N. Moaddabi, “Compensation
of DGs impact on overcurrent protection system of smart micro-grids,”
in 22nd Int. Conf. and Exhibition on Electricity Distribution, June 2013.
[30] M. Baran, H. Hooshyar, Z. Shen, and A. Huang, “Accommodating high
PV penetration on distribution feeders,” IEEE Transactions on Smart
Grid, vol. 3, no. 2, pp. 1039–1046, June 2012.
[31] H. Hooshyar, M. Baran, and L. Vanfretti, “Coordination assessment of
overcurrent relays in distribution feeders with high penetration of PV
systems,” in IEEE PowerTech, June 2013, pp. 1–6.
[32] V. A. Papaspiliotopoulos, V. A. Kleftakis, P. C. Kotsampopoulos, G. N.
Korres, and N. D. Hatziargyriou, “Hardware-in-the-loop simulation of
protection blinding and sympathetic tripping phenomena in modern
distribution grids,” ENERGY, vol. 2012, pp. 1–1.
[33] G. Bernardi and J. de Carvalho Filho, “High-resistance neutral grounding
in industrial systems and the ground fault protection,” in 2013 Brazilian
Power Electronics Conference, Oct 2013, pp. 1276–1282.
[34] H. Ravindra, M. Faruque, P. McLaren, K. Schoder, M. Steurer, and
R. Meeker, “Impact of PV on distribution protection system,” in North
American Power Symposium, Sept 2012, pp. 1–6.
[35] M. McGranaghan, T. Ortmeyer, D. Crudele, T. Key, J. Smith, and
P. Barker, “Renewable systems interconnection study: Advanced grid
planning and operations,” Sandia National Laboratories, 2008.
[36] R. Meeker et al., “High penetration solar PV deployment Sunshine
state solar grid initiative (SUNGRIN),” Florida State University, Tech.
Rep., May 2015. [Online]. Available: http://www.caps.fsu.edu/pdfs/
Final Report SUNGRIN 20160104.pdf
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/ACCESS.2016.2589545, IEEE Access

15

[37] D. Jones and J. Kumm, “Future distribution feeder protection using

directional overcurrent elements,” IEEE Transactions on Industry Ap-
plications, vol. 50, no. 2, pp. 1385–1390, March 2014.
[38] A. Sinclair, D. Finney, D. Martin, and P. Sharma, “Distance protection
in distribution systems: How it assists with integrating distributed
resources,” IEEE Transactions on Industry Applications, vol. 50, no. 3,
pp. 2186–2196, May 2014.
[39] P. Naik, M. Bahadornejad, N. Nair, and V. Vyatkin, “IEC 61850 based
smart distribution protection: Solutions for sympathetic tripping,” in
IEEE PES Innovative Smart Grid Tech. Asia, Nov 2011, pp. 1–7.
[40] R. Seguin et al., “High-penetration PV integration handbook for distri-
bution engineers,” National Renewable Energy Laboratory, Golden, CO
(United States), Tech. Rep., 2016.
[41] T. Seegers et al., “Impact of distributed resources on distribution relay
protection.”
[42] R. Hebner et al., “Energy storage on future electric ships,” ESRDC
report, March, 2008.
[43] C. R. Mason, The art and science of protective relaying. John Wiley
and Sons, 1967.
[44] R. J. Broderick, J. E. Quiroz, M. J. Reno, A. Ellis, J. Smith, and
R. Dugan, “Time series power flow analysis for distribution connected
PV generation,” Sandia National Laboratories, Tech. Rep., 2013.
[45] R. C. Dugan, “OpenDSS Manual.” [Online]. Available: http://smartgrid.
epri.com/SimulationTool.aspx
[46] “IEEE guide for protective relay applications to distribution lines,” IEEE
Std C37.230-2007, pp. 1–100, 2007.

2169-3536 (c) 2016 IEEE. Translations and content mining are permitted for academic research only. Personal use is also permitted, but republication/redistribution requires IEEE permission. See
View publication stats http://www.ieee.org/publications_standards/publications/rights/index.html for more information.