You are on page 1of 10

Electrical Power and Energy Systems 123 (2020) 106209

Contents lists available at ScienceDirect

Electrical Power and Energy Systems


journal homepage: www.elsevier.com/locate/ijepes

Overvoltage analysis and protection of lightning arresters in distribution T


systems with distributed generation

Jonathan Snodgrassa, , Le Xieb
a
Department of Electrical and Computer Engineering, University of Wisconsin, Madison, WI USA
b
Department of Electrical and Computer Engineering, Texas A&M University, College Station, Texas, USA

A R T I C LE I N FO A B S T R A C T

Keywords: The objective of this paper is to address an issue that arises from the increasing penetration of distributed energy
Distributed generation resources in power systems: the design and analysis of protective devices with more complex topology and
Microgrids power flow patterns. In particular, this paper investigates lightning arrester overloading and failure from fault-
Renewable energy resources induced overvoltages. Currently, the length of time from the fault inception until the lightning arrester’s failure
Lightning arresters
is unknown, forcing utility companies to assume the worst-case scenario and install more costly and complex
protection schemes than might otherwise be needed.
In this paper, the Thevenin Equivalent Impedance method is proposed to analyze a distributed generation
(DG) source’s effect on the transformer high side voltage. After examining the voltage transients and determining
the magnitude of the overvoltage, an optimal and cost-effective protective relaying strategy is developed and
implemented. To complete this study, various types of DG sources were modeled and simulated using two test
systems. Finally, the implementation of the suggested solution of intentional islanding operation of the dis-
tribution system is discussed. This solution allows the DG source to continue to supply a portion of the dis-
tribution system’s load, thereby increasing the reliability of the system.

1. Introduction fault that causes the distribution grid to inadvertently island will cause
at worst a line-to-line voltage drop across the line-to-ground connected
One of the major challenges facing the world today is that of global lightning arresters, thus overloading the arresters. Without an analysis
climate change and the resulting effects on the environment. In re- method to determine the possibility and severity of these overvoltages,
sponse, there have been efforts to decrease greenhouse gas emissions by utility companies are forced to assume the worst-case scenario, and
installing new renewable energy sources to eventually replace fossil- must implement a distribution system protection scheme that is po-
fuel based generation [1]. Many of these renewable energy sources, tentially much more complex and costlier than otherwise required.
such as photovoltaic panels or wind turbines, are integrated directly Thus, a tractable yet sufficiently accurate method to analyze the voltage
into distribution systems and are thus termed distributed generation profile of a distribution system with integrated DG is of great need.
(DG) resources. DG resources pose fundamental challenges in today’s This paper presents an analysis method that uses a Thevenin
control and protection framework of distribution systems, since the equivalent circuit to calculate the magnitude of the overvoltage seen by
traditional design philosophy is based on the premise of unidirectional the lightning arresters. By using the overvoltage magnitude to de-
power flow from transmission down to the distribution systems [2,3]. termine the time to failure for the lightning arresters, a more appro-
While there are numerous additional challenges to overcome when priate protective relaying scheme may be utilized. This prevents a
integrating DG into future or existing energy systems [4–6], the focus of distribution system operator (DSO) or utility company from im-
this paper is on updating distribution system protection schemes to plementing an overly conservative and costly protection scheme.
account for potential damage caused to surge arresters (also termed Additionally, if intentional islanding schemes are utilized, the DG
lightning arresters) by DG sources. source can continue to operate during a transmission-level fault, in-
If a distribution system with adequate distributed generation ca- creasing the reliability of the distribution grid by supplying a portion of
pacity is connected to the transmission or sub-transmission grid the distribution system load.
through a delta-wye substation transformer, a persistent line-to-ground A brief summary of the paper is provided. Section 2 explains how a


Corresponding author.
E-mail address: jsnodgrass2@wisc.edu (J. Snodgrass).

https://doi.org/10.1016/j.ijepes.2020.106209
Received 23 December 2019; Received in revised form 6 May 2020; Accepted 20 May 2020
0142-0615/ © 2020 Elsevier Ltd. All rights reserved.
J. Snodgrass and L. Xie Electrical Power and Energy Systems 123 (2020) 106209

DG source can cause a fault-induced overvoltage that can damage the 3. Review of existing analysis
lightning arresters. Section 3.1 summarizes work in the existing lit-
erature regarding overvoltages caused by DG in an ungrounded or 3.1. Existing analyses regarding fault-induced overvoltage damage to
delta-connected distribution system. Section 3.2 summarizes other litghtning arresters
analysis methods used to calculate distribution system voltages and
compares them to the proposed method of using Thevenin’s equivalent Several other sources have theorized about the line-to-line over-
circuit. Section 3.3 explores proposed modifications that distribution voltage created across the lightning arresters and documented the po-
system operators (DSOs) or utility companies can make to existing tential problems that this overvoltage condition would create. The
distribution system protection schemes to prevent overvoltage damage analysis done in [9] shows how the loss of a ground source in a wind
to lightning arresters. Section 4 presents an analysis method using turbine collector system can cause ferroresonance-induced overvoltage
Thevenin’s equivalent circuit to determine distribution system voltages, failure of lightning arresters. In [10], it is shown that an ungrounded
and includes a description of how the DG source, loads, and distribution wye system could experience up to a 1.73 per unit overvoltage on the
system were modeled. Section 5 shows simulations of distribution unfaulted phases for a single line-to-ground fault. The analysis in [11]
systems with DG and demonstrates the existence of a 1.73 per unit indicates that a 1.73 overvoltage can also occur from an open phase
overvoltage caused by a permanent line-to-ground fault as well as the fault. The authors in [12] introduce the idea of cogeneration (DG)
effectiveness of Thevenin’s equivalent circuit in determining the dis- connected to a distribution system, and present the same problem as
tribution system voltage. Finally, Section 6 shows the time to failure for addressed in this paper: a distribution-level DG source is interconnected
lightning arresters for given overvoltage magnitudes. Section 6 also to the transmission grid through a delta-wye transformer. According to
presents modifications to a distribution system protection scheme that [12], during a ground fault, the line-end breakers will open, but the DG
prevent damage to the lightning arresters, which are selected based on breaker will not open due to the delta connection preventing fault
the time to failure. current from flowing. Until the DG breaker trips on underfrequency or
undervoltage, the line section between the transformer and the line-end
breakers will operate as an ungrounded 3-wire system, and the un-
2. Lightning arrester operation and temporary overvoltage faulted phases “may reach 1.73P.U. This overvoltage may have no ef-
fect on gapped arresters, but the metal oxide arresters on this feeder
2.1. Problem overview will conduct current on the overvoltage. If the overvoltage is high, the
metal oxide arresters will fail unless the duration of the overvoltage is
An increasing amount of distributed generation is being integrated short.” [12]
into distributions systems. Though there are many challenges that such While all of these papers explain that neutral shift overvoltages can
installations pose, one such difficulty is the occurrence of a line-to-line occur in theory, none of them conducted simulations to illustrate or
(1.73 per unit) overvoltage on the high side of a delta-wye transformer prove this point. Thus, in Section 5.1 the effect of a line-to-ground fault
caused by inadvertent islanding of the distribution system. This issue is on an ungrounded (delta) system is examined. It is shown that the re-
illustrated in Fig. 1 and explained below. sulting neutral shift overvoltage does indeed result in an overvoltage of
In the case of a persistent line-to-ground fault (shown as an x in approximately 1.73 pu voltage on the unfaulted phases.
Fig. 1), the transmission level line-end breaker(s) will open within 1–3 Additionally, existing work suggests replacing semiconductor-based
cycles. Once the line-end breakers have opened, the DG source will be arresters with gapped arresters [11] or upgrading the arresters so they
connected to the transmission grid via an ungrounded delta trans- are rated for line-to-line voltage instead of the customary line-to-neu-
former. During this period until the DG breaker or distribution sub- tral voltage [8,10]. However, replacing or upgrading the arresters in-
station breaker trips, the lightning arresters on the non-faulted phases curs significant cost to the distribution system operator (DSO) or utility
will experience up to a line-to-line voltage drop instead of the cus- company, in addition to reducing the effectiveness of the arresters in
tomary line-to-neutral voltage drop, as explained in [7]. If the over- protecting against transient overvoltages and switching surges [10].
voltage is severe enough, and the DG source isn’t tripped within 1–2 Additionally, the authors in [9] suggest closing a ground switch 20 ms
cycles, then this overvoltage condition will cause semiconductor-based after the feeder breaker opens, thus causing the wind turbines in the
lightning arresters to fail [8]. As explained in [7], the delta transformer collector system to trip to avoid overvoltage damage to the lightning
creates a break in the zero-sequence transformer impedance, thus no arresters. Unfortunately, this solution also has the disadvantage of in-
fault current flows, making fault detection much more difficult since creased wear on the wind turbine breakers, as well as eliminating the
traditional overcurrent relays cannot be used. potential for intentional islanding to continue serving any connected
load.
Instead of following the previously suggested recommendations in
the existing literature, this paper proposes calculating the voltage
magnitude at the distribution system substation and appropriately
modifying the protection scheme to avoid damage to the lightning ar-
resters. Specifically, Section 4 presents a method using Thevenin’s
equivalent circuit to calculate the resulting voltage profile of the dis-
tribution system after the substation breaker opens. The resulting
magnitude of the substation voltage will determine the duration until
the failure of the lightning arresters, which informs the decision about
which modification should be implemented to the distribution system
protection scheme. This analysis and potential modifications are ex-
plored in Section 6.

3.2. Previously proposed voltage calculation methods

Fig. 1. One-line of distribution system showing transmisison- or subtransmis- There are currently three methods available for calculating the
sion-level breakers, 3 single-phase lightning arresters, fault location (marked distribution system voltages to determine the magnitude of the over-
with an x), delta-wye transformer, and connected DG source. voltage at the substation: unbalanced 3-phase distribution power flow,

2
J. Snodgrass and L. Xie Electrical Power and Energy Systems 123 (2020) 106209

calculations utilizing symmetrical components and general circuit In such situations without a complete model of the distribution system,
analysis techniques. However, the first two methods have difficulties in it is advantageous to utilize the Thevenin equivalent circuit since all
calculating the fault-induced overvoltage as explained below. The circuit analysis techniques require approximations and estimates for the
general circuit analysis techniques are utilized to determine the effec- distribution system model.
tiveness of the proposed method of utilizing Thevenin’s equivalent Additionally, using mesh current or node voltage circuit analysis
circuit. techniques to calculate the node voltages in a large unbalanced dis-
Power flow is a solution technique that utilizes numerical methods tribution system is computationally expensive, and thus difficult to
to solve equations describing the real and reactive power flow on a perform on-line. Thus, even if a full distribution system model is
network in terms of bus voltage magnitude and phase angle [13]. Un- available, the DSO could still utilize the Thevenin equivalent circuit to
balanced power flow models the mutual coupling between phases in a calculate the substation voltage using real-time load data to adaptively
power system in addition to unbalanced loads [14]. DG resources can update the distribution system protection scheme, as explained more in
be modeled as either PV busses [15] or both PV and PQ buses [16]. Section 6.
The benefits of 3-phase unbalanced power flow are the readily
available open-source software for modeling and solving the problem
[17,18]. The disadvantages of distribution power flow are that the so- 3.3. Current utility solutions
lution method requires a slack (or distributed slack) bus that provides
the necessary real and reactive power injections to balance load and Currently, protection engineers mitigate the risk of lightning ar-
losses with generation. Normally, the distribution substation, or point rester overloading and failure by installing a transmission-level un-
of interconnection (POI), is set as the reference or slack bus [16], but dervoltage monitoring relay with a direct transfer trip signal to the DG
once the transmission-level breakers open and the distribution system is source’s main breaker [23]. This transfer trip command could initiate
(at least temporarily) islanded, the transmission grid can no longer be from the remote substation relays, or a relay on the high side of the
used as the reference bus. If the DG source(s) can be re-dispatched to local substation transformer. While this approach is robust and reliable,
supply the entire load of the distribution system, the power flow can be it causes over-tripping of the DG source for every transmission line
solved. However, if the DG source(s) are inadequate to supply the entire fault, including transient faults. This means the DG source must wait a
load, unbalanced power flow can no longer be used. However, the set amount of time before reconnecting to the transmission grid and
proposed Thevenin equivalent circuit analysis technique overcomes this must go through the resynchronization process before connecting to the
difficulty by modeling constant power or current limited sources as grid. This results in lost revenue for the DG, and increased wear and
explained in Section 5.3 tear on the DG interconnection breaker. Also, there is currently no
Another approach used to determine the magnitude of the fault- analysis process to determine if the power output of the DG source is
induced overvoltage is the method of symmetrical components. sufficient to cause overvoltages capable of damaging the lightning ar-
Originally proposed by Fortescue in 1918 [19], symmetrical compo- resters, i.e. determining if the distribution system load reduces the
nents uses a linear transformation to decompose an unbalanced net- substation voltage enough to mitigate the problem. Therefore, the
work into three balanced sequence networks: the positive, negative and worst-case scenario is always assumed, and undervoltage relays with
zero sequences [20,21]. The advantage of sequence network approach direct transfer trip are implemented for all capacities of distribution
is the treatment of both the unbalanced power flow in the distribution level interconnected DG. Thus, a simple, but sufficiently accurate
network as well as the analysis of the transmission line fault. method of determining the overvoltage conditions resulting from a
However, the fundamental problem with using symmetrical com- persistent line-to-ground fault needs to be developed.
ponents to compute unbalanced distribution load flow is that trans-
forming phase impedances to sequence impedances requires im- 4. Analysis using thevinin’s equivalent circuit
pedances to be finite. This is normally not an issue for three-phase
transmission or distribution system primary feeders, but distribution Given the drawbacks and inadequacies of the previously proposed
systems often contain single- or two-phase laterals that directly supply methods to analyze the high-side transformer voltage to determine if
loads. The missing phases have zero-entries in their corresponding line the transmission line lightning arresters will be overloaded, a simpler,
admittance matrices, which results in infinite impedance in the corre- first-order method must be developed and tested. Thus, the distribution
sponding impedance matrices. Thus, the impedance matrices for single system is modeled using Thevenin’s equivalent impedance with the
and two-phase systems cannot be converted to sequence impedances. substation as the output node. This allows for calculation of the sub-
This prohibits analysis using symmetrical components from being used station voltages using ohm’s law and doesn’t require the use of sequence
to compute voltages in distribution system with single or two-phase networks, unbalanced load flow calculations, in-depth circuit analysis,
laterals. or numerical solvers.
Traditional circuit analysis methods include well-known techniques
such as mesh current or node voltage [22], which can be utilized to
calculate the voltages in a circuit model of a distribution system. Such 4.1. Thevenin’s equivalent circuit overview
methods can be utilized for single-phase or unbalanced three phase
systems, and for systems where the DG source capacity is below the As shown in Section 5.1, the voltage transients are negligible after
load capacity. Thus, circuit analysis techniques are utilized as a the transmission breakers open, thus allowing the authors to use a
benchmarking method to determine the accuracy of Thevenin’s steady state analysis method to determine the transformer high-side
equivalent circuit method in calculating the distribution system vol- voltage and determine if the lightning arresters would fail. Therefore,
tages. Additionally, if a complete distribution system model is available, the distribution system network was modeled using Thevenin’s
it can be utilized to compute the worst-case error when utilizing the equivalent impedance [22], with the DG source modeled as an ideal
Thevenin equivalent method for adaptive on-line calculations of fault- voltage source, the primary distribution feeder modeled using an
induced over-voltages, as explained in Section 6. equivalent network impedance, and the total distribution system load
However, the DSO often does not have a complete model of the represented using a constant impedance model of the equivalent load,
distribution system including the exact locations of loads on the feeder as shown below in Fig. 2. As explained in Section 4.3 below, the model
laterals, as explained in Section 4.3. Instead, approximations are made can be adapted to incorporate additional knowledge of the distribution
for the locations of the loads on the primary and secondary feeders, thus system from the distribution system operator, and a modified voltage
reducing the accuracy of the model and resulting voltage calculations. divider equation is used.

3
J. Snodgrass and L. Xie Electrical Power and Energy Systems 123 (2020) 106209

and primary method used to measure the equivalent circuit impedance


was injecting a test current and measuring the resulting voltage [22].
The method is as follows: first deactivate the sources by setting their
values to zero. This results in a short (zero voltage) for voltage sources,
and an open circuit (zero current) for current sources. Then, a test
current source is connected to the output node, and the voltage drop
from node one to node two of the current source is measured. Using
ohm’s law, the system impedance is calculated by dividing the mea-
sured voltage by the injected test current.

4.5. Voltage computation

Fig. 2. Thevenin Equivalent Network Model. Once the equivalent impedance is calculated, the DG voltage source
and equivalent distribution system load are used to compute the voltage
4.2. DC source modeling at the output node, i.e. the substation. Ohm’s law in the form of a
voltage divider equation is utilized to calculate the substation voltage.
Since the total load of the IEEE 13 node system is 3.74 kVA [24], a 5
kVA microturbine was connected to node 675. As explained in Section 5. Numerical examples and results
5.1, the synchronous reactance of the microturbine was used as the
source impedance since the machine would be operating in steady state First, the authors utilized the IEEE 4 node test feeder [26] to test the
after the transmission line breakers opened. Based on typical values authors’ hypothesis that a steady state analysis is sufficient to determine
from [25], the reactance X was set to be 0.17 pu and the X/R ratio is 25. the fault-induced overvoltage on the transmission system. Then, the
Because the total distribution system load was less than the max- authors employed the IEEE 13 node test feeder [26] to examine the
imum power output of the DG source, the source was modeled as an accuracy of the proposed Thevenin’s equivalent model compared to a
ideal voltage source. Although the authors did not fully model power full unbalanced distribution load flow. For the base case, there was a
electronically interfaced generation, if a power electronically interfaced maximum error of 0.02% between the phase voltages calculated from
DG source was used, the equivalent impedance of the DG source as well the Thevenin’s equivalent impedance and those simulated using a nu-
as power electronics and any connected transformers would be utilized. merical circuit analysis solver in Simulink. Additionally, the load was
If the power electronically interfaced DG source has a power rating increased, and different system topologies were modeled to increase the
below the total distribution system load, an iterative approach can be effective loading of the system, as explained below in Section 5.2.
used to estimate the voltage and current supplied by the DG, as ex-
plained in Section 5.3. 5.1. IEEE 4 node test feeder

4.3. Load modeling The first step in testing the validity of the Thevenin equivalent
circuit approach was to examine the transient voltage waveforms to
The system detail level employed using the Thevenin’s equivalent determine if they had significant impact on the peak voltage seen across
impedance can vary based on the modeling detail of the distribution the lightning arresters. The hypothesis is that voltage transients would
system model available to the distribution system operator. The authors be minimal since negligible fault current flows after the transmission
assumed that the distribution system operator (DSO) only knows the breakers are opened, and thus a steady-state analysis will be sufficient
impedances of the primary distribution feeder(s) and the total kVA of to determine the magnitude of the fault-induced overvoltages. To test
the distribution system load but does not know the exact placement of this hypothesis, the IEEE 4 node test feeder was modeled and simulated
the load in the distribution system model. Following the practice in the in the EMTP transient program, ATP [27]. This model is shown in
utility industry, the authors modeled the primary distribution feeders Fig. 3. The IEEE 4 node case [28] is a small distribution system con-
and placed the aggregate load at the substation, which was the furthest sisting of an infinite bus representing the transmission grid, a delta-wye
from the DG source. This mirrors the common practice of distribution transformer, and two short, nearly symmetrical distribution lines. The
protection engineers of modeling the aggregate distribution system load primary purpose of this distribution system is to test simulation models
at the end of the primary feeder for fault calculations. of transformers when there is an unbalanced load on the distribution
However, if the distribution system operator has a more detailed system. Thus, this system worked well to examine the transient voltages
model of the distribution system including impedances of the secondary caused by a line-to-ground fault on the high side of a delta-wye trans-
feeders, laterals and locations of the loads on the feeders or laterals, former, as well validate the line-to-line overvoltage on the lightning
then that information could be included in the model. The DG source arresters, as described in Section 2.
would be placed at the proposed location in the distribution system First, a base case was established for the balanced load specified in
model, and the feeder(s) and loads would be modeled according to the [20] by simulating a single phase-to-ground fault on the high side of the
knowledge of the distribution system operator. If a complete distribu- transformer. The voltage at the high side of the transformer is shown
tion system model is available, circuit analysis techniques such as node below in Fig. 4. To aid in viewing the results, the authors set the peak
voltage or mesh current could be utilized in addition to Thevenin’s voltage value of the DG source to be the nominal line-to-line voltage.
equivalent circuit for an offline calculation of the voltage at the sub- Thus, the pre-fault voltage seen across the lightning arresters (con-
station. nected line to neutral) is 24.83 kV, and the post fault voltage is 43 kV,
corresponding to the line-to-line voltage of the pre-fault condition. This
4.4. Equivalent system impedance validates the assertions from [10,11] that a bolted line-to-ground fault
on an ungrounded (delta) system results in an approximately 1.73 pu
The authors are modeling the case where the distribution system (line-to-line voltage) overvoltage to be seen on the non-faulted phases.
operator does not know the precise location of the loads in the dis- However, the overvoltage is only seen on the high side of the delta-wye
tribution system. Thus, the impedances of the primary feeders are uti- transformer so the DG source will not experience an overvoltage at its
lized, and the equivalent system load is placed at the substation. To terminals.
calculate the Thevenin equivalent impedance of the network, the first Also, the transient voltage spikes in Fig. 4 are noticeably

4
J. Snodgrass and L. Xie Electrical Power and Energy Systems 123 (2020) 106209

Fig. 3. ATP Simulation of IEEE 4 Node System Model.

insignificant, thus showing that a steady-state post fault voltage fully accurate individual load data. In the IEEE 13 node test feeder, both test
characterizes the behavior of the system. As seen below, there is no current injection and Voc/Isc methods were used to determine the
overshoot or significant ramping from the pre-fault to post-fault voltage equivalent impedance.
levels of the system, thus validating the authors’ hypothesis. This is Additionally, the system load was increased by connecting two of
most likely because a negligible amount of fault current flows in the the IEEE 13 node systems together in series by connecting the substa-
system, as explained in [7]. tion of a second system to node 675 of the first system, and the DG
source to node 675 of the second system. This is the “series” config-
uration below. The “parallel” configuration is achieved by adding a
5.2. IEEE 13 node test feeder copy of the IEEE 13 node system to the test distribution system with the
DG at node 675, and connecting substation of the two systems together.
After the transient and steady state voltages were examined using The “series parallel 1” system is the “series” system with an additional
the IEEE 4 node feeder, a larger system, the IEEE 13 node test feeder, IEEE 13 node system connected in parallel, i.e. by connecting the
was utilized to examine the effect of DG placement and load values on substations together. The “series parallel 2” configuration is similar,
the substation voltage. The characteristics of the IEEE 13 node feeder except there are two IEEE 13 node systems in parallel with the series
[29] are that the system is operated at 4.16 kV, the lines are short and configuration. These configurations were created to compare the effect
relatively highly loaded, and the system has unbalanced loads and of increasing the feeder impedance and composite system load with a
shunt capacitor banks. The Simulink model of the IEEE 13 node test Simulink simulation of a larger and more heavily loaded distribution
feeder that was included in Matlab 2015b is shown in Fig. 5. system.
The equivalent system impedance can be calculated by the utility Additionally, the load levels were modified as shown below in
company using known distribution line parameters and models, and Table 1. The different load levels were selected to demonstrate the
SCADA load data from the substations or smart meter data for more

Fig. 4. Voltage Transients for Single Line-to-Ground Fault.

5
J. Snodgrass and L. Xie Electrical Power and Energy Systems 123 (2020) 106209

Fig. 5. Simulink Simulation of the IEEE 13 Node Feeder System.

Table 1 modified configuration of the IEEE system as explained above in


Equivalent Substation Load for Modified Configurations of IEEE 13 Node Section 5.2. Notice that the primary feeder connecting the DG system
System. from node 675 to the substation primarily determines the equivalent
Configuration Phase A Phase B Phase C impedance as “seen” from node 675 to the “output” node of the sub-
station. Notice how the equivalent impedance nearly doubles from the
Base Case 678-240i 600-268i 723-301i base case to the “series” case which connects two IEEE 13 node systems
Parallel 1755-649i 1144-403i 1695-598i
in “series,” which highlights the contribution of the primary feeder to
Series 1357-480i 1200-537i 1446-602i
Series Parallel 1 3111-1129i 2344-940i 3141-1200i the equivalent impedance. The reason the impedance didn’t completely
Series Parallel 2 4866-1778i 3489-1344i 4836-1798i double is due to the connected load in the distribution system.
For each of the load scenarios in Table 1, the equivalent impedance
was calculated using a nominal system line-to-line voltage of 4.16 kV.
effectiveness of the Thevenin equivalent circuit in calculating the sub- V Z
The voltage divider equation, Vsub = Z DG+ Load was utilized to calcu-
Load ZFeeder
station voltages for a range of distribution system load levels. Since the late the substation voltage, Vsub for each load scenario. ZFeeder is the
DG is rated at 5kVA, the first 3 scenarios have a total loading level impedance of each configuration of the IEEE 13 bus system as listed in
below the rating of the DG. The last two load scenarios were created to Table 2. However, if less modeling detail is available for the distribu-
examine the distribution system voltages when the total load was tion system then the impedance of the feeder(s) and/or lateral(s) con-
higher than the DG rating. Finally, the results for the calculations and necting the DG to the distribution system can be used for ZFeeder .
simulations are summarized in the subsequent section. Finally, the resulting substation voltage for each configuration of
the IEEE 13 node feeder system is shown below in Table 3. The re-
5.3. Results and discussion sulting voltages are displayed for the phase with the highest per unit
voltage since it will result in the worst-case overloading of the lightning
Table 2 records the measured equivalent impedance of each
Table 3
Table 2 Calculated Substation Voltages for Each Modified IEEE 13 Node System
Measured Thevenin Equivalent Impedance of Modified Configurations of IEEE Configuration.
13 Node System.
Test Case Voltage using Voltage using Phase % Error
Configuration Phase A Phase B Phase C Configuration Thevenin’s Circuit Simulink

Base Case 0.2353 + 0.4431i 0.2251 + 0.4219i 0.223 + 0.4229i Base Case 0.9693 0.9694 B 0.02%
Parallel 0.2353 + 0.4431i 0.2251 + 0.4219i 0.223 + 0.4229i Parallel 0.958 0.951 B 0.78%
Series 0.473 + 0.9388i 0.4444 + 0.9114i 0.4412 + 0.901i Series 0.904 0.929 B 2.60%
Series Parallel 1 0.5052 + 0.7577i 0.4743 + 0.7886i 0.4755 + 0.7419i Series Parallel 1 0.838 0.889 B 5.74%
Series Parallel 2 0.4994 + 0.619i 0.4809 + 0.6869i 0.4761 + 0.6161i Series Parallel 2 0.786 0.844 B 6.92%

6
J. Snodgrass and L. Xie Electrical Power and Energy Systems 123 (2020) 106209

arresters on the high side of the transformer. Also, phase B also had the
worst calculation error of the 3 phases for all scenarios.
The calcuation method utilizing Thevenin’s equivalent circuit re-
sulted in lower voltages than the Simulink simulation. This is to be
expected, since placing the load at the substation and the DG source at
node 675 places the load and DG on “opposite” sides of the distribution
system. Thus, there is more current flowing through the main feeder at
the substation in the Thevenin model than there is in the full dis-
tribution system model. This reduces the substation voltage in the
Fig. 7. Thevenin Equivalent Network Model Using Current Source.
Thevenin equivalent analysis compared to the Simulink simulation.
If additional distribution system modeling detail is available, the
loads can be placed on their corresponding points along the main feeder be utilized until the rated current or temporary overcurrent limit of the
(s) to increase the calculation accuracy. Then, a series of voltage divider DG source is reached. At which point, the voltage source in Thevenin’s
equations, or the node voltage method, could be utilized to calculate equivalent circuit in Fig. 1 is replaced with a current source supplying
the substation voltage. However, as seen in table 3, the largest error is the DG’s rated current as shown in Fig. 7. For higher load levels, the
6.92%, which indicates that the Thevenin equivalent circuit method equivalent resistance decreases, consequently reducing the voltage at
provides a good first-order approximation to determine the distribution the substation calculated using Ohm’s law (V = IZ ). If there are a
system substation voltages. However, if the calculated distribution substantial number of constant current loads in the distribution system,
system voltage is close to overloading the lightning arresters during a such as motors, they can be modeled as current sources in series with
permanent line-to-ground fault, then either a more comprehensive the equivalent system load.
analysis method can be utilized, or a more conservative system pro-
tection scheme from Section 6.4 should be implemented. 6. Suggested solutions

5.4. Power limited renewable generation models As shown in previous sections, Thevenin’s equivalent circuit can be
utilized to calculate the voltage at the low side of the delta-wye
Once the test systems were established, a simple model for power transformer. This voltage is then multiplied by the 3 voltage increase
electronic interfaced DG sources was developed. Current industry caused by the delta connected transformer to determine the per unit
practice is to model such sources as current-limited (to 1.5 pu) voltage voltage overload at the high side of the transformer. After the high-side
sources to reflect the regulating effect of the power electronics. Thus, a voltage is determined, a mitigation strategy must be determined and
simple model of these sources is a constant power source, since the implemented. If the lightning arresters will not overload before the DG
power electronic interfaces can only supply a maximum amount of source trips on anti-islanding detection, no further actions must be
power to the system. taken. But if lightning arrester overloading and damage will occur, one
Since the distribution system loads in Simulink are modeled as of the following two strategies must be implemented: either replacing
constant impedance loads, a constant power source is created using the the lightning arresters with ones with higher maximum continuous
following approach: Using the measured current as feedback, a operating voltage (MCOV) ratings, or the distribution system breaker or
Simulink controlled voltage source is varied to supply constant power DG source must be tripped before the lightning arresters will be da-
to the system. A Simulink model of the proposed system is shown below maged, as explained in the following 3 sub-sections.
in Fig. 6. Since two quantities are fixed, the load impedance Z and the
power output of the source, there need to be two degrees of freedom to 6.1. Lightning arrester sizing considerations
satisfy Ohm’s law and the electrical power equation. Thus, these two
equations, P = IV and V = IZ are used to calculate the voltage and When sizing lightning arresters for distribution systems, several
current produced by the power-limited DG source. Then, these voltages considerations must be made, as indicated in [30–32]. Per [8], arresters
and currents can be used to determine the voltage drop across the are selected to have a duty cycle ranging from 0.72 to 1.13 per unit of
network and thus the voltage at the substation. the distribution system line-to-line voltage and 0.75 to 0.94 per unit of
The constant power model described above and shown in Fig. 6 can the transmission system line-to-line voltage. Since the MCOV rating is

Fig. 6. Simulink Constant Power Model of Power Electronically Interfaced Generation.

7
J. Snodgrass and L. Xie Electrical Power and Energy Systems 123 (2020) 106209

always lower than the duty cycle rating [32], this results in the typical line-to-line voltage of the distribution or transmission system, as sug-
distribution system arrester having an MCOV rating that is between gested in [8,31]. This is practical for a future distribution system with a
1.06 and 0.61 of the line-to-line conductor voltage, and transmission delta-wye transformer that is being designed with integrated DG; the
arresters with MCOV ratings between 0.80 and 0.59 of the line-to-line lightning arresters should be selected with an MCOV of at least the line-
voltage. For the most conservative analysis, the lowest rating in this to-line system voltage. However, for existing systems with numerous
range can be utilized, but if the DSO or utility company knows the arresters installed in the distribution or transmission system, this option
MCOV ratings of their arresters, they can utilize the exact MCOV ratings is not cost effective compared with the recommendations in the fol-
in the analysis that follows. lowing sub-section.

6.2. Overvoltage induced time to failure for lightning arresters


6.4. Proposed distribution system protection schemes
Per the IEEE 1547–2018 standard, DG sources must remain con-
nected to the distribution system to provide reactive power support For a static distribution system protection scheme to protect against
during a fault [33]. Depending on the category of the DG source, the DG lightning arrester failure, the worst-case (shortest) time to failure for
source many not trip until a time ranging from 0.6 s to 10 s has elapsed the lighting arresters should be computed using the lightest distribution
for undervoltage levels ranging from 0.7 to 0.5 pu. Additionally, if a DG system load and rated DG output. The resulting required modifications
source is not configured for unintentional islanding, the DG source must to the distribution system protection scheme depend on whether is-
disconnect within 2 s of detecting an island condition. Thus, based on landing is desired and whether the DG will trip on undervoltage or anti-
this information, the lightning arresters should be able to withstand a islanding detection before the lightning arresters will overload.
temporary overvoltage caused by the delta-wye transformer of between If the DG source is connected to the distribution system using a
1 and 10 s, depending on the category of the DG source. delta-wye transformer and the permanent line-to-ground fault is on the
When subjected to an overvoltage of sufficient duration and mag- distribution system as shown in [7], then the DG source itself must be
nitude, metal oxide surge (MOS) arresters will fail. These temporary tripped to prevent damage to the lightning arresters on the distribution
overvoltage (TOV) ratings and curves are determined using tests spe- system. However, if the permanent fault is on the transmission system
cified in [32]. Utilizing Tables 1–3 in [8], it is determined that dis- that is connected to the distribution system via a delta-wye transformer,
tribution system lightning arresters will overload after 1 s at lower then the distribution system can be islanded to increase the distribution
threshold voltages ranging from 1.51 pu to 1.94 pu, and will overload system reliability. This can be accomplished by tripping the low-side
within 10 s beginning at 1.44 to 1.86 pu voltage. The minimum over- distribution system breaker that is the closest to the delta-wye substa-
load threshold for overloading transmission arresters after 1 s ranged tion transformer.
from 1.48 pu to 2.0 pu voltage and after 10 s transmission arresters will If the calculated voltage does not result in lightning arrester failure
overload beginning at 1.4 to 1.9 pu voltage, depending on the rating of before the DG source would otherwise trip, an intentional islanding
the arrester chosen. From this range, it is clear that some arresters, scheme can be utilized to allow the distribution system to continue
particularly those with duty cycle ratings at or above the line-to-line functioning with the DG supplying some or all of the loads. To suc-
voltage, will not overload after 1 s of being exposed to a 1.73 per unit cessfully island the distribution system by tripping the substation
overvoltage. However, arresters that are rated closer to the nominal breaker, under/over voltage and under/over frequency relays are ty-
line-to-neutral distribution or transmission system voltage will overload pically utilized [7]. If islanding is not desired, no modification to the
and fail within 1 s. distribution system protection scheme is necessary. This saves DG
For example, if a 3.0 kV arrester from Table 1 in [8] was installed on owners the expense of paying for unnecessary additional protection
the 4.16 kV IEEE 13 bus system, these arresters would fail in 1 s when equipment described in Section 3.3, and from losing revenue un-
subjected to 1.52 pu voltage, which equates to a transformer secondary necessarily.
side voltage of 0.88 per unit. Thus, this particular lightning arrester However, if the lightning arresters would fail before the DG breaker
would be overloaded for the total distribution load scenarios below trips on anti-islanding or undervoltage, then either the substation
8500 MW in Table 1, i.e. the base case, series, and parallel system breaker or DG source breaker need to be tripped, depending on whether
configurations. Thus, the distribution system protection scheme must islanding is desired. If the lightning arresters will overload within a few
be modified to island the distribution system or disconnect the DG cycles, then the DG or distribution breaker should be tripped at the
before the lightning arresters will be overloaded. inception of the fault. One such method described in [23] utilizes a
If the MCOV ratings and temporary overvoltage (TOV) curves of the transformer-high side undervoltage relay to detect the phase-to-ground
lightning arresters are known, then the exact time to failure can be fault and send a direct transfer trip to the DG source breaker. This is
determined using the calculated distribution system voltage. If this in- seen in the base case of the IEEE 13 system, where the lightning ar-
formation is not available, the worst case must be assumed, and the resters would overload in approximately one cycle, which is insufficient
lowest arrester duty cycle rating for the given distribution system vol- time to trip the DG source breaker after the transmission line-end
tage in Table 1 in [8] should be utilized. For example, the IEEE 13 node breakers have opened.
feeder base case substation voltage of 0.969 pu results in a 3.0 kV If the time to overload is greater than a few cycles but still results in
lightning arrester being overloaded in 0.02 s, or slightly over one cycle. arrester failure, then the time delay for the existing distribution system
If the computed time to failure results in the arresters overloading anti-islanding protection scheme can be decreased to avoid damage to
before the DG source trips on anti-islanding detection or undervoltage the lightning arresters. If islanding is desired, an intentional islanding
protection, then one of the following solutions in the following sub- scheme can be designed to successfully island the system before the
sections should be implemented. lightning arresters would be damaged by the fault-induced un-
grounded-delta overvoltage.
6.3. Replacing and upgrading lightning arresters After the distribution grid successfully islands, load shedding relays
are used to balance the DG generation capacity and the load. Then, the
As proposed by other papers in the existing literature, the gapless DG source is switched to droop control [34] based grid forming mode
semiconductor-based metal-oxide surge (MOS) arresters can be re- [35] in order to match small changes in the load power. After the
placed with gapped arresters due to their higher TOV withstand curves. transmission level fault has been completely cleared, the distribution
Another solution involves replacing the MOS arresters with arresters system is resynchronized and reconnected to the transmission grid
having a maximum continuous operating voltage (MCOV) equal to the following the procedure described in [7].

8
J. Snodgrass and L. Xie Electrical Power and Energy Systems 123 (2020) 106209

6.5. On-line voltage calculations and adaptive relaying strategies assumes the lightning arresters will overload for any load level or ca-
pacity of installed DG. Using this current approach, the DG source is
Another advantage of utilizing Thevenin’s equivalent method is that tripped for every single line-to-ground fault on the high side of the
the protection schemes mentioned in the preceding section can be delta-wye transformer, reducing revenue for the DG owners.
adaptively updated to reflect the current load level and maximum DG As established in this paper, only permanent (or bolted) line-to-
output. For example, if the current wind speed or incident solar energy ground faults result in the overvoltage at the high side of the delta-wye
results in the DG output being below its rated value, and the distribu- transformer. If calculating the distribution system voltages results in
tion system load is higher than the worst-case (lowest) value, the time adequate time until the lightening arresters would fail, a protective
to failure will be longer than the worst-case calculations descrived in relaying scheme could be adopted that waits until the transmission or
earlier sections. This time increase could allow a less-aggressive re- substation feeder breaker tries to reclose before tripping the DG source.
laying scheme to be implemented during low DG output and/or higher If the transmission or substation breaker recloses successfully, the DG
load conditions to avoid unnecessarily over-tripping the DG source. source can remain online and avoid interruptions of service.
Without this adaptive methodology, the worst-case scenario must al- However, if the DG source has sufficient capacity to both damage
ways be assumed, and the maximum DG output and minimum dis- the lightning arresters before the transmission breakers lock out, and
tribution system load level must be utilized to determine the worst-case supply the majority of the distribution system load, the distribution
(shortest) time to failure for the lightning arresters. system could be islanded to increase system reliability. By intentionally
For these online calculations, the worst-case error for the Thevenin islanding the distribution grid instead of always tripping the DG source
equivalent method should be calculated as seen in Table 3. Then, the breaker for all faults, the DG source may supply a portion, or all, of the
Thevenin method could scaled by the maximum error amount and be distribution system load, thus increasing the reliability of the dis-
utilized in real-time to determine the overvoltage time to failure for the tribution grid and reducing lost revenue for the DG owners. Finally,
arresters. This will determine if the DG source breaker or substation modeling the distribution system using Thevenin’s Equivalent
breaker needs to be tripped more quickly, or if the DG can stay con- Impedance allows the voltage calculations to be performed in real-time,
nected for a longer duration to participate in LVRT or wait until the allowing the distribution system protection scheme to be updated
transmission-level breakers try and reclose. adaptively based on the system load level and current, forecasted or
scheduled DG output.
7. Summary and conclusions This paper provides the initial phase of a full systematic framework
of co-designing transmission and distribution protection systems.
This paper examines future design and analysis of protective re- Future research areas include conducting stability studies to analyze the
laying schemes in distribution systems with integrated distributed effects of large mismatches between the DG capacity and load levels
generators. Specifically, when examining the impacts of inter- when the system islands, in addition to studies to improve DG and grid
connecting new distributed generation (DG) sources into an existing or synchronization and reconnection.
new distribution grid, studies must be performed to ensure that light-
ning arresters on the high side of a delta-wye distribution transformers CRediT authorship contribution statement
will not fail from a temporary overvoltage. This overvoltage condition
can happen when a persistent single line-to-ground fault occurs on the Jonathan Snodgrass: Investigation, Methodology, Software,
high side of the distribution transformer, increasing the voltage of the Writing - original draft. Le Xie: Conceptualization, Supervision.
unfaulted phases to a maximum of 1.73 pu volts. In this paper, the
Thevenin Equivalent Impedance method was used to calculate the Declaration of Competing Interest
magnitude of the overvoltage, which determined the proper protective
relaying strategy to be implemented. The authors declared that there is no conflict of interest.
As shown in previous sections, there are several proposed methods
of calculating distribution system voltages including unbalanced dis- References
tribution load flow, symmetrical components, and circuit analysis
techniques such as mesh current or node voltage. It was established that [1] ERCOT, “ERCOT Analysis of the Impacts of the Clean Power Plan, Final Rule
symmetrical components cannot be utilized for distribution systems Update,” 2015.
with single or two-phase laterals, and the remaining methods required [2] K. Kauhaniemi and L. Kumpulainen, “Impact of distributed generation on the pro-
tection of distribution networks,” in 2004 Eighth IEE International Conference on
complete detailed models of the distribution system. Additionally, dis- Developments in Power System Protection, 2004, pp. 315-318 Vol.1.
tribution load flow requires that the DG capacity be adequate to supply [3] Le TD, Petit M. Directional relays for distribution networks with Distributed
the entire load of the distribution system. Since the distribution system Generation. 11th IET International Conference on Developments in Power Systems
Protection (DPSP 2012) 2012:1–6.
model is often incomplete and lacks full models of the feeder laterals, [4] Conti S. Analysis of distribution network protection issues in presence of dispersed
another analysis method utilizing Thevenin’s equivalent circuit was generation. Electr Power Syst Res 2009;79:49–56.
proposed, which does not require a full model of the distribution [5] Doyle MT. Reviewing the impacts of distributed generation on distribution system
protection. IEEE Power Engineering Society Summer Meeting 2002:103–5.
system.
[6] El-Khattam W, Sidhu TS. Resolving the impact of distributed renewable generation
First, an equivalent network impedance was determined for the on directional overcurrent relay coordination: a case study. IET Renew Power Gener
feeder containing the DG source. Next, an estimate of the lowest load 2009;3:415.
[7] Behrendt K. Protection for unexpected delta sources. 29th Annual Western
level was calculated, which results in the worst-case voltage overload,
Protective Relay Conference. Washington: Spokane; 2002.
and a single equivalent load impedance was modeled. Then, Ohm’s Law [8] L. Pryor, “The Application and Selection of Lightning Arresters,” General Electric
was used to determine the voltage at the substation node of this Company2008.
equivalent circuit consisting of the DG source, the system impedance, [9] Schoene J, Pallem C, McDermott T, Walling R. Evaluating the Response of Surge
Arresters to Temporary Overvoltages. presented at the IEEE PES T&D Conference
and the load impedance. and Exposition 2014.
The Thevenin Equivalent Impedance method is beneficial because it [10] A. Mansoor and F. Martzloff, “The dilemma of surge protection vs. overvoltage
provides a practical and straightforward method to determine whether scenarios: Implications for low-voltage surge-protective devices,” in 8th
International Conference on Harmonics and Quality of Power. Proceedings (Cat.
a given DG capacity level will overload the lightning arresters before No. 98EX227), 1998, pp. 964-969.
the substation breaker will trip and inadvertently island the distribution [11] Walling R, Hartana R, Ros W. Self-generated overvoltages due to open-phasing of
system. This method is a significant improvement over the current ungrounded-wye delta transformer banks. IEEE Trans Power Delivery
1995;10:526–33.
approach utilized by utility companies and DSOs, which always

9
J. Snodgrass and L. Xie Electrical Power and Energy Systems 123 (2020) 106209

[12] Sakshaug E, Burke J, Kresge J. Metal oxide arresters on distribution systems: fun- on Distribution Systems. Energies 2017;10:1864.
damental considerations. IEEE Trans Power Delivery 1989;4:2076–89. [24] Kersting WH. Radial distribution test feeders. IEEE Trans Power Syst
[13] Glover JD, Sarma MS, Overbye T. Power system analysis & design. SI version: 1991;6:975–85.
Cengage Learning; 2012. [25] Suni JCP, Ruppert E, Fajoni F. A guide for synchronous generator parameters de-
[14] Chen TH, Chen MS, Hwang KJ, Kotas P, Chebli EA. Distribution system power flow termination using dynamic simulations based on IEEE standards. The XIX
analysis-a rigid approach. IEEE Trans Power Delivery 1991;6:1146–52. International Conference on Electrical Machines-ICEM 2010;2010:1–6.
[15] A. Abur, H. Singh, H. Liu, and W. Klingensmith, “Three phase power flow for dis- [26] Schneider KP, Mather BA, Pal BC, Ten CW, Shirek GJ, Zhu H, et al. Analytic
tribution systems with dispersed generation,” 14th PSCC, Sevilla, vol. 11, 2002. Considerations and Design Basis for the IEEE Distribution Test Feeders. IEEE Trans
[16] Moghaddas-Tafreshi SM, Mashhour E. Distributed generation modeling for power Power Syst 2018;33:3181–8.
flow studies and a three-phase unbalanced power flow solution for radial dis- [27] European EMTP-ATP Users Group e.V., “ATP Software,” 2014.
tribution systems considering distributed generation. Electr Power Syst Res [28] M. L. Baughman, C.-C. Liu, and R. C. Dugan, “IEEE 4 Node Test Feeder,” IEEE Power
2009;79:680–6. Energy Society Distribution System Analysis Subcommittee2006.
[17] Chassin DP, Schneider K, Gerkensmeyer C. GridLAB-D: An open-source power [29] M. L. Baughman, C.-C. Liu, and R. C. Dugan, “IEEE 13 Node Test Feeder,” IEEE
systems modeling and simulation environment. IEEE/PES Transmission and Power Energy Society Distribution System Analysis Subcommittee2004.
Distribution Conference and Exposition 2008;2008:1–5. [30] ABB, “Overvoltage protection: Metal-oxide surge arresters in medium-voltage sys-
[18] D. M. Fobes, C. Coffrin, F. Geth, and S. Claeys, “PowerModelsDistribution.jl: An tems,” 2018.
Open-Source Framework for Exploring Distribution Power Flow Formulations,” in [31] IEEE. IEEE Guide for the Application of Metal-Oxide Surge Arresters for Alternating-
Submitted to 21st Power Systems Computation Conference, Porto, Portugal, 2020. Current Systems. IEEE Std C62.22-2009 (Revision of IEEE Std C62.22-1997)
[19] Fortescue CL. Method of symmetrical co-ordinates applied to the solution of poly- 2009:1–142.
phase networks. Trans Am Inst Electr Eng 1918;37:1027–140. [32] IEEE. IEEE Standard for Metal-Oxide Surge Arresters for AC Power Circuits (> 1
[20] Anderson PM. Analysis of faulted power systems. New York: IEEE press; 1995. p. kV). Revision of IEEE Std C62.11-2005 2012. p. 1–121.
445. [33] IEEE, “IEEE Standard for Interconnection and Interoperability of Distributed Energy
[21] Schweitzer E, Zocholl SE. Introduction to symmetrical components. in proceedings Resources with Associated Electric Power Systems Interfaces,” in IEEE Std 1547-
of the 58th Annual Georgia Tech Protective Relaying Conference. GA: Atlanta; 2018 (Revision of IEEE Std 1547-2003), ed, 2018, pp. 1-138.
2004. [34] R. H. Lasseter, “MicroGrids,” in 2002 IEEE Power Engineering Society Winter
[22] J. W. Nilsson and S. A. Riedel, Electric circuits: Pearson Upper Saddle River, NJ, Meeting. Conference Proceedings (Cat. No.02CH37309), 2002, pp. 305-308 vol.1.
2015. [35] Rocabert J, Luna A, Blaabjerg F, Rodriguez P. Control of power converters in AC
[23] Norshahrani M, Mokhlis H, Abu Bakar A, Jamian J, Sukumar S. Progress on microgrids. IEEE Trans Power Electron 2012;27:4734–49.
Protection Strategies to Mitigate the Impact of Renewable Distributed Generation

10

You might also like