IEEE TRANSACTIONS ON INDUSTRY APPLICATIONS, VOL. 56, NO.

3, MAY/JUNE 2020 2393

Reverse Power Flow (RPF) Detection and Impact on

Protection Coordination of Distribution Systems
Juan Pablo Holguin , Graduate Student Member, IEEE, D. Celeita Rodriguez , Senior Member, IEEE,
and Gustavo Ramos , Senior Member, IEEE

Abstract—Due to the inclusion of distributed generation (DG) in The IEEE 1547 standard defines DG as a great quantity of en-
modern power systems, there are certain changes in the distribution ergy conversion and electric power generation connected in the
and transmission stage, either by impedance reflected by the lines, distribution level. In general, a direct classification of distributed
the increase of short-circuit currents, or the X/R relation, seen
from the different nodes on the grid. Such changes have a direct generators can be established based on their characteristics of
impact on protection coordination, which is the priority of study in direct coupling or by inverter.
this article, but it is important to note that another characteristic The generators of direct coupling are divided into two types of
affected by the DG is the power quality in terms of voltage sags, generators: the synchronous generators (e.g., hydro generator–
swells, and harmonics. One of the significant impacts due to the DG salient pole rotor machine) and asynchronous generators (e.g.,
is the reverse power flow (RPF), which generally occurs when the
generation of a distributed electric power plant exceeds the local wind generator with squirrel-cage rotor). On the other hand, the
load demand, causing power to flow in the opposite direction to second type of generators is connected through the inverter and
normal. This phenomenon can be produced by the intermittency is divided into static conversion (for example, solar panels) and
in the renewable energies, which depends on climatic factors. The rotary conversion (e.g., microturbines) [3].
consequences can be evidenced in the power system with volt- In the same mode, the standard [4] states that in the event
age peaks; therefore, the sensitivity and various parameters of
protection coordination are critically affected, which has further of a failure, the control in the DG system should act as
implications in the power quality. This work proposes a tool to following:
identify possible scenarios of RPFs and a feasible solution is intro- 1) A fault occurs.
duced. A comparative assessment is done to minimize the possibility 2) A fault-detection device in the area electrical power sys-
of an RPF in the IEEE 13 Nodes radial system with conclusive
tem (EPS) detects the fault.
results.
3) The fault-detection device operates a fault-isolating device
Index Terms—Distributed generation, distribution systems, in the area EPS.
protection coordination, reverse power flow (RPF). 4) The fault-isolating device opens, and the area EPS circuit
I. INTRODUCTION becomes islanded or open-phase.
5) The DG interconnection system detects the island, open-
HE increase in electricity demand, the high penetration of
T renewable sources, and the decentralization of the electric
systems have produced an increase in distributed generation
phase, or undervoltage condition.
6) The DG ceases to energize the area EPS.
Distributed generators coupled through an inverter do not have
(DG) connected to the grid. One of the advantages of DG is a significant contribution to the fault or directly affect the protec-
the improvement in the reliability of the network, more effi- tion system [3]. In general, the study is focused on the analysis
ciency, and a reduction of transmission costs and CO2 emissions. of the inclusion of directly coupled distributed generators and
However, given this context, there are changes in the operation, the validation of the effects produced by generators connected
control, configuration, and protection scheme of the network by means of an inverter in the system.
[1], [2]. According to [4] the disconnection or connection of DG
argue the can affect the operation of the total network. This
Manuscript received June 21, 2019; revised November 4, 2019 and January consequence is not the only problem to which it is exposed, there
13, 2020; accepted January 19, 2020. Date of publication January 28, 2020; date
of current version April 24, 2020. Paper 2019-PSPC-0744.R2, presented at the exist other phenomena such as reverse power fluxes (RPF), the
2019 IAS Annual Meeting, Baltimore, MD USA, and approved for publication increase or decrease of the short-circuit current, and the change
in the IEEE TRANSACTIONS ON INDUSTRY APPLICATIONS by the Power Systems in characteristics of impedances in the system.
Protection Committee of the IEEE Industry Applications Society. (Correspond-
ing author: David Celeita Rodriguez.) The reverse power flow (RPF) can be produced by two prin-
Juan Pablo Holguin and Gustavo Ramos are with the Department of Elec- cipal reasons: the first generally occurs when the production
trical and Computer Engineering, Universidad de los Andes, 111711 Bogota, of one or more distributed power plants exceeds the local load
Colombia (e-mail: jp.holguin10@uniandes.edu.co; gramos@uniandes.edu.co).
David Celeita Rodriguez is with the Laboratoire de Gnie Electrique demand or when the local demand is reduced to the point of
et Electronique de Paris, CentraleSupélec, 91192 Paris, France (e-mail: overgeneration in the system, causing power to flow in the
df.celeita10@uniandes.edu.co). opposite direction to normal, and voltage peaks occur [5]–[7],
Color versions of one or more of the figures in this article are available online
at https://ieeexplore.ieee.org. and the second is by a fault current that can flow in the direction
Digital Object Identifier 10.1109/TIA.2020.2969640 opposite to the normal current flow, given the power injection of

0093-9994 © 2020 IEEE. Personal use is permitted, but republication/redistribution requires IEEE permission.
See https://www.ieee.org/publications/rights/index.html for more information.

Authorized licensed use limited to: Auckland University of Technology. Downloaded on June 02,2020 at 21:06:00 UTC from IEEE Xplore. Restrictions apply.
2394 IEEE TRANSACTIONS ON INDUSTRY APPLICATIONS, VOL. 56, NO. 3, MAY/JUNE 2020

DG, which creates an abnormal performance in the protection II. RPF EFFECTS IN PROTECTION COORDINATION
system [8], [9].
This section aims to analyze the impact of DG connected to
As a solution to this problem, the directional overcurrent distribution and transmission networks. The analysis is focused
relay [10] is usually used. However, the network has a differ-
on the consequences of the protection coordination scheme and
ent short-circuit level depending on the level of penetration of
how it can be affected by RPFs, due to surplus production at the
DG and the model generator, as well as if these resources are DG plant, failures produced in the system, and instability in the
isolated or connected to the network. For that reason, the relay
network [20], [21]. Likewise, failure currents in a distribution
would trip in the presence of faults but will not find the correct
system are determined and examined, with and without the
direction for locating the electric failure and this criterium is inclusion of DG, evaluating the change that exists between the
not convenient for the system. Likewise, the implementation of
three cases, for the IEEE 13 nodes system [22].
differential relays in the protection scheme would not be able to
distinguish an overload from fault current, and also the price of
this implementation is highest. Finally, the ANSI 32 function of A. Three-Phase Fault at the Network Equivalent
power flow direction is only able to protect the generator against The magnitude of the current with DG tends to increase
an incoming network flow to isolate the central DG [11].  DG 
If the distributed generator directly feeds a load and also IK−node-fault + IK−node-fault
PowerGrid
> IK−node-fault
PowerGrid
. (1)
has connected a line toward the principal grid equivalent, it is
The current phasor for a node with the inclusion of DG is
necessary to define a correct selectivity criteria of the protection
defined by
system for different scenarios. In a possible scenario of a fault in
the line connected to the DG, the relay protection of the generator DG
IK−node-fault ∠ΘI1. (2)
can view the fault and make the trip for and isolate the generator
of the system. And without DG
This work consists of analyzing three situations of possi- DG
ble RPFs. The first focuses on the appearance of a fault in IK−node-fault ∠ΘI2. (3)
the system connected to the network equivalent, observing Since ΘI1 and ΘI2 are the angles of the current in the failed
the contribution of the DG to the failure. This case includes the node with and without inclusion of DG, it is assumed
analysis of voltage-level variations with and without DG. The
second case evaluates the failure that occurs in the distribution ΘI1 ≈ ΘI2. (4)
system, analyzing the contributions made by the DG and how
these contributions to the failure are modified [12]. The last case However, the voltage angles are modified, but not the voltage
examines the RPF at both distribution and transmission levels magnitude
facing an excess of production in the DG plant and a constant  DG+PowerGrid 
VK−node-fault ≈ VK−node-fault
PowerGrid
. (5)
demand in the distribution system [13]; further this article takes
in account the importance of modifing the X/R relation with and Therefore,
without DG.
The objective is to observe how the power supplied by the ∠ΘV 1 = ∠ΘV 2 . (6)
network equivalent is affected when connected in a distribution
system and what is the modification of the fault currents in In general, the voltage angle is modified in such a way
relation with the magnitude and the angles. Furthermore, note that it lags by at least 180◦ . This depends on the transformers
that the DG not only has positive effects in the grid, such as the downstream and their type of connection.
increase of the electric power system reliability and provision
of ancillary services, among others, but has other problems like B. Three-Phase Fault at the Distribution Level
that mentioned in the previous paragraph [12], [14]. Given this situation, (1) remains the same; however, since ΘI1
The proposed solution consists in indentifying the RPF by and ΘI2 are the angles of the current in the failed node with and
overgeneration, contextualizing, clarifying, and verifying the without inclusion of DG, the angle in this case does not change
magnitudes on the voltage nodes before and after the inclusion
of DG, based on the analysis of the best protection algorithm, ∠ΘI1 > ∠ΘI2. (7)
in this case, the ANSI 59 overvoltage relay protection [15],
[16], for checking the profile voltage in the near nodes and Therefore, the voltage tends to change in magnitude and angle.
deciding on disconnecting the GD to minimize the negative The latter tends to fall behind, depending directly on the node
impact of the DG on the system. The simulations and validation where the failure occurs
of the data are carried out in the ATP, Neplan, and OpenDSS  DG+PowerGrid 
VK−node-fault <= VK−node-fault
PowerGrid
(8)
software. Previous work has demonstrated consistent results
when modeling protection case studies in cosimulation [17], where
real-time hardware in the loop, and playback testing [18], [19].
∠ΘV 1 < ∠ΘV 2. (9)

Authorized licensed use limited to: Auckland University of Technology. Downloaded on June 02,2020 at 21:06:00 UTC from IEEE Xplore. Restrictions apply.
HOLGUIN et al.: RPF DETECTION AND IMPACT ON PROTECTION COORDINATION OF DISTRIBUTION SYSTEMS 2395

Fig. 2. Methodology diagram: test case scenarios and process step by step.

C. Single-Phase Fault at the Network Equivalent

For this case, the magnitude and angle remain approximately
the same for the voltages with or without inclusion of DG
 DG+PowerGrid 
VK−node-fault ≈ VK−node-fault
PowerGrid
(10)
where
∠ΘV 1 ≈ ∠ΘV 2. (11)
Since ΘV 1 and ΘV 2 are the voltage angles of the failed phase Fig. 1. Example: DG contribution under a fault in a radial system.
of the node with and without inclusion of DG, the angle is not
modified. However, the current changes both magnitude and
angle following (1). If the failure occurs in Phase A, then Phase E. Distributed Overgeneration
B and C behave in the following way, with regard to the current
Although DG reduces losses, it causes the voltage to increase
∠ΘIA1 ≈ ∠ΘIA2 in each of the nodes. This behavior can increase the reliability
∠ΘIB1 ≈ ∠ΘIB2 − 180◦ (12) in the nodes that are in the system queue, but it can also have
∠ΘIC1 ≈ ∠ΘIC2 + 180◦. negative effects due to direct overvoltages in these nodes that
Usually the angle of the voltage is modified in such a way that include DG.
it lags or leads by about 180◦ . n=i
  DG+PowerGrid  n=i
 PowerGrid 
Vi > Vi . (15)
D. Single-Phase Fault at the Distribution Level n=1 n=1

Since the system is unbalanced, the possible angular differ- The methodology is summarized in Fig. 2.
ence between phases in a fault situation cannot be characterized
correctly. This effect is because the fault is tied to the different III. CASE STUDY AND TEST SCENARIOS
magnitudes of the load connected to each of the phases. How- The case study is the standard circuit model IEEE 13 nodes
ever, the magnitudes can be characterized as detailed in (1) and test feeder as shown in Fig. 1. This 4.16-kV case is widely used
(10), and the nonfaulted phases to test common features of distribution analysis. The distribution
  system is characterized by being short and relatively highly
DG+PowerGrid
VK−non-fault-phase > VK−non-fault-phase
PowerGrid
. (13) loaded. There is a single-voltage regulator at the substation
DG and the system has both overhead and underground lines, shunt
IK−non-fault-phase +
(14) capacitors, an in-line transformer, and unbalanced loading.
IK−non-fault-phase > IK−phase-fault .
PowerGrid PowerGrid
This section shows the case scenarios and the performance
This stage haracterizes the modified angle which depends on of the proposed methodology. Case A (impact of DG connected
the power supplied by the DG to the grid and the model of the in node 680) only evaluates one DG connected in the system,
connected generator (constant PQ, PV). Case B (identification of RPFs) analyzes the impact with two

Authorized licensed use limited to: Auckland University of Technology. Downloaded on June 02,2020 at 21:06:00 UTC from IEEE Xplore. Restrictions apply.
2396 IEEE TRANSACTIONS ON INDUSTRY APPLICATIONS, VOL. 56, NO. 3, MAY/JUNE 2020

Fig. 3. Case study. (a) IEEE 13 nodes test feeder modified with only one DG connected. (b) IEEE 13 nodes test feeder with three different DGs connected.

different levels of DG penetration (see Fig. 1), and the last case 3) Case 3. Distributed Generation – 3200 kVA: In this case,
only takes into account the evaluation of one DG in node 680. optimal system operation is not possible through automatic
reconfiguration of taps. Since in the position (2, 0, 2), the
surges remain on the system. In this case, in node 680 Phase B,
where the voltage reaches a value of 1.0535 p.u., reaching
A. Impact of Distributed Generation Connected in Node 680 the standardized limits. It is possible to show that the reactive
1) Case 1. Distributed Generation – 400 kVA: The dis- power injected by the network equivalent does not manage
tributed generator simulated in Neplan corresponds to a pho- to control the voltage profiles in the system, specifically in
tovoltaic power plant with an installed capacity of 400 kVA, at Phase B.
a power factor of 0.9. It delivers an active power of 120 kW per
phase to the grid and reactive power of 58.1 kVAr per phase. The
results, in this case, show that although the taps (Configuration B. Identification of RPFs
7, 6, 7) of the main regulator of the transformer remain fixed, 1) RPF Due to Failure Produced in Network Equivalent That
no overvoltage is found in the system nodes. The lowest voltage Represents the Transmission System: (Note: For this scenario,
profile is still present in Phase C of node 684 with a value of only evaluate one DG connected in the 680 node.)
0.958 p.u. Also, the highest voltage profile of 1048 p.u. is present Three-phase fault in node 650: In the presence of a three-
in node 675. Regarding system losses, they are significantly phase fault to ground in the upper node connected to the network
reduced by 20.56 kW. equivalent of the system and no DG, a short-circuit current equal
2) Case 2. Distributed Generation – 1600 kVA: Fixed Taps to 29.972 kA is presented at 84.3◦ with a voltage magnitude
of Regulator RG-60: By maintaining the initial conditions of the close to zero, and a voltage angle equal to 183.6◦ [23]. Due to the
taps in the main transformer (Configuration 7, 6, 7), significant inclusion of DG in node 680, the fault current increases by 4 kA,
overvoltages were found. This performance is the consequence because it is fed by the network equivalent and the photovoltaic
of the DG being forced to work at a certain power and the power station. The angle of the fault current remains constant at
transformer to deliver specific voltages defined by the taps. It is 84.3◦ ; however, the voltage angle changes significantly and is
evident that the voltage limits are exceeded in Phase B because set to −83.7◦ .
it is the line with the lowest connected load. The nodes that show Single-Phase Fault in Node 650: In case of a single-phase fault
overvoltages are at 680, 675, 671, and 692. in the equivalent network system, the magnitudes of the voltages
Automatic Regulation Taps: Through the automatic control in the nonfaulted lines is 2.642 kV at an angle of 240◦ and 120◦ ,
of taps in the transformer, it is possible to eliminate the nodes respectively. While in the failed phase the voltage takes a lower
with overvoltages in the system; in this way, the regulator would magnitude at an angle of −11.16◦ , the short-circuit current at
go from initial taps (7, 6, 7) to (5, 3, 5). Due to the notorious node 650 takes a value of 29.972 kA at an angle of −84.29◦ , and
contribution of the power of the photovoltaic power station, the the current magnitudes of the other phases are set at magnitudes
network equivalent decreases its power delivered to the system; close to 0 kA with angles of 180◦ and −90◦ , respectively. With
this is notable in the decrease of the taps. the inclusion of the DG, the magnitude of the LN voltage in the

Authorized licensed use limited to: Auckland University of Technology. Downloaded on June 02,2020 at 21:06:00 UTC from IEEE Xplore. Restrictions apply.
HOLGUIN et al.: RPF DETECTION AND IMPACT ON PROTECTION COORDINATION OF DISTRIBUTION SYSTEMS 2397

current to increase due to the inclusion of the photovoltaic power

plant in failed phases (B and C) reaching a value of 29.975 kA
at 155.71◦ and 29.975 kA at 35.71◦ . Phase A has a small current
magnitude at an angle of −90◦ . The voltage in phase A assumes
a value of 2.642 kV at a 0◦ , while in phases B and C, the voltage
has a small magnitude at an angle of 180◦ and 135◦ .
2) RPF Produced by Failure in the Distribution System:
Three-Phase Fault in Node 632: Due to failure without DG, the
initial short-circuit current Ik takes a value close to 11.403 kA at
an angle of 67.39◦ . The line-to-ground voltage at node 632 takes
Fig. 4. Currents extracted from ATP model—Three-phase fault without DG.
a value of 0.123 kV at an angle of 139.89◦ . In the presence of
a fault with DG, the initial short-circuit current Ik is 15.810 kA
at −69.42◦ , the line-to-ground voltage at node 632 is equal to
0.099 kV at 150.85◦ .
Single-Phase Fault in Node 632: The single-phase failure
produces an extensive alteration of the voltages. For this case,
the magnitude of the voltage of the failed phase (A) is small at
an angle of 206.57◦ . While, on the other hand, the voltage of
Phase B reaches 2.898 kV at 220.26◦ and at Phase C, it is equal
to 2.731 kV at 133.49◦ . The current at node 632 for Phase A
reaches a magnitude of 5.679 kA at −70.54◦ . In Phases B and
C, magnitudes are negligible at 26.57◦ and 165.96◦ . The result
Fig. 5. Currents extracted from ATP model—Three-phase fault with DG.
due to the inclusion of DG causes the voltage in the failed phase
(A) almost become zero at an angle of −15.71◦ . Meanwhile, in
Phases B and C, the magnitude of the line-to-neutral voltages
are 3.189 kV at 217.32◦ and 3.114 kV at 137.75◦ , respectively.
The reason why the voltage is not equal is due to the unbalance
of the system. However, the B and C currents are relatively null,
while in the failed phase A, the current reaches a magnitude of
6.760 kA at an angle of −70.97◦ . It is possible to demonstrate
the increase in the magnitude of the fault current in the system
and the increase in the voltages of the nonfaulted phases. As
mentioned before, since the failure is fed by two sources, the
nominal power flow of the system is affected and it is evidenced
with the angles.
Fig. 6. Currents extracted from ATP model—two-phase fault with DG. Two-Phase Ground Fault in Node 632: In this case, the current
of the failed phases B and C are 8.032 kA at 180.74◦ and
7.714 kA at 31.31◦ , respectively. Phase A has a current of
irrelevant magnitude during the fault at a 90◦ . On the other hand,
nonfaulted phases is set at 2642 kV at an angle of 240◦ and 120◦ . the voltage in the nonfailed phase is equal to 2.819 kV at 1.66◦ .
On the other hand, in the failed phase, the voltage assumes an With the presence of DG in the system, the currents of the failed
irrelevant magnitude at an angle of −0.391◦ . phases B and C reach a magnitude of 12.38 kA at 183.921◦ and
The short-circuit current in the node takes a value of 29.974 kA from 11.777 kA to 25.04◦ , respectively. The current in Phase A
at an angle of −84.29◦ , and the current magnitudes of the other is equal to zero. On the other hand, the voltages in the systems
phases are established in magnitudes close to 0 kA with angles are equal to zero in Phases B and C, while in Phase A, the voltage
of 0◦ and 90◦ , respectively. The increase of the fault current is in the node is equal to 3.072 kV at an angle of 1.89◦ .
higher than 2 kA. Given these results at the presence of great variation, it
A Two-Phase Fault to Ground in Node 650: Due to the is necessary to implement other strategy of validation in the
occurrence of a biphasic failure in node 650, the short-circuit OpenDSS Software. This can be done by increasing the number
current in the failed phases (B and C) reaches a magnitude of nodes with DG. Then, a verification routine runs to check
close to 29.972 kA at 155.71◦ and 29 973.01 kA at 35.71◦ , short-circuit currents and the principal affections in all the nodes
respectively. Phase A has a minimal current magnitude at an of the system.
angle of 93.58◦ . The voltage in phase A assumes a value of To model the distributed generators in OpenDSS, there are
2.642 kV at 0◦ angle, while in phases A and B, the voltage has a two different options. First, model 1, which has a performance
small magnitude at an angle of 10.30◦ and −10.30◦ . A biphasic of active power and constant voltage (PV), represents the charac-
fault in the 650 node causes the magnitude of the short circuit teristics of directly coupled generators and model 2 is of constant

Authorized licensed use limited to: Auckland University of Technology. Downloaded on June 02,2020 at 21:06:00 UTC from IEEE Xplore. Restrictions apply.
2398 IEEE TRANSACTIONS ON INDUSTRY APPLICATIONS, VOL. 56, NO. 3, MAY/JUNE 2020

Fig. 7. Short-circuit currents with and without distributed generators.

TABLE I TABLE III

CHARACTERISTICS OF THE DG IMPLEMENTED IDENTIFICATION OF NODES WITH OVERVOLTAGE DUE TO OVERGENERATION

TABLE II
INCREASE IN THE SHORT-CIRCUIT CURRENT COMPARED TO THE CASE
BASE AND THE CASE MODIFIED WITH THE THREE DG

C. RPF Produced by Overgeneration in the

Distribution System
The simulation carried out seeks to reproduce a situation,
where the load connected in the distribution system can be fed
only by a distributed generator without requiring contribution
sources connected from the transmission level.
In this case, a generator of 4000 kVA is connected to a voltage
of 4.16 kV, with a power factor of 0.9 in node 680 of the IEEE
system 13 nodes. The obtained results show that the distributed
generator becomes the primary power source of the system.
Therefore, the power flows from the DG to the main node of the
equivalent network. This behavior [see Fig. 3(b)] demonstrates
that a possible overgeneration brings RPF consequences. In the
same way, the system exceeds the voltage limits admitted in at
least 9 of the 13 nodes of the system (see Table III).
active and reactive power (PQ), which simulates the actions of However, the main network equivalent which is connected to
generators coupled by an inverter. this system continues to provide power to the network. In Phase
The increase in the short-circuit current is quantified in the A, 64.3 kW is delivered, and in Phase B, 134 kW is delivered.
Table II. The principal nodes affected with the enhancing of the On the other hand, the losses in the system are reduced, from
DG are the nodes that have connected the distributed generators 114.38 kW in the system without DG to 56.42 kW with DG. On
and the adjacent nodes. the other hand, the automatic regulation of taps in the RG-60

Authorized licensed use limited to: Auckland University of Technology. Downloaded on June 02,2020 at 21:06:00 UTC from IEEE Xplore. Restrictions apply.
HOLGUIN et al.: RPF DETECTION AND IMPACT ON PROTECTION COORDINATION OF DISTRIBUTION SYSTEMS 2399

TABLE IV
IDENTIFICATION OF NODES WITH OVERVOLTAGE DUE TO OVERGENERATION
AND VOLTAGE REGULATION ACTIVATED

Fig. 10. Trip during RPF detection using an overvoltage protection strategy.

Fig. 8. IEEE 13 nodes test feeder with DG implemented in ATP.

Fig. 11. X/R relation with and without DG.

Fig. 11) [24]. In the event that an overvoltage is detected and

Fig. 9. ANSI 59 overvoltage relay location.
the relay is activated, it will disconnect the second generator.
The simulation of the protection and the system was carried out
in the ATPDraw software, because it allows the design of the
of the main system contributes to the reduction of overvoltages
protection system, and at the same time, it simulates the load
in the system, but it does not manage to prevent them from
flow.
occurring as shown in Table IV.
The current that the GD injected to the grid has a decrease,
with respect to the initial moment, due to the connection of the
IV. PROTECTION STRATEGY AND OTHER IMPORTANT FACTORS
second generator, that is another signal of RPF, but it does not
MODIFIED WITH THE PENETRATION OF DG have the criterium to make a trip because it could be taken as an
The modeling of the protection method proposed for the increase in the load.
IEEE 13 nodes system consists of the inclusion of the ANSI 59 Another aspect that is modified as well is the X/R relation.
overvoltage function in the model with DG. The modeled relay is This characteristic is affected in the grid, due to the increase of
connected to the observation point (node 680; see Fig. 9), where DG and does not depend on the type of distributed generator. In
two distributed generators of 1600 and 4000 kVA are located, to this case, it is possibly finding a relation of the power installed
assess the operation of the protection. and the decrease of the X/R factor (only in the adjacent nodes
The second generator starts to work in the 0.2 s of the simula- and that has connected generators).
tion. This setting is selected to check the effect overgeneration On the other hand, for a single-phase fault in node 671, the
has on the system and the activation of the ANSI 59 relay (see failure with DG has a displacement gigher than 120◦ that will

Authorized licensed use limited to: Auckland University of Technology. Downloaded on June 02,2020 at 21:06:00 UTC from IEEE Xplore. Restrictions apply.
2400 IEEE TRANSACTIONS ON INDUSTRY APPLICATIONS, VOL. 56, NO. 3, MAY/JUNE 2020

TABLE V Is important to synchronize the distributed generators that will

X/R RELATION OF THE IEEE 13 NODES MODIFIED MINUS X/R
RELATION OF THE IEEE 13 NODES BASE CASE
be connected to the grid, because this can produce a jumping
phase, and cause negative effects in the voltage and the current.
REFERENCES
[1] B. J. Brearley and R. R. Prabu, “A review on issues and approaches for mi-
crogrid protection,” Renewable Sustain. Energy Rev., vol. 67, pp. 988–997,
2017. [Online]. Available: http://www.sciencedirect.com/science/article/
pii/S1364032116305354
[2] J. D. Pico, D. Celeita, and G. Ramos, “Protection coordination analysis
under a real-time architecture for industrial distribution systems based on
the STD IEEE 242-2001,” IEEE Trans. Industry Appl., vol. 52, no. 4,
pp. 2826–2833, Jul. 2016.
[3] S. Kaddah, M. El-Saadawi, and D. El-Hassanin, “Influence of distributed
generation on distribution networks during faults,” Electric Power Com-
pon. Syst., vol. 43, pp. 1781–1792, 2015.
[4] “IEEE application guide for IEEE std 1547(tm), IEEE standard
for interconnecting distributed resources with electric power sys-
tems,” IEEE Std 1547.2-2008, Apr. 15, 2009, pp. 1–217, doi:
10.1109/IEEESTD.2008.4816078.
[5] H. Ravindra, M. O. Faruque, P. McLaren, K. Schoder, M. Steurer, and R.
Meeker, “Impact of PV on distribution protection system,” in North Amer.
Power Symp., Champaign, IL, USA, 2012, pp. 1–6.
[6] J. von Appen, M. Braun, T. Stetz, K. Diwold, and D. Geibel, “Time in the
sun: The challenge of high PV penetration in the German electric grid,”
IEEE Power Energy Mag., vol. 11, no. 2, pp. 55–64, Mar. 2013.
[7] M. Hasheminamin, V. G. Agelidis, V. Salehi, R. Teodorescu, and B.
Hredzak, “Index-based assessment of voltage rise and reverse power flow
phenomena in a distribution feeder under high PV penetration,” IEEE J.
erroneously trip the ANSI 67 Function relay and then operates Photovoltaics, vol. 5, no. 4, pp. 1158–1168, Jul. 2015.
backwards, as seen from node 692. [8] Working Group on Microgrid Protection Systems (WGC30),
“Microgrid protection systems,” IEEE PES Power Syst. Relaying
Control Committee, Aug. 2019, pp. 11–39. [Online]. Available:
V. CONCLUSION https://resourcecenter.ieee-pes.org/technical-publications/technical-
reports/PES_TP_TR71_PSRC_microgrid_082019.html
The inclusion of DG in power grids brings benefits such as [9] L. V. Strezoski, N. R. Vojnovic, V. C. Strezoski, P. M. Vidovic, M. D.
being a support to controlling the injection of reactive power Prica, and K. A. Loparo, “Modeling challenges and potential solutions
and controlling the voltage, the evident reduction of the losses for integration of emerging DERs in DMS applications: Power flow and
short-circuit analysis,” J. Modern Power Syst. Clean Energy, vol. 7, no. 6,
of active power in the system, and the increase of reliability pp. 1365–1384, Nov 2019. doi: 10.1007/s40565-018-0494-1.
in the system. However, it also brings with it consequences [10] B. Kroposki, R. Lasseter, T. Ise, S. Morozumi, S. Papathanassiou, and
such as increased short-circuit currents, voltage variations, and N. Hatziargyriou, “Making microgrids work,” IEEE Power Energy Mag.,
vol. 6, no. 3, pp. 40–53, May 2008.
frequency fluctuations. [11] J. P. Holguin, D. Celeita, and G. Ramos, “Reverse power flow (RPF)
For this reason, it is essential to plan and regulate the amount detection and impact on protection coordination of distribution systems,”
of DG that is implemented in the system to avoid cases such in Proc. IEEE Industry Appl. Soc. Annu. Meeting, 2019, pp. 1–6.
[12] K. N. Bangash, M. E. A. Farrag, and A. H. Osman, “Manage reverse power
as overgeneration, and the appearance of RPFs in the electricity flow and fault current level in LV network with high penetration of small
grid. scale solar and wind power generation,” in Proc. 53rd Int. Univ. Power
This study has illustrated that it can affect the distribution Eng. Conf., 2018, pp. 1–6.
[13] E. J. Aladesanmi and D. G. Dorrell, “Investigation and assessment of the
system, as evidenced in overvoltages, because the connected impacts of reverse power flow on power system network loading under
load was not considerable enough to consume all the power, and high penetration of wind energy,” in Proc. Southern African Universities
the capacity of the transformer connected to the system was not Power Eng. Conf./Robot. Mechatronics/Pattern Recognit. Assoc. South
Africa, 2019, pp. 493–498.
able to optimally control the voltage profiles of the IEEE 13 [14] S. Rahman, H. Aburub, M. Moghaddami, and A. I. Sarwat, “Reverse power
nodes system. flow protection in grid connected PV systems,” in SoutheastCon, 2018,
In the same way, the short-circuit analysis carried out allows pp. 1–5.
[15] “IEEE standard electrical power system device function numbers,
seeing the increase of the fault currents in the failed node, be- acronyms, and contact designations – redline,” IEEE Std C37.2-2008
cause two or more sources will feed the fault. In this case, the first (Revision of IEEE Std C37.2-1996) – Redline, Oct. 3, 2008, pp. 1–62.
is the network equivalent and, second, the distributed generator [16] A. Al-Riyami, K. Burt, G. Manhangwe, P. Pretlove, and S. Georgiopoulos,
“An investigation into alternatives to directional overcurrent protection on
that will contribute with the maximum power of short-circuit that grid transformers to improve the network capacity to accommodate reverse
can contribute. For this reason, the sensitivity of the protection power flow,” in Proc. 12th IET Int. Conf. Develop. Power Syst. Protection,
coordination will be affected by the inclusion of DG, and new Mar. 2014, pp. 1–6.
[17] D. Celeita, G. Ramos, and A. S. Meliopoulos, “Transmission line protec-
parameterization curves for overcurrent relays will have to be tive relay based on recursive least-square filters and weights analysis,” Int.
defined (ANSI 50/51 function). Rev. Modelling Simul., vol. 11, no. 5, pp. 277–287, 2018.
The ANSI 59 overvoltage protection has an essential role in [18] D. Celeita, A. S. Meliopoulos, G. Ramos, and L. Romero, “Dynamic
state estimation for double-end traveling wave arrival identification in
the inclusion of the DG because based on this, it is possible transmission lines,” Electric Power Syst. Res., vol. 170, pp. 138–149,
to configure the ugly limit of the amount of installed power, to 2019. [Online]. Available: http://www.sciencedirect.com/science/article/
avoid affecting the stability of the electrical system. pii/S0378779618304164

Authorized licensed use limited to: Auckland University of Technology. Downloaded on June 02,2020 at 21:06:00 UTC from IEEE Xplore. Restrictions apply.
HOLGUIN et al.: RPF DETECTION AND IMPACT ON PROTECTION COORDINATION OF DISTRIBUTION SYSTEMS 2401

[19] G. De Carne, G. Buticchi, Z. Zou, and M. Liserre, “Reverse power flow David Celeita Rodriguez (Senior Member, IEEE)
control in a ST_FED distribution grid,” IEEE Trans. Smart Grid, vol. 9, received the bachelor’s degree in electronic engineer-
no. 4, pp. 3811–3819, Jul. 2018. ing from the Universidad Distrital Francisco Jos de
[20] M. H. J. Bollen, Y. Yang, and F. Hassan, “Integration of distributed Caldas, Bogot, Colombia, in 2011, and the M.Sc. and
generation in the power system - A power quality approach,” in Proc. the Ph.D. degrees in electrical engineering from the
13th Int. Conf. Harmonics Qual. Power, Sep. 2008, pp. 1–8. Universidad de los Andes, Bogota, Colombia, in 2014
[21] D. I. Doukas, P. A. Gkaidatzis, A. S. Bouhouras, K. I. Sgouras, and D. P. and 2018, respectively.
Labridis, “On reverse power flow modelling in distribution grids,” in Proc. He worked as an Automation Engineer for a few
Mediterranean Conf. Power Gener. Transm. Distrib. Energy Convers., years in low- and medium-voltage applications, and
2016, pp. 1–6. was a Visiting Researcher with the Georgia Institute
[22] W. H. Kersting, “Radial distribution test feeders,” IEEE Trans. Power Syst., of Technology, Atlanta, Georgia, USA. He is cur-
vol. 6, no. 3, pp. 975–985, Aug. 1991. rently a Postdoctoral Researcher of protection algorithms for HV lines with
[23] P. M. Anderson, Analysis of Faulted Power Systems. Wiley-IEEE Press, the CentraleSuplec, France, and an Industry Partner in France. His research
1995, p. 540. [Online]. Available: https://www.sciencedirect.com/science/ interests include protective-relaying control, smart grids, advanced distribution
article/pii/S0378779615002734 automation, fault location, and real-time simulation.
[24] D. Celeita, M. Hernandez, G. Ramos, N. Penafiel, M. Rangel, and J. D.
Bernal, “Implementation of an educational real-time platform for relaying
automation on smart grids,” Electric Power Syst. Res., vol. 130, pp. 156–
166, 2016.

Gustavo Ramos (Senior Member, IEEE) received

the bachelor’s degree in electrical engineering from
Universidad Nacional, Manizales, Colombia, in 1997,
and the M.Sc. and the Ph.D. degrees in electrical en-
gineering from the Universidad de Los Andes, Bogot,
Juan Pablo Holguin (Graduate Student Member, Colombia, in 1999 and 2008, respectively.
IEEE) received the Electrical Engineering and Elec- He is currently an Associate Professor with the
tronic degree from the Universidad de los Andes, Department of Electrical Engineering, School of En-
Bogota, Colombia, in 2018. He is currently work- gineering, Universidad de Los Andes, Colombia,
ing toward the master’s degree at the Department of where he is involved in teaching courses on power
Electrical Engineering, School of Engineering, Uni- electronics, fundamentals of power systems, power
versidad de los Andes, Bogota, Colombia. quality, distribution, and industrial systems design. His research interests include
His research interests include distributed genera- software–hardware platform, power systems simulation, advanced distribution
tion, protection algorithms, modeling and simulation, automation, microgrids, and transients in grounding systems.
electric vehicles, and power quality.

Authorized licensed use limited to: Auckland University of Technology. Downloaded on June 02,2020 at 21:06:00 UTC from IEEE Xplore. Restrictions apply.