This article has been accepted for publication in a future issue of this journal, but has not been

fully edited. Content may change prior to final publication. Citation information: DOI 10.1109/TPWRD.2016.2524654, IEEE
Transactions on Power Delivery
1

Automated Review of Distance Relay Settings

Adequacy After the Network Topology Changes
M. Tasdighi, Student Member, IEEE, and M. Kezunovic, Fellow, IEEE

 Zone 3 phase reach based on apparent impedance of

Z3append
Abstract—A challenge raised in today’s power system is next adjacent line-end faults.
assessing the system protection security and dependability after zl Impedance of the line.
the topology has been changed due to relay operation upon
occurrence of cascading faults or intentional operator switching zladj Impedance magnitude of the next adjacent line i.
i
action. This paper proposes an automated setting calculation
module which could be used to review the adequacy of the z adj
p Line ohms path magnitude to the next adjacent bus i.
i
distance relay settings following network topology changes. The Apparent impedance for three phase fault on remote
calculation procedure is broken down into blocks which could be zrem
bus.
processed in parallel in order to improve the computation speed.
Apparent impedance for three phase fault on next
A new concept called distance-of-impact (DoI) is proposed to ziadj
reduce the computational burden by conducting the calculations adjacent bus i.
on a limited portion of the network affected by the topology adj
zend Apparent impedance for three phase line-end fault on
change. The module performance is tested on synthetic IEEE 118- i next adjacent line i.
bus and real-life Alberta transmission operator systems. A adj Number of next adjacent buses.
N
sensitivity analysis in the form of N-2 contingency impact on the
Required updated column of Zbus corresponding to a
network relay settings coordination is conducted on the test Z col
systems. line-end fault implementation.
Vnpost Post-fault voltage of bus n.
Index Terms—Distance-of-impact (DoI), power system
protection security and dependability, N-2 contingency, phase Vnpre Pre-fault voltage of bus n.
distance settings, relay ranking, topology control, vulnerability. Ri  j Distance relay looking from bus i to bus j.
I. NOMENCLATURE Yij The admittance of the line between buses i and j.
Nbu Number of buses in the network. Ysh Shunt capacitance in pi model of the line
Nbr Number of branches in the network.
VLL Rated line-to-line voltage. II. INTRODUCTION
PERATIONAL schemes are being changed in today’s re-
I rating
30
The line current rating.
Relay reach in primary ohms at a power factor angle
O structured power systems. As an example, deploying the
renewable distributed generation is changing the traditional
Z relay
of 30 degrees. way of operation [1]. Other examples include the transmission
Z1 Zone 1 phase reach in primary ohms. line switching mostly aimed towards reducing the system
Z2 Zone 2 phase reach in primary ohms. operational cost in real time [2]-[5] or the load shedding
Z3 following a contingency [6]-[8]. Multiple switching actions
Zone 3 phase reach in primary ohms.
may also be associated with cascading line tripping [9]-[11] as
Z 2l Zone 2 phase reach based on line ohms only. well as transmission line taken out of service due to
Zone 2 phase reach based on the remote bus fault maintenance purposes. As a result, setting coordination of the
Z 2app
apparent impedance. distance relays may be affected due to the change of the
Z 3l Zone 3 phase reach based on line ohms only. network short circuit values following a topology change [1].
Zone 3 phase reach based on apparent impedance of The tool to review the adequacy of the protection coordination
Z3appbus of the distance relays neither before the planned topology
next adjacent bus faults.
change, e.g. for corrective purposes, nor after the topology
change, e.g. due to maintenance purposes, is available. Hence
M. Tasdighi and M. Kezunovic are with Department of Electrical and an assessment whether a change in relay settings is needed to
Computer Engineering, Texas A&M University, Texas, USA. (E-mail: maintain the security and dependability of the power system
m.tasdighi@tamu.edu; kezunov@ece.tamu.edu).
The financial support for this research comes from the Advanced
protection at such times is typically not performed.
Research Projects Agency-Energy (ARPA-E), U.S. Department of Energy There are several reported studies which investigated the
(DOE), under the Green Electricity Network Integration (GENI) project transmission line relay operation under abnormal conditions
“Robust Adaptive Topology Control”, ARPA-E award no. DE-AR0000220.

0885-8977 (c) 2015 IEEE. Personal use is permitted, but republication/redistribution requires IEEE permission. See http://www.ieee.org/publications_standards/publications/rights/index.html for more information.
This article has been accepted for publication in a future issue of this journal, but has not been fully edited. Content may change prior to final publication. Citation information: DOI 10.1109/TPWRD.2016.2524654, IEEE
Transactions on Power Delivery
2

especially power swings [12]-[17]. Several relay ranking proposed and implemented with comprehensive
schemes have been proposed to identify vulnerable relays in a simulations.
network. Singh et al. [12] have ranked the relays based on  The results of parallelization using supercomputing
Lyapunov stability criterion related to the power swing facilities.
severity. Relay margin concept has been used in [13] to  The results from the Alberta transmission operator
measure the closeness of a relay from issuing a trip signal. system that demonstrate the effectiveness and
Reference [14] has proposed a new approach to locate all the robustness of the approach in realistic user setting.
electrical centers following an unstable swing and simplify the  Discussion of how the computational burden can be
visual monitoring of all the R-X plots. For stable swings, the reduced for implementation on real sized systems.
concepts of the branch norm, fault norm, and system norm are
defined to rank the power swings, faults or contingencies, and The rest of the paper is organized as follows. Section III
detect an out-of-step condition respectively. Seethalekshami et describes the creation of short circuit databases necessary for
al. [15] have used the branch loss sensitivity measure the phase distance setting calculations. The basic rules for
presented in [16] to propose a relay ranking index (RRI). This calculating zones 1, 2 and 3 of a distance relay are discussed in
index is defined as the ratio of normalized apparent impedance this Section. The general procedure for the proposed setting
seen by the relay to the corresponding branch loss sensitivity. calculation, as well as the distance-of-impact measure to asses
The less the RRI value, the more the relay is probable to miss- relay setting adequacy are presented in Section IV. Section V
operate under power swing and voltage instability conditions. illustrates simulation results along with sensitivity analysis.
The performance of various conventional power swing Concluding remarks and the main contributions are
detection algorithms for the relays on the series compensated summarized in Section VI.
lines are compared in [17].
Overloaded lines, as a result of a line-tripping contingency, III. PHASE DISTANCE SETTING RULES
could also lead to relay miss-operation which is known as Utilities, all over the world, follow different rules in setting
“load encroachment” into the relay operating characteristic. calculation of the distance relays depending on their approach
This happens when the load apparent impedance gets so small to operating the network. The setting procedure followed in
that it falls into the protection zone of a relay. Calculating the this study is the same as the default procedure in CAPE [21].
relay apparent impedance from a load flow analysis could be A modern distance relay has several elements which provide
used to detect such a case [18]-[19]. many protection functions in a single package. In this study,
The above mentioned efforts have studied the miss- the focus is on the phase distance elements and mho settings of
operation of distance relays under cases such as power swing, different zones. There are two ways to calculate the zone
voltage instability, and load encroachment. The focus of this settings: one is based on the line ohms only, which is not so
study is to investigate the adequacy of the network relay practical, and the other, which is used here, is to consider both
settings for a new (evolving) network topology and identify the the line ohms and the apparent impedance of different fault
consequent vulnerable relays at selected locations in the types seen by the relay [23], Fig. 1.
transmission system. The proposed automated approach for To obtain the initial mho settings, in regards to the apparent
adequacy checking of distance relay settings is able to improve impedances, three types of fault calculation, as shown in Fig.
the transmission system operators (TSOs) actions by providing 1, are implemented: a) three-phase fault on remote bus, b)
them with a decision making tool to assess the adequacy of three-phase fault on the next adjacent bus, and c) three-phase
distance settings for an evolving network topology, especially line-end fault on the next adjacent line. The maximum torque
the long-term ones, so that a proper action can be taken either angle (MTA) is considered the same as the zone 1 line angle,
before or after the topology change as needed. For example, in i.e., MTA   Zl .
the scenario that the operator is provided with a list of The apparent impedances are checked as follows to make
switching actions for corrective purposes, e.g. load shedding sure they are valid during the setting procedure:
or cost reduction, he/she is able to assess the switching
candidates in regards to their impact on the protection security
and dependability when selecting the best option. If the
topology has already changed due to maintenance purposes or
Remote Bus Next Adjacent Bus
cascading tripping as a consequence of relay mal-operation, Forward Direction
the operator could assess the protection security and Relay
Zone 1 Line Next Adjacent Line
dependability for the current topology and take proper action.
Previous study [20] proposed the general framework for the F3
approach and briefly discussed the impacts of switching
F1 F2
actions on distance relay settings coordination, which was
tested on rather small-sized systems. The new contributions
reported here relate to: Figure 1. Fault types used for phase distance setting calculation: remote
bus fault (F1), next adjacent bus fault (F2), and line-end fault (F3)
 The concept of distance of impact (DoI), which is

0885-8977 (c) 2015 IEEE. Personal use is permitted, but republication/redistribution requires IEEE permission. See http://www.ieee.org/publications_standards/publications/rights/index.html for more information.
This article has been accepted for publication in a future issue of this journal, but has not been fully edited. Content may change prior to final publication. Citation information: DOI 10.1109/TPWRD.2016.2524654, IEEE
Transactions on Power Delivery
3

 zrem  10  Zl & IV. RELAY SETTING CALCULATION AND ADEQUACY CHECK

z
 rem  The proposed module contains algorithms which check the
zrem    zrem  MTA  (1) adequacy of the existing relay settings for the new system
 4
 0 topology after switching. It performs fast relay setting
otherwise
calculation for the new topology and compares the new setting
 ziadj  10  Zl & values with the current settings. Figure 2 shows a general
 adj flowchart of the proposed setting calculation module.
z 
ziadj iN adj   i  ziadj  MTA  (2) The critical challenge for actual implementation of this
 4 methodology is that the setting coordination check of distance
 0
 otherwise relays in a transmission operator (TOP) sized system is a
 adj
zend  10  Zl &
significantly time consuming task and should be automated
 adj i [21]-[23]. Identifying the relays whose settings get affected
z 
adj
zend   endi  zend
adj
 MTA  (3) due to a change of the network short circuit values following a
i iN
adj

 i
4 network topology change could be considered as an initial step
 0 in conducting automated settings coordination check.
 otherwise
The phase distance setting rules are as follow: Performing the setting coordination check on the affected
relays should also be fast enough. As a result, the focus of this
A. Zone 1 Setting Rule study is also to make the setting calculation process faster by
Z1  0.8  Zl (4) investigating how to reduce the problem size and calculation
burden. This paper illustrates the methodology and
B. Zone 2 Setting Rule demonstrates its effectiveness using a real-life network
  example.

Z 2l  max 1.2  Zl , Zl  0.2  min zladj

i
  


(5)
A. Short-Circuit Model Data Preparation
 iN adj 
The input for the relay setting calculation module is: 1)
Z 2app  1.2  zrem (6)
short-circuit model data for buses, branches, generators, 2)

Z 2  max Z 2l , Z 2app  MTA  (7) power flow data, and default relay settings and 3) list of
network topology changes. Having recognized the network
C. Zone 3 Setting Rule topology, the module builds up the Zbus for the whole network

 
Z3l  1.2  max z adj
p i iN adj
(8)
which is used in fault calculations. Furthermore, a list of
network distance relays and their adjacent buses and branches
 1.1 max  z 
is obtained from the network topology. For each relay the
Z3appbus adj
i (9)
iN adj

Z3append  1.1 max zend

adj
 i
 (10)
iN adj


Z3  max Z3l , Z3appbus , Z3append  MTA  (11)
Then, the load encroachment is evaluated for all zones to
prevent phase protective relay settings from limiting the
transmission system loading capacity while maintaining
dependability of the network protection. According to NERC
[24], the relay performance should be checked for 150% of the
highest seasonal rating of the lines at 0.85 per unit voltage and
a power factor angle of 30 degrees:
0.85  VLL
30
Z relay  (12)
3 1.5  I rating

Furthermore, according to setting rules followed by CAPE

[21], the zone settings are checked not to over-reach 20% of
the transformer impedance in order not to interfere with the
distance relay settings on the lines located after the
transformer.

Figure 2. General flowchart of the relay setting calculation module

0885-8977 (c) 2015 IEEE. Personal use is permitted, but republication/redistribution requires IEEE permission. See http://www.ieee.org/publications_standards/publications/rights/index.html for more information.
This article has been accepted for publication in a future issue of this journal, but has not been fully edited. Content may change prior to final publication. Citation information: DOI 10.1109/TPWRD.2016.2524654, IEEE
Transactions on Power Delivery
4

buses and branches on which the bus fault and line-end fault corresponding type of fault. The voltage and current values are
should be implemented respectively to obtain the relay settings then used to calculate the associate apparent impedances. To
are determined. Having identified the required fault improve the calculation speed, parallel computation could be
calculations for the setting procedure of each distance relay in performed on the three fault type calculation independently.
a network, this method could be implemented on a system with Figure 3 shows the general flowchart of implementing parallel
different types of relays without causing an interference with computation for N tasks each of which might contain several
setting calculation of the distance relays in that network. The sub-tasks. For the parallel computation to be implemented the
module could be run for the identified distance relays in the tasks should be independent from each other, i.e., there should
network to assess their settings. Moreover, any exclusive be no flow of data required between the tasks for each of them
condition such as the type of the relay used at specific points to be completed. In that case, all the tasks could be submitted
in the network or specific setting procedures (rules) followed to a group of workers (computing nodes) called pool of
by a network operator could be predefined in the module so workers. Each worker might include several processing cores.
the relays are set accordingly. The access to input data is provided for all the workers so they
can use the same input data. However, the worker does not
B. Fault Databases Preparation
need any data obtained by other tasks.
During the fault calculations several updates to Zbus are Now, as an example, let’s discuss how the line-end fault
required depending on the fault type. This prevents repetitive database preparation could be parallelized. Considering to-end
and excessive Zbus calculation if not necessary. The sparsity of the lines, the goal is to obtain the voltages of all the buses
oriented compensation methods are used to perform updates to for the faults implemented on all the to-ends of the lines. In
Zbus [25]. Zbus should specifically be modified by updating the other words, (13) to (17) should be calculated for i = 1:Nbr.
required column when implementing a line-end fault as As it could be understood from these equations, they could be
assumedly another bus is added to the network. The process of conducted for branch n completely independent from those of
calculating the voltages of all the buses for the case of a line- branch m. For each implemented fault, the voltage of all the
end fault implemented on the j side of the line from bus i to buses could be calculated and stored separately. This allows
bus j, are formulated as follows using branch-oriented implementation of the parallel computation. The same process
compensation method [25]: could be implemented for the from-end of the lines and also
i
T for the remote bus fault calculations. The latter could be
U Nbu1  0 Yij 0 (13)
conducted much faster as no change to Zbus is required to
i j T implement it. For a fault implemented on bus n, Zcol is simply
0 0 1 0 0 (14) the nth column of Zbus and there is no need to conduct (16)-
V Nbu2  
0 0 1 0 0  (17). The network Zbus is the only common data fed into the
three blocks. Having the voltages at both ends of the branches
0 Yij 
  (15) corresponding to each fault case together with their
Yij (Yij  Ysh )  impedances, the branches currents could be easily calculated.

 
1
W  I 22    V T  Z bus  V 

X  Zbus  V  W  V T  Z bus (16)

Z   Zbus  X   U
1 Z 
Z col    (17)
Yij  Ysh  U  Z  1 ( Nbu 1)1
T

 Z col 
Vnpost  Vnpre 1  coln1  n  {1,..., Nbu} (18)
 Z( Nbu 1)1 
 
C. Implementation of Parallel Computation
The algorithm computation burden mostly relates to the
creation of different fault type databases, as shown in the three
blocks highlighted in Fig. 2. Specifically, line-end fault
database preparation for two ends of the line is a very time
consuming task. This is because the power system Zbus is a big
order sparse matrix for which operations such as inversion and
multiplication shown in (16)-(17), require more computation
efforts from the processor. Each of the fault databases contains
the bus voltage and branch current values for the
Figure 3. General implementation of parallel computation for N tasks

0885-8977 (c) 2015 IEEE. Personal use is permitted, but republication/redistribution requires IEEE permission. See http://www.ieee.org/publications_standards/publications/rights/index.html for more information.
This article has been accepted for publication in a future issue of this journal, but has not been fully edited. Content may change prior to final publication. Citation information: DOI 10.1109/TPWRD.2016.2524654, IEEE
Transactions on Power Delivery
5

D. Distance-of-Impact (DoI) Concept DoI = 2

For the algorithm to be practical for real-time applications
e DoI = 1
in TOP-sized networks, the calculation burden and i c g
corresponding time should be reduced further. For this
purpose, we have investigated the distance-of-impact (DoI)
concept, which determines how far from the line that a b
experiences a switching action we could expect the relay j f d h
settings to be affected. This has been verified from conducting
numerous simulations on both the 118-bus and Alberta
transmission operator systems. In the case that the switching
Figure 4. Illustrating the concept of DoI
actions impact on the relay settings is limited to a certain
electrical distance from the switching location, the calculations transformer protection and it is assumed that transformers are
are then focused on the portion of the network within that normally protected by differential relays.
distance. Figure 4 illustrates an example to clarify the DoI
A. Sensitivity Analysis
concept. For the switched transmission line a-b, the DoI of one
includes the buses c to f with their corresponding branches and In the first step of the simulations, a sensitivity analysis to
relays, and DoI of two includes those of the buses c to j. investigate the impact on the network relay settings by N-2
The proposed module could be used in practice to assess contingencies with 2 lines switched out has been done using
multiple switching impacts on the network relay settings. The IEEE 118-bus test system. The N-2 contingencies are chosen
settings are calculated for the new system topology and to be studied because of two reasons: 1) the number of cases to
compared with the previous ones to identify the affected be investigated and 2) they are practical switching scenarios in
relays. The module’s output contains the list of the relays today’s power systems. Conducting each case is significantly
whose settings have changed beyond an acceptable margin. It fast as the size of the system is rather small. All N-2
provides the network operator with an extra decision making contingencies for which the power flow solution converges
tool to deal with the impact of switching actions on the have been considered (13945 cases in total). A sensitivity
security and dependability of power system protection. analysis is used to detect and rank the probable system
protection vulnerabilities following a network topology change
V. SIMULATION RESULTS leading to N-2 contingency.
Multiple results from the sensitivity analysis conducted on
The performance of the proposed module is tested on the
IEEE 118-bus system could be seen in Table I. The relays
synthetic IEEE 118-bus [26] and real-life Alberta transmission
whose settings change for a greater number of N-2
operator system [27]. The actual calculated relay setting values
contingencies are identified as vulnerable points in the network
are verified by comparing them with the output of CAPE
protection. A relay is considered being affected if its zone 2 or
commercial package used in the project reports submitted to
zone 3 experiences a change beyond 5% of the base network
ARPA-E, U.S. Department of Energy [28]. The relays are
settings. Zone 1 is not a concern as it is only based on the line
assumed to be set only in the forward direction as shown in
impedance. Top 10 vulnerable points (critical relays) in the
Fig. 1. The transmission lines do not have mutual coupling and
network protection following N-2 contingency cases are
transformer protection is neglected for the sake of simplicity.
ranked in Table I based on the number of N-2 contingency
In general, the impact of mutual coupling is limited to the
cases which affect them. The participation ratio for a relay
relays on the lines with such feature and could be modeled by
means the ratio of the number of N-2 contingency cases which
proper changes to the network Zbus so it does not harm the
have affected the relay to the total number of contingency
effectiveness and generality of the proposed approach and it
cases. Top 10 N-2 contingency cases according to their
does not introduce a significant additional computational
impacts on the network relay settings, and number of affected
burden. The distance protection is not the best option for
relays, are also shown in Table I. The lines participating in the
Table I. Results of sensitivity analysis on the IEEE 118 bus test system
Critical Relays Critical N-2 Contingency Cases Critical Lines
Rank
Relay Participation Ratio (%) Lines (from-to) No. of Affected Relays Lines (from-to) Participation Ratio (%)
1 R98-80 18.32 60-61 & 82-83 21 82-83 35.74
2 R57-56 16.9 60-61 & 82-96 18 94-100 27.58
3 R58-56 12.35 54-56 & 82-83 18 82-96 26.45
4 R16-17 10.24 49-51 & 82-83 18 60-61 23.13
5 R70-24 10.2 31-32 & 82-83 18 11-12 19.42
6 R62-60 10.13 15-19 & 82-83 18 31-32 19.34
7 R54-55 10.06 11-12 & 82-83 18 49-51 18.54
8 R105-106 9.98 82-83 & 100-106 17 54-56 18.25
9 R59-54 9.76 82-83 &100-104 17 100-103 17.56
10 R17-15 9.72 60-61 & 94-100 17 100-104 17.52

0885-8977 (c) 2015 IEEE. Personal use is permitted, but republication/redistribution requires IEEE permission. See http://www.ieee.org/publications_standards/publications/rights/index.html for more information.
This article has been accepted for publication in a future issue of this journal, but has not been fully edited. Content may change prior to final publication. Citation information: DOI 10.1109/TPWRD.2016.2524654, IEEE
Transactions on Power Delivery
6

majority of the N-2 contingency cases with significant impacts may be prohibitive for the control room setting today, the
on the relay settings could also be identified from the simulation time could be improved to a desired level just using
sensitivity analysis. Table I shows top 10 of such critical lines. the ordinary high-end control room computers. With the pace
In the second step of the simulations, the same sensitivity of the technology development, the use of the supercomputers
analysis was conducted for 1000 N-2 contingency cases on in the control room may be feasible in the near future.
Alberta transmission operator system. The top 10 N-2
C. The Role of Distance-of-Impact
contingency cases with major impact on the relay settings are
shown in Table II. The size of this system is rather big (2585 Another focus of the sensitivity analysis was to search for
buses and 2970 branches) and the relay setting calculation the DoI of the switching actions. For this purpose, a search
process is time-consuming, specifically the process of creating space of the branches starting from the ones adjacent to both
and updating the line-end fault databases. The parallel ends of the line participating in the switching action is created.
computation and supercomputing facilities have been deployed The search space grows based on the network connectivity
to conduct the contingency cases as will be discussed in the graph till it covers all the affected relays on the network
next section. branches. For IEEE 118-bus test system, it was observed that
the impact of N-2 contingency cases is limited to the relays on
B. The Role of Parallelization the branches within DoI=3 of the switching action as it could
Parallel computation technique is implemented to increase be seen for the top 10 cases in Table III. Figure 6 shows one-
the calculation speed. For this purpose, Texas A&M line diagram of IEEE 118-bus test system for which the
University supercomputing facility [29] with the access to the switching actions 60-61 and 82-83, the contingency case
maximum of 32 workers (nodes) of the facility to conduct the ranked first in Table I, are highlighted in red. The neighboring
simulations has been employed. branches up to DoI of three are highlighted in blue and as it
The simulation time to run the setting module for each N-2 could be seen the affected relays, which are shown by red
contingency case of 118-bus system is insignificant already arrows, are within the DoI. By the use of DoI concept, the
(less than a second) when using only one processing node of relays for which the short circuit values should be updated are
supercomputing facilities even without employing parallel
computation. Therefore, to show the effectiveness of the
12000
parallel computation technique in improving the calculation
speed on IEEE 118-bus test system, it is deployed in 10000
conducting all the contingency cases (13945 cases) together
Simulation Time (s)

when having access to different numbers of workers. On the 8000

other hand, running the setting calculation module for each
contingency case of the real sized Alberta transmission 6000

operator system is significantly time consuming. The

4000
improvements from parallel computation could be seen when
implemented on even one contingency case. For Alberta 2000
system, the setting module is run for the case ranked first in
Table II having access to different numbers of workers. Figure 0
1 5 10 15 20 25 30
5 (a) and (b) show how parallel computation could Number of Workers
significantly improve the module calculation speed based on (a)
4
the available numbers of workers for IEEE 118-bus system 4
x 10

and Alberta system respectively. In Fig. 5, one worker

3.5
represents running the module without parallelization. These
results provide a perspective how parallel computation 3
Simulation Time (s)

benefits escalate especially when the module is conducted on a 2.5

TOP-sized network. While the cost of supercomputing facility
2

Table II. N-2 Contingency cases affecting major relay settings 1.5
Rank Lines (from-to) No. of Affected Relays
1 89-91 & 579-585 29 1
2 420-865 & 666-1691 29 0.5
3 420-865 & 1318-1344 25
4 207-590 & 666-1200 23 0
1 5 10 15 20 25 30
5 208-581 & 242-253 23 Number of Workers
6 35-331 & 167-737 22 (b)
7 297-483 & 669-677 22
8 35-331 & 666-1670 22 Figure 5. Simulation time based on the number of workers; (a) Running
the module for all the N-2 contingency cases together in IEEE 118-bus
9 152-988 & 1431-1484 22
system; (b) Running the module for the contingency case ranked 1 st in
10 63-821 & 136-514 21 Table II for Alberta system

0885-8977 (c) 2015 IEEE. Personal use is permitted, but republication/redistribution requires IEEE permission. See http://www.ieee.org/publications_standards/publications/rights/index.html for more information.
This article has been accepted for publication in a future issue of this journal, but has not been fully edited. Content may change prior to final publication. Citation information: DOI 10.1109/TPWRD.2016.2524654, IEEE
Transactions on Power Delivery
7

Table III. Distance of impact for top 10 cases in both systems Table IV. Portion of the network within DoI for both systems
Rank 1 2 3 4 5 6 7 8 9 10 IEEE 118-bus (DoI=3) Alberta (DoI=5)
IEEE 118-bus Rank No. of No. of No. of No. of
3 2 3 3 3 3 3 2 2 2 Buses Branches Buses Branches
system
DoI 1 56 113 239 420
Alberta TOP
5 5 5 3 4 5 5 5 4 4 2 60 120 310 569
system
3 65 122 242 467
the ones within the DoI of the switching action leading to a 4 73 133 303 542
5 64 131 168 325
significant reduction in computational burden. 6 72 142 236 447
The sensitivity analysis for the contingency cases in Alberta 7 59 120 198 348
transmission operator system points the maximum of DoI to be 8 50 94 176 305
5 as shown in Table III for the previously obtained top 10 9 50 94 364 660
10 59 114 229 407
contingency cases. The number of buses and branches within
DoI = 3 and DoI = 5 for the top 10 cases in IEEE 118-bus and These simulations are conducted while deploying 1 worker
Alberta transmission operator systems respectively are shown for IEEE 118-bus system and 30 workers for Alberta system.
in Table IV. The number of relays which settings are to be As it could be seen from Fig. 7, simulation time could be
checked is two times the number of branches in DoI. As significantly reduced depending on the required number of
mentioned before, creating and updating the line-end fault line-end fault simulations especially in a real sized system.
database is the most time consuming part of the setting Considering the DoI, the calculations could be done
calculation process. Following the network topology change, exclusively for the relays in the portion of the network within
the line-end fault values for the relays within DoI should be the distance. This leads to significant time savings in
updated. Figure 7 (a) and (b) shows the simulation time for the simulations as shown in Fig. 8 (a) and (b). The calculation
cases ranked first in Tables I and II respectively based on time for 10 cases of Tables I and II could be compared
different number of implemented line-end faults to obtain the respectively between 2 scenarios: 1) running the setting
updated values. Different numbers of line-end faults are calculation module for the whole system and 2) doing it for the
obtained based on the number of relays within the DoI as it portion of the system within DoI = 3 for 118-bus system and
increases from the switching action location.

Figure 6. One-line diagram of IEEE 118-bus test system

0885-8977 (c) 2015 IEEE. Personal use is permitted, but republication/redistribution requires IEEE permission. See http://www.ieee.org/publications_standards/publications/rights/index.html for more information.
This article has been accepted for publication in a future issue of this journal, but has not been fully edited. Content may change prior to final publication. Citation information: DOI 10.1109/TPWRD.2016.2524654, IEEE
Transactions on Power Delivery
8

1
1

0.8
Simulation Time (s)

0.8
0.6

Simulation Time (s)

0.4 0.6

0.2
0.4
0
116 194 239 294 335 378
Number of Implemented Line-End Faults
0.2 for the whole network
(a)
for DoI=3 from the switching action
1200
0
1000 1 2 3 4 5 6 7 8 9 10
Simulation Time (s)

The Rank of N-2 Contingency Case (Table III)

800
(a)
600
1500
400

200

Simulation Time (s)

0
108 195 321 420 573 734 889 1038 1187 1332 1000
Number of Implemented Line-End Faults
(b)
Figure 7. Simulation time based on the required number of line-end
faults; (a) Case ranked 1st in Table I in IEEE 118-bus system; (b) Case
500
ranked 1st in Table I in Alberta transmission operator system
for the whole network
DoI = 5 for Alberta transmission operator system. The parallel for DoI=5 from the switching action
computation has been deployed in both scenarios for Alberta
0
system. In the first scenario, the simulation time for all cases is 1 2 3 4 5 6 7 8 9 10
The Rank of N-2 Contingency Case (Table III)
almost the same while in the second scenario it changes based
on the network connectivity graph. As it could be seen from (b)
Figure 8. Simulation time comparison between with and without
Fig. 8, the calculation time has been reduced significantly and
implementing DoI; (a) Top 10 cases of IEEE 118-bus system; (b) Top
real-time identification of vulnerable relays following a 10 cases of Alberta transmission operator system
network topology change becomes more practical.

VI. CONCLUSION ACKNOWLEDGEMENT

The contributions of this work are: The authors gratefully acknowledge Dr. Gurunath Gurrala’s
 The proposed algorithm allows a novel way of setting contribution in developing the setting calculation module at
calculation and evaluation under changing network the initial steps of this study.
topology. The authors also acknowledge Electrocon International
Inc., especially Dr. Ashok Gopalakrishnan, for providing
 The proposed parallel computation technique
generous technical assistance in conducting this study.
significantly reduces the computation time and makes
Special thanks are due Texas A&M high performance
the approach applicable for real-time analysis.
research computing center (HPC) for providing access to the
 The proposed calculation module follows the same
supercomputing facilities for this study.
relay setting procedure as CAPE commercial package
and is tested and verified on real-sized Alberta
REFERENCES
transmission operator systems.
[1] M. Kezunovic, et al., "Reliable Implementation of Robust Adaptive
 The proposed distance of impact (DoI) metric is Topology Control," 47th Hawaii International Conference on System
realized to handle numerous cases of network Sciences (HICSS), pp. 2493-2502, Jan. 2014.
topology changes in the form of N-2 contingencies, [2] K. W. Hedman, S. S. Oren, and R. P. O’Neill, “Optimal Transmission
Switching: Economic Efficiency and Market Implications,” Journal of
which significantly reduces the computation burden. Regulatory Economics, vol.40, no.2, pp.111-140, 2011.
The proposed decision making tool can allow the utility [3] A. Khodaei and M. Shahidehpour, “Transmission Switching in
staff to assess the impact of multiple switching actions and Security-Constrained Unit Commitment,” IEEE Trans. Power Syst., vol.
network topology changes on network protection security and 25, no. 4, pp. 1937-1945, Nov. 2010.
[4] R. P. O’Neill, et al., “Dispatchable Transmission in RTO
dependability leading to a proper setting coordination action. Markets,” IEEE Trans. Power Syst., vol.20, no.1, pp.171-179,
Feb.2005.

0885-8977 (c) 2015 IEEE. Personal use is permitted, but republication/redistribution requires IEEE permission. See http://www.ieee.org/publications_standards/publications/rights/index.html for more information.
This article has been accepted for publication in a future issue of this journal, but has not been fully edited. Content may change prior to final publication. Citation information: DOI 10.1109/TPWRD.2016.2524654, IEEE
Transactions on Power Delivery
9

[5] K. W. Hedman, et al., “Smart flexible just-in-time transmission and [28] ARPA-E Robust Adaptive Transmission Control (RATC) Project
flowgate bidding,” IEEE Trans. Power Syst., vol. 26, no. 1, pp. 93-102, granted under GENI contract 0473-1510, ARPA-E award number DE-
Feb. 2011. AR0000220, [Online]. http://smartgridcenter.tamu.edu/ratc/web/
[6] A.R. Escobedo, E. Moreno-Centeno, K.W. Hedman, "Topology Control http://arpa-e.energy.gov/?q=slick-sheet-project/automated-grid-
for Load Shed Recovery," IEEE Trans. Power Syst., vol.29, no.2, disruption-response-system
pp.908-916, March 2014. [29] High performance research computing (HPC) [Online]. Available:
[7] W. Shao and V. Vittal, “Corrective switching algorithm for http://sc.tamu.edu.
relieving overloads and voltage violations,” IEEE Trans. Power Syst.,
vol. 20, no. 4, pp. 1877-1885, Nov. 2005.
[8] G. Granelli, et al., “Optimal network reconfiguration for congestion
management by deterministic and genetic algorithms,” Elect. Power
Syst. Res., vol.76, no.6-7, pp.549-556, Apr. 2006.
[9] H. Ren, I. Dobson, and B.A., Carreras, "Long-Term Effect of the n-1
Criterion on Cascading Line Outages in an Evolving Power
Transmission Grid," IEEE Trans. Power Syst., vol.23, no.3, pp.1217- Mohammad Tasdighi (S’13) received his B.Sc. and
1225, Aug. 2008. M.Sc. degrees in electrical engineering from
[10] M. Kezunovic, C. Zheng, and C. Pang, “Merging PMU, Operational, University of Tehran, Tehran, Iran, in 2009 and
and Non-operational Data for Interpreting Alarms, Locating Faults and 2012, respectively. Currently, he is a Research
Preventing Cascades,” 43rd Hawaii International Conference on System Assistant pursuing his Ph.D. studies in electrical
Sciences (HICSS), January 2010. engineering at Smart Grid Center, Power System
[11] M. Kezunovic, P. Chen, A. Esmaeilian, and M. Tasdighi," Control & Protection Lab, Texas A&M University,
Hierarchically Coordinated Protection: An Integrated Concept of College Station, TX, USA. His research interests
Corrective, Predictive, and Inherently Adaptive Protection," CIGRE include: adaptive protection of power systems, smart grid, renewable energy
International Conference on Actual Trends in Development of Power sources, energy management in smart microgirds, power system operation.
System Relay Protection and Automation, Sochi, Russia, June 2015.
[12] C. Singh and I. A. Hiskens, “Direct assessment of protection operation
and nonviable transients,” IEEE Trans. Power Syst., vol. 16, pp. 427-
434, Aug. 2001.
[13] F. Dobraca, M. A. Pai, and P. W. Sauer, “Relay margins as a tool for
dynamical security analysis,” Int. J. Electr. Power Energy Syst., vol. 12, Mladen Kezunovic (S’77-M’80–SM’85–F’99)
no. 4, pp. 226-234, Oct. 1990. received the Dipl. Ing. , M.S. and Ph.D. degrees in
[14] S. A. Soman, et al., "Analysis of angle stability problems: a electrical engineering in 1974, 1977 and 1980,
transmission protection systems perspective," IEEE Trans. Power Del., respectively. He is the Eugene E. Webb Professor
vol.19, no.3, pp.1024-1033, July 2004. and Site Director of Power Engineering Research
[15] K. Seethalekshmi, S. N. Singh, S. C. Srivastava, "A Classification Center (PSerc), an NSF I/UCRC at Texas A&M
Approach Using Support Vector Machines to Prevent Distance Relay University. He is also the Director of the Smart Grid
Maloperation Under Power Swing and Voltage Instability," IEEE Trans. Center. His main research interests are digital
Power Del., vol.27, no.3, pp.1124-1133, July 2012. simulators and simulation methods for relay testing, as well as application of
[16] V. Ajarappu, Computational Techniques for Voltage Stability intelligent methods to power system monitoring, control, and protection. He
Assessment and Control, New York: Springer, 2007. has published over 450 papers, given over 100 seminars, invited lectures and
[17] A. Esmaeilian, et al., "Evaluation and performance comparison of short courses, and consulted for over 50 companies worldwide. He is the
power swing detection algorithms in presence of series compensation on Principal of XpertPower Associates, a consulting firm specializing in
transmission lines," 10th International Conference on Environment and automated fault analysis and IED testing.
Electrical Engineering (EEEIC), 2011, vol., no., pp.1-4, 8-11 May Dr. Kezunovic is a Fellow of the IEEE, a Fellow of CIGRE and Registered
2011. Professional Engineer in Texas.
[18] R. Vaidyanathan and S. A. Soman, “Distance relay coordination
considering power swings,” Proc. Int. Conf. Power Syst. Commun. Syst.
Infrastructures for Future, 2002.
[19] Rep. “SIPPROTEC47SA6 Distance Protection Relay for all Voltage
Levels,” Siemens 4.3.2001, p-10.
[20] M. Tasdighi and M. Kezunovic, "Impact analysis of network topology
change on transmission distance relay settings," IEEE Power & Energy
Society General Meeting, vol., no., pp.1-5, 26-30 July 2015.
[21] D. M. MacGregor, A. T. Giuliante, and Patterson, “Automatic relay
setting,” J. of Elec. And Elect. Eng., vol. 21, no. 3, pp. 169-179, 2002.
[22] M. J. Damborg, R. Ramaswami, S.S. Venkata and J.M. Postforoosh,
“Computer aided transmission protection system design Part I &II,”
IEEE Vol. PAS 103, Jan 1984, pp. 51-59.
[23] R. Ramaswami, “Transmission Protective Relay Coordination- A
Computer-Aided-Engineering Approach for Subsystems and Full
Systems,” Ph.D Dissertation, University of Washington Seattle, January
1986.
[24] Rep. “Determination and Application of Practical Relaying Loadability
Rating Version 1,” NERC, June, 2008.
[25] O. Alsac, B. Stott, and W. F. Tinney, "Sparsity-Oriented Compensation
Methods for Modified Network Solutions," IEEE Trans. Power App.
and Sys., vol.PAS-102, no.5, pp.1050,1060, May 1983.
[26] Power System Test Case Archive, Univ. Washington, Dept. Elect. Eng.,
2013. [Online]. Available: https://www.ee.washington.edu/re-
search/pstca/index.html.
[27] Alberta Electric System Operator (AESO) [Online]. Available:
http://www.aeso.ca/transmission/261.html.

0885-8977 (c) 2015 IEEE. Personal use is permitted, but republication/redistribution requires IEEE permission. See http://www.ieee.org/publications_standards/publications/rights/index.html for more information.