I E E E Transactions o n Power Systems, Vol. 13, No.

3, Auguslt 1998


M.K.C. Marwali
S.M. Shahidehpour
Department of Electr ical and Computer Engineering
Illinois Institute of Technology
Chicago, Illinois 606 16
Abstract- Most unit maintenance scheduling packages consider
preventive maintenance schedule of generating units over a on(: or
two year operational planning period in order to minimize the t3tal
operating cost while satisfying system energy requirements and
maintenance constraints. In a global maintenance scheduiing
problem, we propose to consider transmission line maintenance
scheduling and line capacity limits along with generation and line
outages. The inclusion of transmission and network constraint:; in
generating unit maintenance will increase the complexity of the
problem, so we propose
to decompose
the global
generator/transmission scheduling problem into a master problem
and sub-problems using Benders decomposition. In the first stage, a
master problem is solved to determine a solution for maintenance
schedule decision variables. In the second stage, sub-problems are
solved to minimize operating costs while satisfying the network
constraints. Benders cuts based on the solution of the sub-problem
are introduced to the master problem for improving the existing
solution. The iterative procedure continues until an optimal or near
optimal solution is found.
Keywords: Generation maintenance scheduling, transmission
maintenance scheduling, Benders decomposition.

Additional competition and growing complexity in power
generating systems, as well as a need for high service reliability and
low production costs, are provoking additional interests in automatic
scheduling techniques for maintenance of generators, transmission
and related equipment, capable of providing least cost maintenance
In earlier works of thermal generator maintenance scheduling, most
techniques were based on heuristic approaches. These approaches
consider generating unit separately in selecting its optimal outage
interval subject to constraints and an objective criterion for
equalizing or leveling reserves throughout the planning interval [I],
minimizing expected total production costs [6, IO] or leveling the
risk of failure to meet demand [2]. An example of an heuristic
approach would be to schedule one unit at a time beginning with the
largest and ending with the smallest. Most methods, mainly those
based on heuristics, represent only the generation system and do not
take into account the networks constraint effects on generation and
transmission maintenance. Recently [I 31 represented transmission
constraints, but did not recognize interconnection failures and
maintenance, [ 141 recognized the composite generation and
transmission reliability but did not consider transmission
Because of the discrete nature of maintenance schedu,ing,
mathematical programming approaches have fallen into two b-oad
categories: integer programming (branch and bound) [4-51 and
dynamic programming [3]. More recently, Benders decomposition
has been applied to decompose the problem into a master problem
and a series of sub-problems. The coordination of master and :SubPE-185-PWRS-I 6-09-1997 A paper recommended and approved by
the IEEE Power System Engineering Committee of the IEEE Power
Engineering Society for publication in the IEEE Transactions on Power
Systems. Manuscript submitted May 27, 1997; made available for
printing September 30, 1997.

problems results in the solution of generator maintenance scheduling
as reported in [10,14].
This paper extends the Benders decomposition to include
transmission maintenance and network constraints in the
maintenance scheduling problem. The network is modeled as a
probabilistic problem to include the effect of generation and
transmission outages. Section 2 describes the formulation of
maintenance scheduling with network constraints. Section 3
describes the approach of the maintenance problem using Benders
decomposition. The network model used in the proposed method,
the maintenance sub-problem and operation sub-problem are also
discussed in section 3. A detailed example and the result of the
application of the proposed method to IEEE-RTS [1 I] are presented
in sections 4 and 5.

The maintenance scheduling problem is to determine the period for
which generating units of an electric power utility should be taken
off line for planned preventive maintenance over the course of a one
or two year planning horizon in order to minimize the total
operating cost while system energy, reliability requirements and a
number of other constraints are satisfied.
Generating units are distributed in different regions and
interconnected by transmission lines. This may lead to different
composite reliability levels for a given amount of maintenance
capacity outage. Furthermore, generating unit maintenance should
consider transmission forced and planned outages. When
transmission maintenance and other network constraints are
included, the problem becomes considerably more complex and will
be referred to as an integrated maintenance scheduler (IMS) which
represents a network constrained generation and transmission
maintenance scheduling problem. The methodology for the solution
of this problem is discussed in this paper.
In general there are two approaches which can be employed to
achieve one or more aspects of generation maintenance. These
approaches are fictitious cost and maintenance window. Due to
difficulties in obtaining “field proven” maintenance costs, fictitious
costs are often used to penalize deviations from an ideal
maintenance schedule representing the preferred schedule from the
power plant point of view. In maintenance window, the preferred
schedule is represented by time interval (windows) and the objective
is to minimize the real maintenance cost instead of fictitious cost. In
this paper, we use maintenance window approach.
In order to design an IMS for a practically implementable schedule,
numerous and complex constraints which limit the choice of
scheduling times are incorporated into the solution method. The
constraints which are independent of each other at each time period
are categorized as decoupling constraints and those which are
dependent of each other across the time period are categorized as
coupling constraints.
2.A. Coupling Constraints
The first IMS requirement is that units and lines be overhauled
regularly. This is necessary to keep their efficiency at a reasonable
level, keep the incidence of forced outages low, and prolong the life
of units and lines. This procedure is incorporated periodically by
specifying minimax times that a generating unit may run without

0885-8950/98/$10.00 0 1997 I E E E

is solved to generate a trial solution for maintenance schedule decision variables. seasonal limitations 4. If we consider.’ and l. Therefore (1) corresponds to a mixed-integer programming problem. Once xit or Nkt variables are fixed. generating and line capacity limits. In hydro plants. the second is transmission line maintenance cost. The seasonal limitation can be incorporated into e.g. 0 if unit is off-line for maintenance maximum number of lines in right-of-away k time period in which maintenance of generating unit i start earliest period for maintenance of generating unit i to begin latest period for maintenance of generating unit i to begin duration of maintenance for generating unit i time period in which maintenance of line k start earliest period for maintenance of line k to begin latest period for maintenance of line k to begin duration of maintenance for line k vector of dummy generators which corresponds to energy not served at time period t maximum line flow capacity in matrix term active power flow in vector term maximum generation capacity in vector term vector of (git) power generation for each unit at time t power generation of unit I at time t - Nkt) Zmt In the case of resource constraints. .+d. the network can be modeled as either the transportation model or a linearized power flow model. since there is no constraint across time periods. the corresponding zt would be the number of maintenance crews in area A and a would be a percentage of this crew required per-line for maintenance of lines k. The number of crews in each type required at each stage of overhaul of each machine. desirable schedule d S p (vii) (1) expected value transmission maintenance cost per-line in right-away k at time t generation maintenance cost for unit i at time t generation cost of unit i at time t number of lines available in right-of-way k at time t . Decoupling Constraints Network constraints in each time period are considered as decoupling constraints. 2. . for example that lines 1. The master problem.k {cjt(l-xifj+chf (Nk . Mathematically.N k t . resources availability > ~ i+dk . generation and transmission outages should also be taken into account (composite reliability evaluation).. 2 and 3 are to be maintained simultaneously. For this case. On the other hand constraints (i)-(ii) may be seen as a deterministic problem.1} Nk - N& Nkt = N k ~k~ + di < si + d. one for each time period t. It is assumed here that there is little flexibility in manpower usage that varies the time required for maintenance. Xit =o for si < t xit E XIt for ei 5 t 5 ii {0. T maintenance constraints: =1 for t < ei or t 2 I. seasonal limitation.B.. the set of constraints may be formed as Omk(’k ekstsi. Constraints (iii)-(vii) may be seen as a probabilistic problem. The set of operation sub-problems are then solved using the fixed . The objective of ( 1 ) is to minimize total maintenance and production costs over the operational planning period. z ~ would . is specified. This master problem is a relaxation of the original problem in that it contains only a subset of constraints. of the system.’ values of constraint (i). SOLUTION METHODOLOGY We employ the Benders decomposition. the set of constraints be formed as follows: Nli+N2/+N3/ =3 or Nli+N2r+N3t = 0 If we consider that in each maintenance area we have limited resources and crew. This constraint can be constructed using hydrothermal simulation program I1 41. The problem is decomposed into a master problem and operation sub-problems.. (9 < t 5 “i. peak load balance equation. f o r t s e i . l. + d k (ii) system constraints: { E Cr. and the third is the energy production cost. desirable schedule. cp. Constraints (i) represent the maintenance window stated in terms of the start of maintenance variables (s. In the case of crew constraint. We use transportation model to represent system operation limits. These constraints also can be applied to generation maintenance. only a limited number of machines may be serviced at once due to limited manpower.s6 ). or line maintenance. operational constraints are not completely decoupled since the hydrothermal schedule usually is based on chronological simulation of the generation system. 3. This constraint specifies the minimum number of units necessary to produce a certain amount of energy or to avoid spillage in a hydro plant. Furthermore. in the vector form is N unit maintenance status. Problem (1) has a natural structure that enables it to be decomposed into a maintenance master problem which is a deterministic integer problem and a set of operation sub-problems which are probabilistic linear programs. Constraint (vii) represents the reliability requirement which takes into account all states of the system.1064 The time required for overhaul is generally known. Set of constraints (ii) consists of crew and resource availability. The first term of the objective function (1) is the maintenance cost of generators. ~ . lk ‘+dk’). N ~ ) for 1 crew availability 2. Constraints (iii)-(vi) represent peak load balance. on the other hand. The production cost itself is a probabilistic optimization which takes into account the derated capacity of each generating unit. The unit and the line must be available both before their earliest period of maintenance (e. The available crews can be split into geographical and organizational types. ] i E I ~ ~ c j t ~ j t / S. and other operational constraints such as generation and transmission capacity limits in each state..ek) and their latest period of maintenance (e. the INS can be formulated as follows: I Mn t i. one can determine additional constraints in (1) namely the energy production constraint. and other constraints such as environmental and fuel constraints. i t vt E vector of the demand in every bus at time t node-branch incidence matrix probabilistic vector that defines the state of the system acceptable level of expected energy not served The unknown variables x j t and N k f in ( 1 ) are restricted to integer values. Its optimal value is a lower bound on the optimal value of original problem. or t for Yk E ( o . and hence the number of weeks that a machine is “down” is predetermined. 3. giI has continuous value. In order to avoid over-optimistic planning. ~ ~ be the amount of resource m available in area A for each time t and c m k would be a percentage of this resource required per-line for maintenance of line k. which is an integer programming problem. the resulting operation subproblem can be treated as a set of independent sub-problems.

we use simultaneous decomposition simulation for all generator and transmission states and intervals [12]..A Operation Sub-problems The operation cost is the direct cost associated with load supply.1065 maintenance schedule obtained from the solution of the maiiter problem. verify whether that specific configuration of generators and transmission lines is able to supply that specific load without violating system limits.. The calculation of (3) and (5) involve the reliability evaluation of a composite system. desirable schedule 3. Nkt < ‘k Nkt where w: is the expected fuel cost for period t associated with the n* for t < e i or t > l i + d i for si Nk < t 5 4 + d. ~ ( c p ) . then the fuel cost for period t. ‘The IMS must ensure that sufficient reserve exists to provide a secure supply while minimizing the cost of operation. 4 and . etc. If one or more of sub-problems are infeasible. 3. Update the estimated production cost or load shedding expectation. Thus. < f < $ 1.! - Nkt E(O. If all sub-problems are feasible. given the n* iteration of maintenance schedule. is the solution of feasibility check in its associated state space. If one or more of operation sub-problems are infeasible then for each infeasible sub-problem an infeasibility cut is generated. equipment availability. For each infeasible sub-problem resulting from the nthsolution of the master problem. The multiplier z. .I . Return to step 1 if the accuracy is unacceptable. crew availability 2. the cut is associated with a deterministic instead of probabilistic operation sub-problem. The process continues until a feasible solution is found whose cost is sufficiently close to lower bound. T. The feasible cut is of the form Figure 1. - = (2) for ei 5 t 2 1.yand zkymay be interpreted as expected marginal costs associated with 1 MW decrease in the power capacity. The infeasibility cuts (4) will eliminate maintenance values. which measure the change in either production cost or reliability resulting from marginal changes in the maintenance schedule.1) fort s e i . Select a system state cp. The reliability level. 4... then an infeasibility cut is generated. xit or Nkt.B Maintenance Master Problem The maintenance master problem is the minimization of maintenance cost subject to maintenance constraints as well as feasibility and infeasibility cuts from the operation sub-problems. operating conditions. define load levels.. trial solution. At each iteration the solution of sub-problems generates dual multipliers. Le. 1 may be interpreted as the marginal The multiplier decrease in energy not supplied with a 1 MW increase in either generators or transmission lines available capacity. Le. N n (dual variable is A’) where x*and Nnrepresent the trial solution of the master problem at iteration n. resources availability 3. which are not possible to be scheduled. 2. or I > li. +di. Before minimizing the operation cost. The procedure is as follows: 1. 3.. If all of operation sub-problems are feasible then their solution yield a set of dual multipliers from which a feasibility cut is constructed. N k ) f O r e .T S f + g + r = d ( p ) +oN p~ The solution of the sub-problem is not complicated. The infeasibility cut is SOLVE THE MAINTENANCE MASTER PROBLEM I ’FEASIBLE ~ 7 S. given the nthtrial maintenance schedule. we check the feasibility of the sub-problem as Min S. The cost cuts (6) will tend to increase lower bounds obtained from successive maintenance sub-problem solutions. w.xn (dual variable is A) d(9) r 1 If 5 f(p). Sf+ g+ r = d ( p ) g <g(q). depends on the utilization of available units to satisfy load constraints in each time period subject to maintaining reliability above a certain level.. In order to allow a fast algorithm and efficient calculations. the generation cost in period t can be expressed as SET UP INITIAL MAINTENANCE MASTER PROBLEM 4 The sub-problems may be infeasible due to the fact that the reliability constraints cannot be met. since knowing which generator and transmissions are available during period t allows us to minimize the expected operation cost.. as shown in Figure 1. These dual multipliers are used to form one or more constraints (known as cuts) which are added to the master problem for the next iteration. Maintenance Scheduling Decomposition The initial maintenance master problem is formulated as follow: z Min maintenance constraints X-t =1 Xit Xit for si 5 t 5 si + di =o E {0. seasonal limitations 4. Calculate (3) or (5) for the selected state.

the feasibility check gives E{r}=0.=2 N.= 0. Bus data g3 w =546. For some cases the problem may not converge if the A is too small.5 The solution of the maintenance master problem for the second iteration is: N. known as Benders cut. we solve the initial maintenance master problem: Min z 200*(2-N2. in this sample study.. 4.i ) 3. The problem is defined as: We are to perform maintenance on at least 2 lines and one generator within the study period.=O z=500 Given the trial maintenance schedule from iteration 2.Xjt) + c i t ( z k - Nki)] 11 maintenance constraints : for t < ei or t t li + di for si < t < si + di Xjt = 1 Xit = 0 zit E for ei 5 t 5 li {0.) < z N.95*0. Loads are assumed constant during the study period.=2 xI= 1 x. =O h.+N.5* ( l .<2 x. line and load input data in per unit are given in Tables 1. since E{r]+r2+r3}>0.<t<l.85 + 0.25 0. These multipliers are used to form a linear constraint.5 ij 2 . Table 4.95 I. Table 5..~ z + ) 0. is returned to the maintenance problem which is modified and solved again to determine a new trial maintenance plan.) +0..98 power flow: hf.5 0.= 1 x..025 E{r)=O. + d..6 for s.9 I.4 2 2 2 f o r t S e i or t 2 l i + d i - if all subproblems are feasible then the feasibilty cut is: (7) if one or more subproblems are infeasible then the infeasiblity cut is: infeasible subproblem Min S. + r I =I -h.1 0. all lines are assumed to be perfectly reliable.i 0 -2*0. Feasibility Check State Spaces There is a trade-off between A and number of iteration.=l x. < t < s.x?)< 0.)+ 300*(1.NIJ +0. and 3..2. A.5 The solution is: N.. From our experience.. written in terms of maintenance variable x. These bounds can be used as an effective convergence criterion...U.The cost is set arbitrarily to w=1000 because the solution is infeasible. The smaller the A the larger the number of iteration. The infeasible cut is as follows: 0.k cit (1 . For convenience.. = 0. < 2.6 < g..n Maint. 4 x. An are multiplier vectors at the nth iteration The important feature of the Bender decomposition is the availability of upper and lower bounds to the optimal solution at each iteration.<I x. < 2*0.. we check the feasibility of operation sub-problem given the first trial of maintenance schedule. This constraint.285 0.045 0.095 0.9*2.< 1 x.5 -2*0.5 < g . zr . N ~ fore.I 1 crew availability e Line Wline # of lines 1-2 2-3 1-3 (P.5 0. +A2+ g2 + r2 = 3 n 100 200 300 Max Capacity/line (P.+N.5 The primal solution for all state spaces in the feasibility check are given in Table 4.+x2+x.Cost/Unit IFOR z t i.95*0. + g.=O N. Generator Data h i t I Min Canacitv IMax Caoacitv I Operation I Maint. 0.1066 Min Table 2.95 The above LP solution is infeasible..<o 05f.... +h3+g3+ r?=I z .5 o<g.11 Nkt < '-k Nkt N k 0.=O ~ 3 0 0 Then. We assume the study period has only one time interval.4225 which is less than ~ = 0 . The first state soace formulation of feasibility check is as follows N<N . The generator. seasonal iimitations 4 desirable schedule 2 resources availability N~~ E(o.5 (2..<2 N.. The convergence criterion is 2(W-Z) (8) (W+Z) Load balance at bus 1 Load balance at bus 2 Load balance at bus 3 Generator 1 limit Generator 2 limit Generator 3 limit Line 1-2 flow limit Line 1-3 flow limit Line 2-3 flow limit -&-A..u. setting A to 1% is good enough. The system reliability requirement ( E ) is 0..<2 N.5. 5 2 N.5 p.NL2)+ 100*(1-x. 5 .1 0.85 The dual price of the operation sub-problem: generator: h. .98*3* ( 0 .. A"'.2 0. The critical point in the decomposition is the modification of objective function based on the solution of operation sub-problem. 100*(2. EXAMPLE We use a three-bus system as an example..=O N.=2 x.5 0.005 0.=2 N.= 0 I.The primal solution for all state spaces of feasible sub-problem can be seen in Table 5.5 <Ai< 2*0.T.25 . Associated with the solution of the operation sub-problem is a set of dual multipliers which measure changes in system operating costs caused by marginal changes in the trial maintenance. Feasible Sub-txoblem State Soaces I Table 1. < 2.~.)+200*(1-~2)+ S.= 0.<l feasibility and infeasibility cuts from previous iterations Vl E 100 0. = 0..1125 0.. 1 where: n is the current number of iterations.=l x.Cost ($/line) First. The forced outrage rate (FOR) of each generator is given in Table 2.T Yl+r2+r3 xg E 0.5 (0.U.)+ 300*(2-N.Nl.

00/MBtu for Oil #2. Table 10).5*( 1-~.8 18-29 10000 850. For Case 0.21-24 18-22.29 18.28.0 0. Some generation and transmis:. Cases 1 and 2).24.19.25. 20 demand sides.)+ 300*(1. power flow in each line except for lines 1-3 and 11-14 is within limits for most periods.69147 In Case 0 with no network constraints.24-29 I 20.21 27. [r Size 1 Fuel I Heat Ratel 1 MW I 1 I 20 76 100 1 I I I 1 I I I Case I I 1 I Ia I I O Unit l l 1 1 5 1 I l l 1 34 2 5 Table 8. Details of transmission line characteristics can be found in [I 11.27 20.24 2 1. In iteration two.)-9*2. The imposed transmission limits increase the cost of operation.29 .28 .25-29 18-27 20-29 18. Here.22-29 22.xi)+ 200*(1-~2)+ 100*(1-~. Table 9.1067 The dual price of the operation subproblem is: generator: ngi= -9. &=I% of load ) 2 1 36. and the scheduling of inefficient unit for on-line duty. Table 8 shows the change in operating cost over the study period. In these tables. Table 6. O&M Cost of Generating Unit (12-week horizon. the 12-week horizon is between 18-29. Case 0 is converged in two iterations. The corresponding line maintenance schedule and generator loading points of the weekly load are shown in Tables 10 and 11 respectively.5*(1-~2) The solution of the maintenance master problem for the third iteration is: Ni2=2 N2?=0 N. at least 4 transmission lines in this area. Case 1 does not impose any transmission limits.=O ~ 5 4 25 6 We stop here since FW which means the cost is equal to the lower bound.29 23.23 23. transmission limits are imposed on the optimization problem.E$ 22. ~ = l % of load ) 5. the sub-problems are infeasible in all peiiods. $/kWNr.N12)+200*(2-N23)+ 300*(2-N. Generating Units Operating Cost Data O&M Cost I Maintenan<= Btu/kWh Fixed I Variable\ Window I D u r z a I 103$/Yr. A three-month study period of summer weeks.5*2. 23 buses and 38 transmission lines.( (week) Oil #2 14500 I 6.22 18. (Unit 5 in Case 2.25. is considered. Weeks on Maintenance 24. The average fuel price is $3.21 23.25-9.25 27.19. This system is made o f 32 generating units.29 18. indicating a shift from units that use inexpensive fuel to those with more expensive fuels and inefficient units.3/MBtu for Oil #6.24 Maintenance 4-9 6-10 2-4 5-10 4-9 3 -9 Weeks Down 28 23 29 19 18-21. During the tkreemonth period.25-29 18-22. Generating Units Considered Table 7 gives the operating characteristics of the generating uni'ts in this area.68846 36.)+546. weeks up and down in Table I O refer to periods in which units are used for supplying the load.28 I buses 1-2 5-10 Weeks Up 18-23.19. the problem is the classical unit maintenance scheduling.0 Oil#6 Cases 1 and 2 study the effect of transmission limits on maintenance schedule.5 ICg2 = -9 xg3 =0 power flow: nfi2=0 xns= 0 =fii= 0 The feasible cut for the second iteration is: z >100*(2.ion facilities in a particular area need maintenance within the study period. Transmission limits also affect the maintenance schedule (Unit 2 starting on the week of maintenance. The maximum transmission flow over the three-month study period is shown in Table 9..19 20. sub-problems are feasible and the final cost is given in Table 8.9 18-29 0. The coverage of area includes buses 1 through 10.0 I 18-29 I 2 Coal 12000 760. weeks 18-29.. are expected to be maintained. In the first iteration. Table 6 gives the generating units which are to be maintained.26-29 18-26.=2 x!= I n2=l x.22-29 I 18-22.19 26.23-29 20-29 18.29 18.0 I 5.CASE STUDY We apply the proposed method to the IEEE-RTS [I I].19.25-29 18-20. Unit 5 is not used in all time periods. $I.24 28. Table 7.Z/MBtu for coal and $2. Overflow on Transmission Lines (12-week horizon. In Case

Vol. The units and lines maintenance schedule are shifted to lower peak loads.” IEEE Trans. 1972.” IEEE Trans. With transmission limitations. O&M Cost of Generating Unit Case I Cost (IO6$) 12-week I 36. CONCLUSIONS This paper presents a decomposition approach based on the duality theory for generationitransmission maintenance scheduling with network constraints. pp. on Power Apparatus and Systems. 1168-1 174. “Adjusting Maintenance Schedules to Levelize Risk. Dillon. Quintana. 2. Indonesia. “Decomposition Approach to Automated Generation /Transmission Expansion Planning. Zurn. S. N. Garver. With transmission limits. PAS-91.F. Le. Egan. units are loaded in decreasing order of operating costs. 8 J. on Power Systems. The test results demonstrate that the limits on transmission line capacity affect the loading of units and increase the generation by expensive and inefficient units. PAS-91.C. 1970. at Illinois Institute of Technology. 1975. May 1992.L. Jan.1068 In practice. 2057-2063. 538-547. “A Method for Planning Economic Unit Commitment and Maintenance of Thermal Power Systems. Lasdon. T. Singh. May 1992. IO. K. 6. 726-733. Using the proposed decomposition method. ~ = l % of load ) Maintenance 5. pp. 1-2 and 1-5 affects the output of units. Vol. J. May 1966. Z. pp. “Optimal Generator Maintenance Scheduling Using Integer Programming.S. 11. “An Experimental Method of Determination of Optimal Maintenance Schedules in Power systems Using the Branch and Bound Techniques. Sept. 5. Dopazo. 734-743. T. REFERENCES 1. 14. M. maintenance of unit 1 is shifted from weeks 23-24 to weeks 34-35. “IEEE Reliability Test System. 1. Silva. AI-Khamis. SMC-6. Since 1989. No.. No. No.D.5. pp. Palmer. Vol. 8. No. Reliability Test System Task Force of the Application of Probability Methods Subcommittee. Yellen. Limiting lines 1-3. on Power Apparatus and Systems. 1991. 2.L. 7. on Power Apparatus and Systems. Shahidehpour is a professor in the ECE Department and Dean of Graduate College at Illinois Institute of Technology./Feb./Dec. New York. Hara. Aug. MacMillan. BIOGRAPHIES M. Morsztyn. &=I% of load Table 13. Presently. Generating Unit Maintenance in Case 2 (52-week horizon. 1972. Also units 2. Lemonidis. Since units and lines are not forced to be maintained within 12 weeks.” IEEE Trans.H. PAS-94. 2047-2054. ” IEEE Trans. H. W.” IEEE Trans. “Optimal Generating Unit Maintenance Scheduling for Multi-area System with Network Constraints.695-701. pp. Vol. 5 . “A Decomposition Approach to Unit Maintenance Scheduling. resulting in an increase in the overall cost of operation. Indonesia. Optimization Theory for Large Systems.H. pp.K. pp. J. additional complex constraints are imposed on the maintenance scheduling problem. 3. et al. May 1995.PAS-85. The test system was applied over a year-long horizon of 52 weeks Tables 12 and 13 show the maintenance schedule for generating units and transmission lines respectively. Vol. pp. Merrill. Jan.12 1 3. loading of all units is affected and the inefficient unit 5 has to be brought on-line for weeks 23 and 24 to supply generation deficit. 11. ACKNOWLEDGMENTS The authors would like to appreciate the financial support provided by the Power Systems Laboratory at Illinois Institute of ’Technology. PAS-94. 137-144. 1.Vemuri. H.M./Oct. “A Technique for the automated Scheduling of the Maintenance of Generating Facilities. 30743080. ” IEEE Trans.. PAS-104. 1979. Chen. PAS-98.67934 I 4. Toyoda. No.” IEEE Trans. C. pp. G. Vol. pp. Honda.T.V. L. 1985. . J.69147 52-week 36. Sept. and Cybernetics. Marwali received his BS degrees in electrical engineering from Bandung Institute of Technology. 1537-1545./Oct. Nov. “A New Approach to Reliability Evaluation of Interconnected Power Systems Including Planned Outage and Frequency Calculations. No. 1983. 1976.39 11. 7. the cost can be reduced as seen in Table 14 and a better distribution of risk can be achieved. on Power System.M.M. on Power Apparatus and Systems. Vol.. Shahidehpour is a member of the Technical Committee of PICA and serves on the Board of Directors of HKN. 31. he has been with The Ministry o f Research and Technology Republic of Indonesia. pp. Dr. No. V. A. and the Agency for the Assessment & Application of Technology (BPPT). 6./Feb. on Power Systems. Vol. “Solving an Electricity Generating Capacity Expansion Planning Problem by Generalized Benders Decomposition. in 1989 and MS from IIT in 1994. Christiaanse. Transmission Maintenance Schedule in Case 2 (52-week horizon. which in turn affects the loading of unit 2.R. 7. Vol. 665-671.Kimura. Vol 10. 84100. L.” IEEE Trans. K. 36 Table 14. 38.H.F. 1. “Transmission Constrained Maintenance Scheduling of Generating Units: A Stochastic Programming Approach.427-436. Nov.” IEEE Trans. This in turn increases the unit operating cost affecting the loading of units as shown in Table 10. on Man. Deng. on Power Apparatus and Systems. Bloom. 1 6. available units in one time period may become less attractive as compared to those in some other time periods when availability is even more crucial. 4 and 5 are shifted from weeks 27-28 to weeks 38-39. No. 12. Pereira et al. No. No. Maintenance 4-9 3-9 2. No. Table 12. Vol. The extension of the generation maintenance scheduling to include transmission maintenance and network constraints is applicable to the problem of maintenance with probabilistic data.” IEEE Trans. S. 13 L. he is completing his Ph. Vol. weeks 18-19 to 11-12. 6.” IEEE Trans. on Power Apparatus and Systems. pp. L. pp.” Operation Research. on Power Systems. on Power Apparatus and Systems. March/April 1975.” IEEE Trans.2. E. and weeks 27-28 to 38-39 respectively. 9 M. “Generator Maintenance Scheduling via Successive Approximation Dynamic Programming. weeks 20-21 to 12-13. No. Vol.