IEEE TRANSACTIONS ON POWER DELIVERY, VOL. 27, NO.

3, JULY 2012 1415

Coordination of Directional Overcurrent

Relays Using Seeker Algorithm
Turaj Amraee

Abstract—Coordination of directional overcurrent relays in a problem should be modified to consider all possible scenarios.
multiloop subtransmission or distribution network is formulated The pickup setting of the relay remains between the upper and
as an optimization problem. In this paper, the coordination of di- lower values at which all short circuits in the line section are
rectional overcurrent relays is formulated as a mixed-integer non-
linear programming problem and is then solved by a new seeker seen. The pickup setting should be above the largest possible
optimization technique. Based on the act of human searching, in load current and below the minimum short-circuit current, with
the proposed seeker technique, the search direction and step length a security margin. Each relay has two main and backup zones.
are determined in an adaptive way. The proposed method is imple- The setting should also provide backup protection for neighbor
mented in three different test cases. The results are compared with feeders.
previously proposed analytic and evolutionary approaches.
The purpose of the time dial setting is to enable relays to co-
Index Terms—Coordination, directional overcurrent relay, opti- ordinate with each other. By providing a family of curves, two
mization, seeker algorithm.
or more relays, seeing the same fault, can operate at different
times. Therefore, the main goal of the coordination problem is
I. INTRODUCTION to find optimal time dial setting (TDS) and plug multiplier set-
ting (PMS) with a minimum operating time of primary relays
and under topological and transient changes of network con-

O VERCURRENT protection could be used as the primary

protection in distribution or subtransmission networks.
Directional overcurrent protection is necessary for ring or mul-
figurations. Therefore, there is an evident need to improve the
overcurrent coordination problem to find optimal and robust set-
tings under topological or operational changes in actual power
tiple source circuits to limit relay tripping for faults in only one systems. This need motivates the work reported in this paper.
direction. The directional relaying is carried out to provide relay
coordination between all of the relays that can see the same fault. B. Literature Review
The aim of the coordination problem is to find the optimal time
dial setting and current plug setting of the relays subject to some Various approaches have been proposed to solve the direc-
constraints. These constraints range from relay characteristics tional overcurrent coordination problem. They could be catego-
to the topological changes of the network. Generally, the co- rized in two different approaches. The first approach is based on
ordination problem is a mixed-integer nonlinear programming topological analysis, including graph-theoretic and functional
problem which could be solved by well-known mathematical or dependencies techniques. In graph-theoretic methods, the con-
evolutionary techniques. cept of breakpoint has been used [1]–[3]. In [1], the concept of
linear graph theory has been extended to identify and analyze all
A. Motivation and Problem Description simple loops of the network in both directions considering the
minimum set of breakpoints and all primary and backup relay
The coordination problem could be formulated as a linear or pairs. In the functional dependencies method, which is a system-
nonlinear programming problem. In the linear model, only the atic topological analysis technique, the constraints on the relay
time dial settings are optimized and the pickup current settings settings are formulated by a set of functional dependencies [3].
are fixed at values between the maximum load current and min- The second category is based on the optimization techniques
imum fault current. However, in the nonlinear programming ap- in which the coordination problem is formulated as a linear pro-
proach, based on the relay characteristic, time dial and pickup gramming problem and the TDSs are then determined via linear
current settings are optimized simultaneously. Also, the discrete programming solvers for fixed values of the pickup settings [4].
natures of the settings add more complexity to the coordina- The optimization problem could be solved by analytic
tion problem. Indeed, the coordination problem is a mixed-in- methods (e.g., Simplex method) [4]–[6] or evolutionary tech-
teger nonlinear programming problem. Due to the uncertainties niques [7]–[10]. In the linear programming approach, the
of load and topological changes, the obtained settings are not TDS is optimized. To optimize time dial and pickup current
optimal over all possible scenarios. Therefore, the coordination settings simultaneously, the coordination problem will be a
nonlinear programming model. Recently, the applications of
Manuscript received July 17, 2011; revised January 25, 2012; accepted Feb- evolutionary techniques have been used in optimal coordination
ruary 24, 2012. Date of publication April 10, 2012; date of current version June of directional overcurrent relays, genetic-algorithm (GA)-based
20, 2012. Paper no. TPWRD-00606-2011. methods [7]–[9], and the particle swarm optimization (PSO)
The author is with the Faculty of Power Engineering, K. N. Toosi University
of Technology, Tehran 1431714191, Iran (e-mail: [email protected]). algorithm [10], [11]. They have been used to solve the overcur-
Digital Object Identifier 10.1109/TPWRD.2012.2190107 rent coordination problem.
0885-8977/$31.00 © 2012 IEEE
1416 IEEE TRANSACTIONS ON POWER DELIVERY, VOL. 27, NO. 3, JULY 2012

The major weakness of the previously proposed methods, in- operating times of primary relays for faults at their associated
cluding mathematical and evolutionary approaches, lies in the zones. Therefore, the objective function is written as follows:
risk of being trapped in local optimal settings especially under
topological changes such as transient changes (i.e., the relay
opens at one end of the line section) in actual networks. (3)
C. Contributions
Any directional overcurrent coordination method should sat- Inverse time inverse definite minimum time (IDMT) character-
isfy the following requirements. istics are selectable from a choice of some IEC/IEEE curves
1) R1: The total operating times of primary relays should conforming to the following formula:
be minimized while maintaining coordination among all
relays.
(4)
2) R2: The pickup and time dial settings should be robust
under all possible topological and operational scenarios.
3) R3: The optimization technique should be capable of
where and are constant parameters. For IEC curves, is
finding the near-global optimal settings.
zero. For standard inverse-type relays, the IDMT characteristics
The gap that this paper intends to fill is the improvement of the
are assumed as 0.14. Another formula-
third requirements in addition to the other requirements. Re-
tion has been proposed to approximate the relay characteristic
garding this issue, the coordination problem is formulated as
called the Sachdev nonlinear model [5].
an optimization problem and is then solved by a new seeker al-
gorithm. In the proposed seeker technique, based on the act of
(5)
human searching, the search direction and related step length
are determined in an adaptive way. Also, the concept of seeker
where
mutation enables the seeker algorithm to escape from local op-
tima. The obtained results are compared with other previously
proposed analytic and evolutionary approaches. (6)

D. Paper Organization
The rest of this paper is organized as follows. In Section II, (7)
the structure and formulation of the directional overcurrent co-
ordination problem are described. The details of the new seeker
where and are constants. Tap settings is used to adjust
algorithm are described in Section III. The simulation results of
the minimum current in the relay for which the relay will just
applying the proposed method over three test systems are pre-
pick up. This provides great flexibility in the relay application
sented in Section IV. Finally, the conclusions are provided in
and permits the same relay to be used at many locations. The
Section V.
concept of relay pickup tap setting could be formulated by
II. PROBLEM FORMULATION (8)
The coordination of directional overcurrent relays in a multi-
loop system is formulated as an optimization problem. The coor- where is the primary pickup current and stands for the
dination problem, including objective function and constraints, CT ratio.
should satisfy all three aforementioned requirements.
B. Constraints
A. Objective Function
The coordination problem has two types of constraints, in-
The aim of the coordination problem is to minimize the total cluding the constraints of the relay characteristic and coordina-
weighted sum of operating times of primary relays as follows: tion constraints. Relay constraints include limits of relay oper-
ating time and settings. Coordination constraints are related to
(1) the coordination of primary and backup relays.
1) Constraint of Relay Operating Time: The operating time
where variables and stand for pickup tap setting of a relay is a function of the pickup current setting and the fault
and time dial setting of the th relay, respectively. The operating current seen by the relay. Based on the type of relay, the oper-
time of the th relay at location (i.e., ) is defined as ating time is determined via standard inverse curves or analytic
formula. The bounds on operating time are expressed by
(2)
(9)
where is the short-circuit current seen by the th relay for the
fault at location . It should be noted that the goal of the coor- where and are the minimum and maximum oper-
dination problem is to minimize the total weighted sum of the ating times of the th relay at the th location.
AMRAEE: COORDINATION OF DIRECTIONAL OVERCURRENT RELAYS USING SEEKER ALGORITHM 1417

2) Constraint of Time and Current Settings: The limits on A. Search Direction

TDS and PTS are expressed by In a population-based search, such as the human search, each
individual shows three different behaviors: 1) self-interest or
(10) egotistic behavior; 2) group-interest or altruistic behavior; and
3) proactiveness behavior [12]. Each seeker or individual in a
(11) population has two extreme types of cooperative behavior. The
egotistic behavior is entirely proself and the altruistic behavior
For phase relays, determining the maximum load current and
is entirely progroup [12]. The position of each seeker is updated
minimum fault current considering engineering experiences and
as follows:
planning studies for different network topologies is required.
The pickup current should be greater than the largest possible (14)
load currents and lower than the minimum fault current with
a reasonable security margin. This is the tradeoff between the where ,
security and dependability features of the relay. and refer to the optimization variable
3) Constraint of Coordination: The operating time of a with dimension , individual step length, and search direction,
backup relay should be selected to be greater than the operating respectively. The search direction has four components which
time of the related primary relay. An intentional time margin are determined as follows.
called the coordination time interval (CTI) is added to the • Personal Direction
operating time of the first backup relays. If for a given fault at In a population, each seeker is uniformly self-interested; in
location , the th relay may be the first backup of the th relay, other words, each seeker has a direction toward his or her
the coordination constraint could then be expressed by best position; this direction is expressed as follows:
(12) (15)
For transient topologies where one relay of a zone has op- where and are the current position and his-
erated, the coordination constraint should be satisfied without torical best position of the th seeker. is the search
considering the tripping sequence direction toward personal best position of the seeker.
• Local and global directions
(13)
Based on the simulation of social behavior of a groups of
where is the value of under transient topology. It should agents, each seeker has a progroup behavior. This cooper-
be noted that is determined via (4) in which the magnitude ative behavior could be represented by two directions to-
of (i.e., fault current magnitude in normal configuration) is ward the local and global best positions. The local best po-
replaced with (i.e., fault current magnitude in the transient sition is the best position in each subpopulation, and the
configuration). The value of varies from 0.2 to 0.3 s. global best position is the best position among all seekers
4) Topological Changes: By considering all probable sce- in all subpopulations. These two directions are formulated
narios of topological changes of the network, the optimal setting via (16) to (17).
is the setting that provides the minimum total operating time for
(16)
all possible configurations.
(17)
III. SEEKER OPTIMIZATION TECHNIQUE
where is the local best position of the th seeker in
Seeker optimization is a computational algorithm which im- the th subpopulation. is the global best position
itates the behavior of human searching considering its memory, in the population.
experience, uncertainty reasoning, and social learning [12]. • Historical direction
Some applications of the seeker algorithm could be found in Each seeker has a memory that retains the best experience.
[12]–[14]. All evolutionary techniques are initialized by gener- In other words, each seeker could adjust its future direction
ating a population of random individuals. Here, each individual toward the global solution based on the past experiences
of the population is called a seeker. Total search space or pop- obtained during searching the solution space. This proac-
ulation could be divided randomly into some subpopulations. tive learning behavior is modeled by
All subpopulations share their findings to escape from the local
optimum. The key parts of seeker optimization algorithm are (18)
the procedure of determining search direction and step length.
In analytic gradient-based optimization techniques, the search where , and is the better
direction and step length are determined based on the deriva- solution.
tives of the objective function and constraints. In evolutionary The overall direction is determined based on a random
optimization techniques, the search direction and step length combination of all directions. The actual search direction
are determined only based on the values of objective functions of the th seeker, (i.e., ), is deter-
with penalty terms related to constraints. The major parts of the mined based on a compromise among the aforementioned
seeker optimization technique will be described. four components of search direction. The overall direction
1418 IEEE TRANSACTIONS ON POWER DELIVERY, VOL. 27, NO. 3, JULY 2012

of the th element of the search direction is selected

by the following proportional rule [12]:
if
(19)

where refers to the th dimension of the th seeker

and is a uniform random number between , and
is defined as follows. In the set
is the percent of the
number or number of .

B. Step Length
Fig. 1. Mechanism of the intersubpopoulation learning.
In analytic gradient-based optimization algorithms, the step
length or learning factor could be adjusted in an adaptive way.
In other words, near the extremum, the step length should
be shorter and vice-versa. In evolutionary population-based
optimization techniques, such as the human search, the step
length toward the optimal point could be interpreted as a fuzzy
judgment. Here, in the fuzzification step, a linear membership
function and in the defuzzification step, a bell-type member-
ship function are used. To design a fuzzy rule for step-length
determination, all the fitness values in the population are sorted
from smallest to largest. A number is then assigned to each
seeker based on its rank in the population. This number (i.e., Fig. 2. Single-line diagram of the three-bus test case.
) is used as the input of the fuzzy system
with a linear membership function as follows: C. Seeker Mutation
Although the addition of the global search component to
the overall search direction reduces the probability of being
trapped in local optimum, it is beneficial to add an inter-
(20) population learning mechanism. Suppose that there are
where and are the membership value and the rank of the subpopulations. The worst seekers in each subpopulation
th seeker in the sorted population, respectively. The maximum are replaced by the best positions of the remaining
and minimum membership values occur at the global best and subpopulations. For example, if there are three subpopulations,
the worst positions in the population, respectively. Here, a bell the two worst seekers of each subpopulation are replaced with
membership function as is used. Based on the best seekers of the two remaining subpopulations as shown
the bell function, the membership values for the members out- in Fig. 1.
side the are fixed at . The value of as an
indicator of the search radius is a random vector which is deter- D. Steps of the Seeker Algorithm
mined as
The proposed seeker algorithm is carried out in the following
(21) steps.

where is the best seeker in the same subpopulation. The Step 1) Generate an initial population of seekers with a
weighting factor is used to decrease the step length toward size of .
the future Step 2) Split the population in different
subpopulations.
(22) Step 3) Set iteration 1.
Step 4) Evaluate the fitness function for each seeker.
where is the iteration number. Step 5) If the maximum number of iteration is reached,
Finally, the step length will be defuzzified using the bell func- then end; else continue.
tion as follows: Step 6) Determine the personal best, local best, and
global best positions.
(23) Step 7) Calculate the four components of the search
direction.
where generates a random number in the interval Step 8) Calculate the step length for all seekers.
. Step 9) Update the position of each seeker.
AMRAEE: COORDINATION OF DIRECTIONAL OVERCURRENT RELAYS USING SEEKER ALGORITHM 1419

TABLE I TABLE II
SHORT-CIRCUT RESULTS,THREE-BUS SYSTEM PICKUP TAP SETTING AND CT RATIO, THREE-BUS SYSTEM

TABLE III
TIME DIAL SETTINGS, THREE-BUS SYSTEM, LINEAR MODEL

Step 10) Mutate the population using the

intersubpopulation learning strategy.
Step 11) Construct the new population, increase the
iteration and go to Step 4).

IV. MINLP ALGORITHM TABLE IV

TDSS, THREE-BUS SYSTEM, MINLP MODEL
In this paper, the mixed-integer nonlinear programming is
solved using a standard branch-and-bound algorithm (SBB).
SBB is a new solver for mixed-integer nonlinear programming
(MINLP) models. It is based on a combination of the standard
branch-and-bound method known from mixed integer linear
programming and some of the standard NLP solvers already
supported by GAMS [15].

V. SIMULATION RESULTS
The proposed algorithm is simulated over three different test
cases. In each case, the obtained results are compared with pre-
viously proposed algorithms. The effectiveness of the proposed the primary relay has opened the circuit breaker at one end
seeker algorithm is verified for continuous and discrete (i.e., of the line. The value of CT ratio and related tap settings
mixed-integer) models. In this paper, the coordination problem is given in Table II. To make a clear comparison, the lower
is carried out for phase relays. It should be noted that for the limit of time dial setting is fixed at 0.1 s. Therefore, the
coordination of ground relays, the same procedure is followed. coordination problem is a linear programming problem.
Due to general similarity of the standard IDMT curves for both For this test case, the seeker parameters are assumed as
phase and ground relays (i.e., in view of linearity or nonlin- (0.95,0.0111), (0.1,0.9),
earity of the objective function and related constraints), the and . The results obtained by the proposed
simulations are carried out only for phase relays. A computer method and the previously proposed methods are given
program is developed to implement the seeker algorithm using in Table III. All three methods give the same results. Re-
Matlab, and executed on a Pentium IV 3.06-GHz PC with garding the convergence time, the time taken by the seeker
512-MB RAM. algorithm to reach the optimal setting is about 6.45 s after
45 iterations.
A. Case 1: Three-Bus Test Case • Mixed-integer nonlinear model
In this case, the proposed algorithm is simulated for a Now it is assumed that the pickup tap setting is a dis-
three-bus test case. The results are presented for linear (i.e., crete variable between 1.5–5 in steps of 0.5. The opti-
fixed PTS and continuous TDS) and mixed-integer nonlinear mization parameters are assumed as given in the previous
(i.e., discrete PTS and continuous TDS) models. case. The new results are given in Table IV. A standard
• Linear model branch-and-bound algorithm has been used to solve the re-
The proposed algorithm is simulated over a three-bus test sulting mixed-integer nonlinear programming problem. It
case. Data of this test case could be found in [4]. To make can be seen that the proposed seeker technique gives better
a clear comparison, the results of short-circuit analysis for settings rather than the analytic SBB procedure. The time
faults at the midpoint of lines are given in Table I. The taken by the seeker algorithm to reach the optimal setting
transient configuration refers to the configuration in which is about 10.45 s after 85 iterations.
1420 IEEE TRANSACTIONS ON POWER DELIVERY, VOL. 27, NO. 3, JULY 2012

TABLE VII
CT RATIO FOR THE RELAYS OF A 15-BUS SYSTEM

TABLE VIII
Fig. 3. Single-line diagram of the eight-bus test case. NEAR-END 3 SHORT-CIRCUIT RESULTS, 15-BUS SYSTEM

TABLE V
NEAR-END 3 SHORT-CIRCUIT RESULTS, EIGHT-BUS SYSTEM

TABLE VI
OPTIMAL TDS, 8-BUS SYSTEM

cuit faults is given in Table V. For this test case, the seeker
parameters are assumed as
and 100. According to
Table VI, it can be seen that the proposed method gives a lower
B. Case 2: Eight-Bus Test Case total operating time for primary relays. The time taken by the
The proposed algorithm is simulated over an eight-bus test seeker algorithm to reach the optimal setting is about 50.45 s
system. The system data could be found in [10]. The test case after 169 iterations.
has 14 relays as shown in Fig. 3. The TDSs range continuously
C. Case 3: 15-Bus Test Case
from 0.1 to 1.1. To make a clear comparison, seven discrete
pickup tap settings (i.e., 0.5, 0.6, 0.8, 1.0, 1.5, 2.0, and 2.5) are The proposed technique is implemented in a 15-bus test net-
assumed. The ratio of current transformer of relays (1, 2, 4, 5, work. This case is a highly distributed generation (DG) pene-
6, 8, 10, 11, 12, 13) and (3,7, 9, 14) are assumed as (1200:5) trated distribution network as shown in Fig. 4. Each generator
and (800:5), respectively. A coordination interval of 0.3 s is has a synchronous reactance of 15% with 15 MVA and
considered. The short-circuit current for near-end short-cir- 20-kV ratings. The external grid has 200-MVA short-circuit
AMRAEE: COORDINATION OF DIRECTIONAL OVERCURRENT RELAYS USING SEEKER ALGORITHM 1421

Fig. 6. Variations of the best solution found for 100 different program runs,
15-bus system.

TABLE IX
OPTIMAL SETTINGS OF THE 15-BUS SYSTEM

Fig. 4. Single-line diagram of the 15-bus network.

TABLE X
OPTIMAL SETTINGS OF THE 15-BUS SYSTEM, CONTINUED

Fig. 5. Convergence of the seeker algorithm to the final solution, case 3.

capacity. The reactance of each line section is

km. The short-circuit analysis for near-end faults
and based on the International Electrotechnique Commission
standard are given in Table VIII. The test case has 42 relays and
82 backup-primary pairs. It is assumed that the TDS is greater
than 0.10 s. The pickup tap setting for current pickup varies be-
tween 0.5 to 2.5 in steps of 0.5. The resulting mixed-integer non-
linear programming problem is solved using a standard branch-
and-bound technique and a new seeker algorithm. For this test
case, the seeker parameters are assumed as
, , and 120.
The optimal settings found by two approaches are given in Ta- of mixed-integer variables, the proposed seeker optimization
bles IX and X. It could be deduced that in case technique yields better results. While most of the mathematical
1422 IEEE TRANSACTIONS ON POWER DELIVERY, VOL. 27, NO. 3, JULY 2012

approaches are guaranteed to converge only under special con- [5] B. Chattopadhyay, M. Sachdev, and T. Sidhu, “An on-line relay coor-
vexity assumptions, the proposed seeker algorithm could pro- dination algorithm for adaptive protection using linear programming
technique,” IEEE Trans. Power Del. , vol. 11, no. 1, pp. 165–173, Jan.
vide superior results for mixed-integer nonlinear programming 1996.
problems. The computational time to reach the optimal setting [6] H. Abyaneh, M. Al-Dabbagh, H. Karegar, S. Sadeghi, and R. Khan, “A
new optimal approach for coordination of overcurrent relays in inter-
is about 406.3 s after 385 iterations. Also, the convergence of connected power systems,” IEEE Trans. Power Del., vol. 18, no. 2, pp.
the seeker algorithm is illustrated in Fig. 5. One major advan- 430–435, Apr. 2003.
tage of the seeker algorithm is its robustness for different runs [7] F. Razavi, H. A. Abyaneh, M. Al-Dabbagh, R. Mohammadi, and H.
Torkaman, “A new comprehensive genetic algorithm method for op-
of the algorithm. Based on Fig. 6, it can be seen that the best timal overcurrent relays coordination,” Elect. Power Syst. Res., vol. 78,
solution could be found in a significant number of trials. no. 4, pp. 713–720, 2008.
[8] P. Bedekar and S. Bhide, “Optimum coordination of directional over-
current relays using the hybrid GA-NLP approach,” IEEE Trans. Power
VI. CONCLUSION Del., vol. 26, no. 1, pp. 109–119, Jan. 2011.
[9] A. Noghabi, J. Sadeh, and H. Mashhadi, “Considering different net-
Coordination of directional overcurrent relays was formu-
work topologies in optimal overcurrent relay coordination using a hy-
lated as a mixed-integer nonlinear programming problem and brid GA,” IEEE Trans. Power Del, vol. 24, no. 4, pp. 1857–1863, Oct.
was then solved by a new seeker optimization technique. In 2009.
the proposed seeker technique, based on the act of human [10] H. Zeineldin, E. El-Saadany, and M. Salama, “Optimal coordination
of overcurrent relays using a modified particle swarm optimization,”
searching, the search direction and step length were determined Elect. Power Syst. Res., vol. 76, no. 11, pp. 988–995, 2006.
in an adaptive way. The proposed technique was applied over [11] M. Mansour, S. F. Mekhamer, and E.-S. N. El-Kharbawe, “A modified
three different cases. The effectiveness of the proposed tech- particle swarm optimizer for the coordination of directional overcurrent
relays,” IEEE Trans. Power Del., vol. 22, no. 3, pp. 1400–1410, Jul.
nique was verified for continuous and discrete variables. The 2007.
results showed that the proposed seeker technique is capable of [12] C. Dai, Y. Zhu, W. Chen, and X. Zhang, “Seeker optimization algo-
finding superior TDS and PMS settings in linear and nonlinear rithm for optimal reactive power dispatch,” IEEE Trans. Power Syst.,
vol. 24, no. 3, pp. 1218–1231, Aug. 2009.
models. The obtained results verified that the proposed seeker [13] B. Shaw1, V. Mukherjee, and S. Ghoshal, “Seeker optimisation al-
covers the weakness of the previously proposed evolutionary gorithm: Application to the solution of economic load dispatch prob-
lems,” Inst. Eng. Technol. Gen., Transm. Distrib., vol. 5, no. 1, p. 8191,
techniques.
Aug. 2011.
[14] C. Dai, Y. Zhu, and W. Chen, “Seeker optimization algorithm for dig-
REFERENCES ital IIR filter design,” IEEE Trans. Ind. Electron., vol. 57, no. 5, pp.
1710–1718, May 2010.
[1] M. Damborg, R. Ramaswami, S. Venkata, and J. Postforoosh, “Com-
[15] Gams Release 2.50, A Users Guide. 1999. [Online]. Available: http://
puter aided transmission protection system design part I: Alcorithms,”
www.gams.com
IEEE Trans. Power App. Syst., vol. PAS-103, no. 1, pp. 51–59, Jan.
1984.
[2] M. H. Dwarakanath and L. Nowitz, “An application of linear graph Turaj Amraee was born in Loristan, Iran, in 1979.
theory for coordination of directionalovercurrent relays,” in Proc. He received the Ph.D. degree in power system
Elect. Power Problems—Math. Challenge Conf., Seattle, WA, 1980, engineering from Sharif University of Technology,
no. 1, pp. 104–114, SIAM. Tehran, Iran, and Grenoble Institute of Technology,
[3] L. Jenkins, H. Khincha, S. Shivakumar, and P. Dash, “An application Grenoble, France, in 2011.
of functional dependencies to the topological analysis of protection Currently, he is an Assistant Professor in the Elec-
schemes,” IEEE Trans. Power Del., vol. 7, no. 1, pp. 77–83, Jan. 1992. trical Engineering Department, K.N. Toosi Univer-
[4] A. Urdaneta, R. Nadira, and L. P. Jimenez, “Optimal coordination of sity of Technology, Tehran. His research interests are
directional overcurrent relays in interconnected power systems,” IEEE power system dynamics, system protection schemes,
Trans. Power Del., vol. 3, no. 3, pp. 903–911, Jul. 1988. blackouts, and power system operations.