You are on page 1of 9

IEEE TRANSACTIONS ON POWER DELIVERY, VOL. 26, NO.

4, OCTOBER 2011 2171

A Reliability-Centered Approach to an Optimal


Maintenance Strategy in Transmission Systems
Using a Genetic Algorithm
Jae-Haeng Heo, Mun-Kyeom Kim, Geun-Pyo Park, Yong Tae Yoon, Member, IEEE,
Jong Keun Park, Senior Member, IEEE, Sang-Seung Lee, Member, IEEE, and Dong-Hyeon Kim

Abstract—Electric power transmission utilities try to maximize d12, d22 Decisions to do maintenance.
profit by reducing electricity supply costs and operation costs while
maintaining their reliability. Developing maintenance strategies is d11, d21 Decisions to do no maintenance.
one of the effective ways to achieve these profitable goals. The re-
liability-centered maintenance approach is a key method in pro- J Total expected cost.
viding optimal maintenance strategies. It considers the tradeoffs MC Maintenance cost.
between the upfront maintenance costs and the potential costs of
reliability losses. Since a transmission system is a group of different RC Repair cost.
kinds of equipment and the reliability of the electric facilities varies
with time, an equipment state model using a modified semi-Markov GC Generation cost.
chain is proposed. In addition, a genetic algorithm is used to find OC Outage cost.
the optimal maintenance strategies from a large class of possible
maintenance scenarios. These optimal maintenance strategies have th system state.
been tested on an IEEE 9-bus system and an IEEE 118-bus system;
the results show that the proposed method minimizes the total ex- Line flow vector for the system state .
pected costs.
LC Load curtailment.
Index Terms—Equipment state model, genetic algorithm (GA),
modified semi-Markov chain, reliability-centered maintenance PG Generation vector that has element .
(RCM).
PD Load vector that has element .
Lower vectors for the generator.
NOMENCLATURE
Upper limit vectors for the generator.
N Normal state. Transmission capacity.
D1, D2 Deterioration states. NL Number of load buses.
M1 Weak maintenance. NG Number of generator buses.
M2 Strong maintenance/refurbishment. Search space.
F Failure states. Decision vector.
S Real-time sensor. Ps Population size.
T Random variable.
Manuscript received April 27, 2010; revised October 08, 2010; accepted June
05, 2011. Date of publication September 01, 2011; date of current version Oc- R Uniformly distributed random number .
tober 07, 2011. This work was supported in part by KESRI and in part by the
Human Resources Development of the Korea Institute of Energy Technology CDF(t) Cumulative distribution function.
Evaluation and Planning (KETEP), which are funded by the Korea Govern-
ment Ministry of Knowledge Economy (MKE). (No. 2011-0539). Paper no.
t Transition time from the present state to the next
TPWRD-00301-2010. state.
J. H. Heo, Y. T. Yoon, G. P. Park, and J. K. Park are with the School
of Electrical Engineering, Seoul National University, Seoul 151-742, Transition rate.
Korea (e-mail: parkjk@snu.ac.kr; ytyoon@snu.ac.kr; pgp7926@snu.ac.kr;
jhheo78@snu.ac.kr). Repair rate.
M. K. Kim is with the Department of Electrical Engineering, Dong-A Uni-
versity, Busan 604-714, Korea (e-mail: mkkim@dau.ac.kr).
S. S. Lee is with the Korea Electrical Engineering and Science Research I. INTRODUCTION
Institute (KESRI), Seoul National University, Seoul 151-742, Korea (e-mail:
sslee6@snu.ac.kr).
D.-H. Kim is with the Burns & McDonnell Engineering Company, Inc.,
Kansas City, MO 64030 USA (e-mail: dkim@burnsmcd.com).
Color versions of one or more of the figures in this paper are available online
at http://ieeexplore.ieee.org.
I N the deregulated power market, electric power transmis-
sion utilities need to reduce the electrical supply costs and
the operational costs in order to achieve price competitiveness.
Digital Object Identifier 10.1109/TPWRD.2011.2162752 At the same time, they need to sustain system reliability, which
0885-8977/$26.00 © 2011 IEEE
2172 IEEE TRANSACTIONS ON POWER DELIVERY, VOL. 26, NO. 4, OCTOBER 2011

is a critical factor for the power system industry. One of the


ways to decrease costs while improving the reliability of the
power systems is to utilize the reliability-centered maintenance
(RCM) method. The RCM approach is the key that provides op-
timal maintenance strategies; it considers the tradeoffs between
the upfront maintenance costs and the potential costs resulting
from losing loads [1]–[5].
Since a transmission system is composed of combinations of
different kinds of aging equipment, all of the elements in a trans-
mission system have different life spans. Even the life spans
of the same equipment may vary due to the different condi-
tions, so it is difficult to follow an equipment-based maintenance
strategy. Many studies have discussed maintenance [6]–[11].
Reference [6] presented a generating unit maintenance schedule
for two interconnected power systems based on the load car- Fig. 1. Basic equipment state model with the modified semi-Markov chain.
rying capability of a unit, and reference [7] proposed a gener-
ator maintenance schedule with transmission constraints. In [8],
a load- and cost-based generating unit maintenance schedule maintenance. The equipment state model is compatible to dif-
was presented that uses fuzzy evolutionary programming. The ferent equipment or systems by adjusting the number of states.
optimal coordination of maintenance scheduling for generating
units and transmission lines was proposed in [9]. However, all A. Basic Equipment State Model
of these papers focused only the generating unit maintenance A semi-Markov chain determines the varying states by proba-
schedule (GMS). In [10], the authors proposed the RCM method bility [12]. This paper proposes an equipment state model using
to be used for transmission systems; however, this particular a modified semi-Markov chain. The proposed equipment state
study does not consider the impact of the maintenance strategy model is shown in Fig. 1. As shown in Fig. 1, D1 and D2 are
on system reliability. Reference [11] focused on the quantified the deterioration states when the equipment ages over time. M1
impact of planned transmission outages on the overall system and M2 are the maintenance methods dependent on the deterio-
reliability, but it does not use a state model that varies with ration state of the equipment. If any maintenance is performed,
time. These studies show the difficulty and the limitations of the equipment state will go back to , which represents the
achieving a maintenance strategy with different kinds of aging normal state. The inspector decides whether to perform main-
equipment. tenance; designating performed maintenance as d12 or d22, or
In order to consider all of the differences in the equipment not as d11 or d21. D2 is different from F. The D2 state repre-
in the system, and to analyze the negative impact of equipment sents that even though all two elements are malfunctioning, the
failure, it is essential to have a state model that has sequences ad- equipment is still in service. However, the F state represents an
dressing equipment aging, failures, and maintenance. This paper out-of-service condition of the equipment.
proposes a modified semi-Markov chain as an equipment state The decisions are made to minimize the total expected cost,
model. which consists of the sum of the customer outage cost, the main-
Since a greater amount of equipment in a maintenance tenance cost, the repair cost, and the generation cost. In the cost
strategy causes an exponential increase in the computer estimation, the equipment state and the expected impact of the
memory requirements, the size of a simulated transmission equipment state to the entire system are considered in order to
system may be limited. In these cases, a heuristic approach is make these decisions. This is the core point of this model.
useful to find a feasible solution close to the optimal solution. A
genetic algorithm (GA) is proposed in order to find the optimal B. Applications to Actual Equipment
maintenance strategies efficiently. The equipment state model is adjusted by changing the
This paper is organized as follows. Section III presents the number of deterioration states. The number of deterioration
equipment state model and Section IV presents the GA process states is defined as the actual equipment aging stages and the
used for the optimal maintenance strategies that minimize the possible maintenance methods. This enables the equipment
total expected cost. A case study is performed on an IEEE 9-bus state model to represent different kinds of equipment in the
system and an IEEE 118-bus system and then compared to the transmission system.
existing time-based method in Section V. Section VI presents In this paper, an overhead line, an insulator, and a tower are
the conclusions. selected as the study models. These three elements are the main
components of a transmission system; the failure of this equip-
ment is related to an entire system failure.
II. EQUIPMENT STATE MODEL
These components have different features and deterioration
In order to build a maintenance strategy for the diverse aging processes. Therefore, the equipment state model needs to be ad-
equipment found in a transmission system, the equipment state justed in accordance with the equipment’s characteristics. The
model is used. The equipment state model uses several states deterioration states in this paper are defined based on the inspec-
to represent the sequence of the equipment aging, failure, and tion criteria used in the transmission utilities. The selection of
HEO et al.: RELIABILITY-CENTERED APPROACH TO AN OPTIMAL MAINTENANCE STRATEGY 2173

TABLE I
STATE CRITERIA OF OVERHEAD LINES

TABLE II
STATE CRITERIA OF INSULATORS

TABLE III
STATE CRITERIA OF TRANSMISSION TOWERS

Fig. 2. Algorithm to find the optimum maintenance strategy using the GA.

The decision vector that minimizes the total expected cost


will be the optimal maintenance strategy for the equipment.

B. Process of Finding the Optimal Maintenance Strategy


the deterioration states may vary with the transmission utility Using the GA
company; however, reasonable assumptions for the equipment
state model were made for this study and summarized in Tables I The algorithm used to find the optimum maintenance strategy
to III. using the GA is shown in Fig. 2. First, the maintenance deci-
sion vector is generated for all of the equipment, and the fitness
of each decision vector is evaluated. Next, the decision vectors
III. OPTIMAL MAINTENANCE STRATEGY USING THE GA are selected from the population to be parents that cross over
through the fitness-based process. Then, crossover and muta-
The GA [13] was selected in order to find the optimal mainte-
tion are adopted. After the maximum numbers of iterations are
nance strategy that minimizes the total expected cost. The equip-
reached, the optimal decision vector is defined to minimize the
ment state and the expected impact of the equipment failure to
total expected cost. This decision vector includes the optimal
the transmission system are considered.
maintenance strategy for all of the studied equipment.
A. Problem Formulation
The total expected cost (J) is the summation of the mainte- C. Generating Population
nance cost (MC), the repair cost (RC), the GC, and the OC as The maintenance decision vector is randomly generated for
seen in (1). The MC and RC are assumed to be based on the an- all of the equipment according to the population size (Ps).
nual cost of the transmission utility companies For example, suppose there are four pieces of equipment.
Two of them have the state model which consists of 3-deteri-
(1) oration states, while the others have the state model which con-
tains 2-deterioration states. Since two pieces of equipment have
In (2), the maintenance decision vector is a chromosome of 3-deterioration states and the others have 2- deterioration states,
the search space the decision vector will be -bit, as shown in
Fig. 3; the full search space will be . The 10-b deci-
(2) sion vector is randomly generated equaling the population size.
2174 IEEE TRANSACTIONS ON POWER DELIVERY, VOL. 26, NO. 4, OCTOBER 2011

Fig. 3. Construction of the decision vector.

Fig. 4. Example of the decision vector.

If any decision vector among the population is like that shown


in Fig. 4, this decision vector (with “1” representing doing main-
tenance, and “0” no maintenance) indicates that the operator
only needs to perform strong maintenance on the first compo-
nent when it is in the D3 state. The second component does not
need any maintenance. The third component has a weak main-
tenance that is only carried out when the state is D1; the fourth
component needs maintenance action in the D1 and D2 states.
This shows that the decision vector includes the information of
the maintenance timing and the method (strong or weak main-
tenance) needed for all of the equipment.

D. Evaluation of the Fitness Function


The total expected cost as the fitness function is evaluated
for each decision vector. The procedure for the fitness function
with a state transition pattern used for the calculation of the total
expected cost is shown as follows.
Step 1) Set the decision vector counter (Dec).
Step 2) Specify the initial state of all the equipment. In this
study, the equipment is initially in the normal state.
Step 3) Estimate the failure patterns and the duration of the
equipment residing in its present state chronolog-
ically using a sequential Monte Carlo Simulation
[14]. For all of the equipment, random numbers (R)
are generated, and the time-to-state transition time Fig. 5. Process of evaluating the total expected cost.
and the repair time (T) are calculated chronologi-
cally using
TABLE IV
LOCATION OF THE EQUIPMENT IN THE IEEE 9-BUS SYSTEM
(3)

where is a uniformly distributed random number


, is the transition rate, and the random vari-
able is the time-to-state transition time.
The duration of the equipment state transition is then
determined as shown

(4)

Step 4) Transit the state of all the equipment based on the


derived failure patterns and the state transition time.
HEO et al.: RELIABILITY-CENTERED APPROACH TO AN OPTIMAL MAINTENANCE STRATEGY 2175

For example, if any equipment has 0.1 and R


is randomly generated to be 0.5, then the transition
time (T) from the state to the D1 state is 1.34
years, as determined by (3). After 1.34 years, the
state of this equipment turns into the D1 state. For
the transition rate and repair rate, were as-
sumed as shown in the Appendix.
Step 5) Check the system topology change caused by equip-
ment failure or repair completion. If anything
changed, it might probably be the load curtailment
or even an entire system outage or a change of the
GC. Our study calculates the impact of the equip-
ment failure by using the GC and OC. The DC-OPF
is expressed as follows:

(5)

(6)
Fig. 6. Result of the GA process for the optimal maintenance strategy.

(7)
(8) F. Crossover and Mutation
(9) Under the binary coding mode, a one-point crossover is a
very effective crossover operation, and so is adopted into the GA
Here, GC is evaluated as the function of process [13]. Mutation is defined by choosing decision vectors
the electrical output of each generator, and stochastically from the population under a small probability of
the OC is calculated by using amounts of mutation, then choosing one bit or some bits of the chosen chro-
load curtailment ( curtailment mosomes stochastically and reversing their values, from 1 to 0
time cost multiplier ). or from 0 to 1. The mutation probability used is 0.01.
is the electrical output of the th generator IV. CASE STUDY AND RESULTS
and is the amount of load curtailment of the
th load. The power balance constraint is shown in The effectiveness of the proposed maintenance strategy was
(6). Equations (7) and (8) are generation limits and demonstrated using an IEEE 9-bus system [15] and an IEEE
load curtailment limits, respectively. In (9), 118-bus system [16]. The test data (bus data, line parameters,
is the line flow vector for the th system state and generation cost) for these two test systems are taken from
and this equation expresses the line capacity limit. the MATPOWER toolbox [17]. The results were then compared
The RC of the failed equipment is also calculated to the conventional time-based maintenance strategies.
based on the repair cost for each piece of equipment A. IEEE 9-Bus System Results
shown in Table XII.
The IEEE 9-bus system consists of 9 lines, 3 generators, and
Step 6) Checkthemaintenanceorder fromthedecisionvector.
3 load buses with a total real consumption of 315 MW. All of
If any maintenance decision has been made, the MC
the equipment is assumed to be installed between the buses, as
is added to the total expected cost. The MC is deter-
shown in Table IV. In reality, many towers and insulators are
mined based on the deterioration states. The state of
installed between two buses; however, only a tower and an in-
the equipment under maintenance will be changed to
sulator are assumed in order to simplify this study. Any failure
in Step 4 after a given maintenance time.
in the three components causes a transmission-line disconnec-
Step 7) Repeat Steps 4)–6) in a given time span.
tion between two buses. The simulation data (transition rate, re-
Step 8) Update the decision vector counter
pair rate, maintenance cost, and outage cost) are shown in the
and repeat steps according to the population
Appendix.
size.
This system has nine lines, nine towers, and nine insulators.
The process of evaluating the total expected cost is shown in
Since the towers and insulators have 2 deterioration states and
Fig. 5.
the overhead lines have 3 deterioration states, the decision
vector is -bit and the search spaces
E. Selection are .
The GA process is shown in Fig. 6; the generation number is
The decision vectors are selected from the population to be 100, the population size for each generation is 200, the -axis
parents that cross over through a fitness-based process. Rank represents the number of generation, and the -axis is the total
selection is applied in order to select these decision vectors. expected cost. The triangles and the dots represent the mean and
2176 IEEE TRANSACTIONS ON POWER DELIVERY, VOL. 26, NO. 4, OCTOBER 2011

TABLE V TABLE VI
OPTIMAL MAINTENANCE STRATEGY AND THE FAILURE REPORT TOTAL EXPECTED COST OF THE PROPOSED MAINTENANCE STRATEGY
FOR THE IEEE 9-BUS SYSTEM IN 30 YEARS AND TIME-BASED MAINTENANCE STRATEGIES FOR THE IEEE
9-BUS SYSTEM IN 30 YEARS

616 over a 30-year time span. The optimal maintenance


strategy was found at the 64th generation.
Table V shows the optimal maintenance strategy selected by
the GA and the failure frequency and the calculation time for the
30 simulation years. It minimizes the total expected cost. This
table shows the maintenance plan for all of the equipment.
The optimal maintenance plan is suggested from the decision
vector shown in Table V. For example, since the optimal mainte-
nance plan for line is [Nothing, Do M2, Do M3], this directs
the inspector to perform maintenance M2 in the D2 state and
M3 in the D3 states; the total expected cost goes to a minimum.
If no maintenance is suggested, such as for tower , the failure
frequency changes; this represents the number of failures that
will occur during the simulation years.
When the maintenance actions are performed following these
parameters in 30 years, the total expected cost is 1,371,457,
616 which consists of the GC, MC, RC, and the OC. The
OC has not occurred to avoid the load curtailment in the op-
timal maintenance plan over 30 years. Moreover, the equipment
failure will occur once at tower .
Table VI shows the total expected cost of the proposed main-
tenance strategy and compares it to the conventional time-based
maintenance. J-1 represents the total cost of performing main-
tenance every year. J-2 and J-3 show the total costs for 2 years
and 3 years maintenance, respectively. J-v indicates the total ex-
pected cost when the lines are maintained every 3 years, the in-
sulators are done every 2 years, and the towers are done every 5
years. J is lower than the time-based maintenance (J-1, J-2, J-3,
and J-v). This illustrates that the proposed maintenance strategy
is more cost effective than the existing periodic maintenance
strategies for all of the equipment.
Fig. 7 shows the comparison of the total expected cost be-
tween the proposed method and the time-based method with
different simulation years. The normalized total expected cost
(in per unit) value with the total expected cost of the proposed
method is used. The plot shows that the total expected cost of
the proposed method is always lower than the time-based main-
tenance cost (J-1, J-2, J-3, and J-v).
For the sake of verifying that the proposed method is always
cost efficient through a change of the system, such as a load in-
crease occurring, the total expected cost of the proposed method
is compared to J-v with the assumption of several cases. Case 1
is the situation where the load of bus # 5 turns 125 into 150 MW
minimum values of the total expected cost for each generation, and Case 2 is when the load of bus #6 changes 90 to 120 MW.
respectively. The minimum total expected cost is 1,371,457, In a similar manner, Cases 3–5 are explained in Table VII.
HEO et al.: RELIABILITY-CENTERED APPROACH TO AN OPTIMAL MAINTENANCE STRATEGY 2177

TABLE VIII
LOCATION OF THE EQUIPMENT IN THE IEEE 118-BUS SYSTEM

Fig. 7. Comparison of the total expected cost (in per unit) between the proposed
method and the time-based method for the IEEE 9-bus system.

TABLE VII
COMPARISON OF THE TOTAL EXPECTED COST
BETWEEN THE PROPOSED METHOD AND J-V FOR THE
IEEE 9-BUS ACCORDING TO A CHANGE OF LOAD

Table VII shows that J is lower than J-v, and the proposed
method is more cost efficient according to load changes.

B. IEEE 118-Bus System Results


The IEEE 118-bus system has 54 generators, 186 lines, and 91
loads. All of the equipment is assumed to be installed between
the buses, as shown in Table VIII.
The IEEE 118-bus system has 15 overhead lines, 10 insula-
tors, and 15 transmission towers. Therefore, the decision vector
is b and the search spaces are
2 . Table IX shows the optimal maintenance strategy, the failure
frequency, and the calculation time. Table X shows the compar-
isonsofthetotalexpectedcostbetween theproposedmaintenance
strategies and the conventional time-based maintenance after the
GA process (the generation number is 100, the population size for
each generation is 200, and the given time span is 30 years). Fig. 8
shows the comparison of the total expected cost (in per unit) ap-
plied to different simulation years. The result shows that the total
expected cost of the proposed method is always lower than the Fig. 8. Comparison of the total expected cost (in per unit) between the proposed
method and the time-based method for the IEEE 118-bus system.
time-based maintenance for the different time spans.

V. CONCLUSION
strategies for different kinds of aging equipment. The notable
In this paper, we proposed an equipment state model using characteristics of this model are:
a modified semi-Markov chain to find the optimal maintenance • modeling aging equipment with deterioration states;
2178 IEEE TRANSACTIONS ON POWER DELIVERY, VOL. 26, NO. 4, OCTOBER 2011

TABLE IX TABLE X
OPTIMAL MAINTENANCE STRATEGY AND THE FAILURE REPORT FOR THE TOTAL EXPECTED COST OF THE PROPOSED MAINTENANCE
IEEE 118-BUS SYSTEM IN 30 YEARS STRATEGY AND TIME-BASED MAINTENANCE STRATEGIES FOR THE
IEEE 118-BUS SYSTEM IN 30 YEARS

TABLE XI
TRANSITION RATES FOR THE EQUIPMENT

TABLE XII
REPAIR RATES FOR THE EQUIPMENT

TABLE XIII
MAINTENANCE COST AND THE OUTAGE COST

• adaptability for different equipment by adjusting the


number of deterioration states based on the inspection
criteria;
• compatibility with a complex system which has various
types of equipment, such as transmission systems.
The failure patterns of the equipment were simulated with a
sequential Monte Carlo method using an equipment state model.
The GA was applied to efficiently find the optimal maintenance
strategies. The proposed maintenance strategy was proven to be
effective by comparing it to the conventional time-based main-
tenance strategies.

APPENDIX
Tables XI and XII show the transition rates and the repair
rates for the equipment, respectively. Table XIII represents the
maintenance cost and the outage cost.
HEO et al.: RELIABILITY-CENTERED APPROACH TO AN OPTIMAL MAINTENANCE STRATEGY 2179

REFERENCES Geun-Pyo Park was born in Korea in 1979. He


[1] M. Rausand, “Reliability centered maintenance,” Rel. Eng. Syst. Safety, received the B.S. and M.S. degrees in electrical
vol. 60, pp. 121–132, 1998. engineering from Seoul National University, Seoul,
[2] G. Levitin and A. Lisnianski, “Joint redundancy and maintenance opti- Korea, where he is currently pursuing the Ph.D.
mization for multistate series-parallel systems,” Rel. Eng. Syst. Safety, degree in electrical engineering.
vol. 64, no. 1, Apr. 1999. His research interests include power system
[3] S. D. Wu and Clements-Croome, “Optimal maintenance policies under reliability.
different operational schedules,” IEEE Trans. Rel., vol. 54, no. 2, pp.
338–346, Jun. 2005.
[4] B. S. Hauge and D. C. Johnston, “Reliability centered maintenance and
risk assessment,” in Proc. Annu. Rel. Maintain. Symp., Jan. 2001, pp.
22–25.
[5] I. P. Siqueira, “Optimum reliability-centered maintenance task fre-
quencies for power system equipments,” in Proc. 8th Int. Conf.
Probabilistic Methods Applied to Power Systems, pp. 162–167. Yong Tae Yoon (M’11) was born in Korea on
[6] F. A. El-Sheikhi and R. Billinton, “Generation unit maintenance sched- April 20, 1971. He received the B.S., M. Eng., and
uling for single and interconnected systems,” IEEE Trans. Power App. Ph.D. degrees from the Massachusetts Institute of
Syst., vol. PAS-103, no. 5, pp. 1038–1044, May 1984. Technology, Cambridge, in 1995, 1997, and 2001,
[7] E. L. da Silva, M. T. Schilling, and M. C. Rafael, “Generation main- respectively.
tenance scheduling considering transmission constraints,” IEEE Trans. Currently, he is an Associate Professor in the
Power Syst., vol. 15, no. 2, pp. 838–843, May 2000. School of Electrical Engineering and Computer Sci-
[8] M. Y. El-Sharkh and A. A. El-Keib, “Maintenance scheduling of gen- ence, Seoul National University, Korea. His research
eration and transmission systems using fuzzy evolutionary program- interests include electric power network economics,
ming,” IEEE Trans. Power Syst., vol. 18, no. 2, pp. 862–866, May 2003. power system reliability, and the incentive regulation
[9] Y. Fu, M. Shahidehpour, and Z. Li, “Security-constrained optimal co- of independent transmission companies.
ordination of generation and transmission maintenance outage sched-
uling,” IEEE Trans. Power Syst., vol. 22, no. 3, pp. 1302–1313, Aug.
2007.
[10] M. E. Beehler, “Reliability centered maintenance for transmission sys-
tems,” IEEE Trans. Power Del., vol. 12, no. 2, pp. 1023–1028, Apr. Jong Keun Park (SM’97) was born in the Republic
1997. of Korea in 1952. He received the B.S. degree in
[11] W. Li and J. Korczynski, “A reliability based approach to transmission electrical engineering from Seoul National Uni-
maintenance planning and its application in BC hydro system,” IEEE versity, Seoul, Korea, in 1973 and the M.S. and
Trans. Power Del., vol. 19, no. 1, pp. 303–308, Jan. 2004. Ph.D. degrees in electrical engineering from The
[12] G. C. Christos and S. Lafortune, Introduction to Discrete Event Sys- University of Tokyo, Tokyo, Japan, in 1979 and
tems. Norwell, MA: Kluwer, 1999. 1982, respectively.
[13] B. R. Moon, Genetic Algorithm. Seoul, Korea: Doo Yang Sa, 2005, Since 1983, he has been an Assistant Professor, an
pp. 2–68. Associate Professor, and a Professor with the School
[14] R. Billinton and R. N. Allan, Reliability Evaluation of Power Sys- of Electrical Engineering, Seoul National University.
tems. New York: Plenum, pp. 400–405. His research areas are power system economics and
[15] R. W. Chang and T. K. Saha, “Maximizing power system loadability by energy policy, and power system analysis.
optimal allocation of SVC using mixed integer linear programming,” Prof. Park is a Fellow with the Institution of Electrical Engineers (IEE). Also,
presented at the Power Energy Soc. Gen. Meeting, Minneapolis, MN, he is the Vice President of the Korean Institute of Electrical Engineers (KIEE),
2010. a member of the National Academy of Engineering of Korea, and the Korean
[16] T. K. Hahn, M. K. Kim, D. Hur, J.-K. Park, and Y. T. Yoon, “Evalu- representative of the study committee SC5 “Electricity Markets and Regulation”
ation of available transfer capability using fuzzy multi-objective con- in CIGRE.
tingency-constrained optimal power flow,” Elect. Power Syst. Res., vol.
78, no. 5, pp. 873–882, May 2008.
[17] R. D. Zimmerman, C. E. Murillo-Sanchez, and D. Gan, MATPOWER
Version 3.2. Natick, MA: MathWorks, 2007.
Sang-Seung Lee (M’99) was born in Goseong,
Jae-Haeng Heo was born in Korea in 1978. He re- Gyeongnam, Korea, on April 2, 1960. He received
ceived the B.S. degree in electrical engineering from the Ph.D. degree in electrical engineering from the
Dankook University, Seoul, Korea, in 2005, and the Seoul National University, Seoul, Korea.
M.S. degree in electrical engineering from Seoul Na- Currently, he is with the Power System Research
tional University, Seoul, where he is currently pur- Department, Korea Electrical Engineering and
suing the Ph.D. degree in electrical engineering. Science Research Institute, Seoul. His research
His research interests include power system interests are North–East Asia power system inter-
reliability. connection, distributed transmission/distribution
load-flow algorithms, nonlinear/adaptive control
theory, power system stabilizers, and reliability-cen-
tered maintenance.

Mun-Kyeom Kim was born in Korea in 1976. He Dong-Hyeon Kim was born in Pusan, Korea, in
received the B.S. degree in electrical engineering 1981. He received the B.S. degree in electrical
from Korea University, Seoul, Korea, in 2004 and engineering from Pusan National University, Pusan,
the M.S. and Ph.D. degrees in electrical engineering Korea, in 2006 and the M.S. degree from Seoul
from Seoul National University, Seoul, Korea, in National University, Seoul, Korea, in 2008.
2006 and 2010, respectively. Currently, he is with Burns & McDonnell En-
He was a Postdoctoral Researcher with The gineering Company, Inc., Kansas City, MO. His
Institute of Information Technology, Department of interest areas are transmission planning as well as
Electrical Engineering, Seoul National University. power system stability and reliability.
Currently, he is an Assistant Professor in the Depart-
ment of Electrical Engineering, Dong-A University,
Busan, Korea. His research interests include electric power network economics,
power system reliability, and the real-time market design in smart grids.

You might also like