2022 IEEE International Conference on Environment and Electrical Engineering and 2022 IEEE Industrial and Commercial Power

Systems Europe (EEEIC / I&CPS Europe) | 978-1-6654-8537-1/22/$31.00 ©2022 IEEE | DOI: 10.1109/EEEIC/ICPSEUROPE54979.2022.9854783

Mixed-Integer Linear Programming based

Maintenance Scheduling of Generating Units
Ali Esmaeel Nezhad Pedro H.J. Nardelli Farideh Ghanavati
Department of Electrical Engineering, Department of Electrical Engineering, Department of Industrial Engineering
School of Energy Systems, LUT School of Energy Systems, LUT and Management, University of Aveiro,
University, 53850, Lappeenranta, University, 53850, Lappeenranta, Aveiro, Portugal
Finland Finland Email: [email protected]
Email: [email protected] Email: [email protected]

Subham Sahoo Gerardo J. Osório

Department of Energy Technology, University Portucalense Infante D.
Aalborg University, 9220 Aalborg, Henrique (UPT), R. Dr. António
Denmark Bernardino de Almeida, 4200-072
Email: [email protected] Porto, Portugal
Email: [email protected]

Abstract—This paper presents a mixed-integer linear The problem of short-term maintenance scheduling problem
programming model for the maintenance scheduling of was studied in [8] using an EP-based algorithm. Ref. [9] used
generating units in the power system. The proposed model is a simulated annealing optimization algorithm to tackle the
investigated for weekly scheduling for one year addressing the maintenance scheduling problem. A mathematical method
crew availability constraint. The maintenance scheduling
combined with the differential evolution method has been
problem is modeled as an optimization problem to determine the
optimal timing for handling the technical constraints of the used in Ref. [10] to solve the single-objective optimization
power generation sector. In addition, the technical constraints generation scheduling problem. The robust optimization
for optimal scheduling of the tasks, like sequential tasks and rest theory has also been suggested in Refs. [11], [12] for the
time of the crews have been addressed in the scheduling generation scheduling problem. The problem of generation
management framework. The weekly peak power and spinning maintenance scheduling of virtual power plants has been
reserve have been considered in line with the economic issues for tackled in [13]. An improved binary particle swarm
power generation in the whole system. The historical market optimization algorithm was utilized in [14] to solve the
clearing price (MCP) and mid-term load forecasting have been generation scheduling problem in power systems. The
considered in the developed model.
proposed model was simulated on the IEEE 24-bus test
Keywords—Maintenance Scheduling, Mixed-Integer Linear system known as the IEEE reliability test system (RTS), and
Programming, Market Clearing Price, Historical Data, Power also, the Kerala power system in India. Ref. [15] introduced
Generating Units. a coordination framework for the generation maintenance
scheduling in electricity markets. The main contributions of
I. INTRODUCTION this paper are as follows:
The increasing demand for electrical energy has been • The maintenance scheduling problem is modeled as a
entering a new stage with new concerns risen regarding standard mixed-integer linear programming.
climate change and environmental emissions. This issue has • The technical constraints for the maintenance crews
already led to starting new investments in the power system have been modeled in this study.
to expand the generation and transmission systems, besides
employing distributed generation systems and demand • A Gantt chart interface for the maintenance scheduling
response programs. However, much more attempts must be and task allocation is provided.
made to promote the operation of the current power system, The rest of this paper is organized as follows. Section II
mitigating the probability of undesired failures, inversely presents the problem formulation. Section III includes the
impacting the power system operation and service quality simulation results, and lastly, some concluding remarks have
[1]–[3]. Accordingly, it is highly required to regularly carry been proposed in Section IV.
out maintenance scheduling to keep the secure operation of
the system. In this respect, a game theory-oriented
maintenance scheduling model has been developed in [4], [5]. II. PROBLEM FORMULATION
The hill climbing technique together with the evolutionary
programming (EP) is employed in [6] to tackle the A. Objective Function
maintenance scheduling problem in the IEEE 30-bus test The objective function of the maintenance scheduling is
system. Ref. [7] presented a fuzzy-EP method to tackle the represented as the maximization of the profits of generation
maintenance scheduling problem using a single-objective companies (GenCos). The profit of the generating units is
optimization model aimed at minimizing the cost. The represented as the historical MCP and the weekly peak loads.
framework was assessed using the IEEE 30-bus test system. There is a logical relationship between the MCP and the load
This work is partly supported by LUT Graduate School, and the Academy of
demand. The objective function includes three main parts, the
Finland via: (a) EnergyNet Research Fellowship n.321265/n.328869 and (b) first term is related to the operating profit of generating units;
FIREMAN consortium n.326270 as part of CHIST-ERA grant CHIST-ERA- the second term addresses the maintenance cost of generating
17-BDSI-003.
978-1-6654-8537-1/22/$31.00 ©2022 IEEE

Authorized licensed use limited to: Polytechnic University of Bucharest. Downloaded on October 19,2022 at 14:10:09 UTC from IEEE Xplore. Restrictions apply.
units and the last term deals with the interruption in the status of the available units while the unit is operating within
maintenance period to guarantee the continuous time interval. its permitted operating bounds if IG is “1”. Constraint (4)
The objective function is as follows: shows the spinning reserve of committed units, while
Max constraint (5) is the limitation of the spinning reserve of
Nω Nu committed units. As constraint (6) indicates, the unit cannot
Z =  [ MCP(ω ) PG (u , ω ) − β (u ) PG (u , ω )] be operated if it is decommitted for maintenance. In other
ω =1 u =1
14444444 4244444444 3 words, the unit would not be absent in the generation
Operational Profit
Nω N u
scheduling if the binary variable IM is equal to “1”. The
−   IM (u, ω )  Fix (u ) + Var (u ) PG max (u )   (1) maintenance time of each generating unit is stated in
ω =1 u =1
1444444442444444443 constraint (7) where MT is applied to the constraint on a
Maintenance Cost weekly basis. Constraint (8) emphasizes that the maintenance
Nω N u Nu
time should be continuous and for any interruption occurring,
−   M (u ) [ SC (u , ω ) + SD (u , ω ) ] − 2 M (u )
ω =1 u =1 u =1
a penalty would be applied to the third part of the objective
1444444442444444443
Interruption Cost function. All binary variables of Eq. (8) are binary variables
The operational profit is modeled as the difference and accordingly, the starting and ending time of each
between revenue of participating in the market and the cost of generating unit maintenance should conform to constraint
power generation during the whole year. The scheduling is (7). Constraints (9) and (10) show the starting and ending
performed for one year by a weekly resolution. Therefore, the weeks of the maintenance scheduling, respectively. The
number of time slots is 52 weeks, i.e. = 52. In this study, limitation of the crews is applied to the problem through
the historical weekly MCP is considered for the revenue constraint (11) in which Lbr(u) is the number of crews for the
calculation. In addition, the linear operating cost function is maintenance of a generating unit. Lbr(u) is determined by the
considered. maintenance scheduling entity as a parameter for each
The second term is related to the maintenance cost of generating unit.
generating units. The maintenance cost of each unit is
expressed by the fix and variable cost terms, while the C. Maintenance Scheduling Specific Constraints
( , ) is the maintenance binary decision variable. In the
case of planned maintenance, this binary variable is 1; Some constraints are particularly significant to the
maintenance team, including the time limitation, available
otherwise, it will be 0.
crews’ limitations, rest time, and also geographical
To guarantee continuous maintenance scheduling, an extra
limitations of a unit. Constraint (12) models the crews’
term has been added to the model as the last term in the
objective function. This term ensures that the outage duration limitation for the case that the simultaneous maintenance of
of each generating unit would be continuous and there is no two units is not possible. In this respect, the units in the set
interruption during the maintenance period to accomplish the {k,l,…,m} cannot enter the maintenance scheduling,
concurrently.
tasks during the predefined time interval. The corresponding
IM (k , ω) + IM (l , ω) + ... + IM (m, ω) ≤ 1 {k , l ,..., m} ∈ u (12)
constraints are as follows:
Furthermore, the maintenance scheduling of similar units in
B. General Constraints a plant should be done in a way that the crews do not have to
Techno-economic constraints of the maintenance move to another plant until finished with such units in one
scheduling problem are as follows: plant. This limitation is addressed in constraint (13).
Nu
Accordingly, the maintenance of unit l would be exactly after
 PG(u,ω) = PL(ω)
u =1
(2) the maintenance of unit k, where the maintenance of the unit
k is indicated on the right-hand side of Eq. (13).
IG(u, ω )PG min (u) ≤ PG(u, ω) ≤ IG(u,ω ) PGmax (u ) (3)
SR(u,ω) = IG(u, ω)PG max (u ) − PG(u, ω) (4)
Start (l ) = Start (k ) + MT (k ) {k , l ,} ∈ u (13)
SR(u , ω ) ≥ SR min (ω ) (5)
In case the crews need rest or an interruption after
IG (u , ω ) ≤ (1 − IM (u , ω ) ) (6) finishing with the maintenance of a unit, the rest would be
Nω used as (14).
 IM (u,ω) = MT (u)
ω =1
(7) Start (m) = Start (l ) + MT (l ) + rest (l ) {l , m} ∈ u (14)
SC (u, ω) − SD(u, ω) = IM (u, ω) − IM (u, ω − 1) ∀ω > 1 (8) For those generating units their maintenance periods are pre-
determined due to the environmental, economic, or fuel
ω × SC (u,ω ) = Start (u) (9)
limitations, the binary variables corresponding to the
ω × SD(u, ω ) = Finish(u ) (10) maintenance of the unit would be assigned to the model as
Nu
parameters. Accordingly, they are not included in the
 Lbr (u) × IM (u,ω ) ≤ Crew(ω )
u =1
(11)
maintenance scheduling. If the time limit is longer than MT,
The constraints of the generating units maintenance it can be modeled by using constraint (15), where Initial and
scheduling are as follows. Constraint 2 indicates the power End specify the permitted time interval for the maintenance
balance equation at each time interval of the scheduling of the unit.
period. The maintenance team must be assured that the 0 if ω < Initial
operable units are adequate to supply the load demand in each 
IM (u, ω ) ≤ 1 if Initial ≤ ω ≤ End (15)
week. The operating constraint of available units is stated in 0 if ω > End
constraint (3). The binary variable IG shows the operation 

Authorized licensed use limited to: Polytechnic University of Bucharest. Downloaded on October 19,2022 at 14:10:09 UTC from IEEE Xplore. Restrictions apply.
 TABLE I. UNIT MAINTENANCE DATA TABLE II. WEEKLY MAINTENANCE DATA

Unit PGmax PGmin MT Fix Var β Lbr Week MCP Demand Week MCP Demand
U01 12 2.4 1 10 5 25.5472 2 W01 36.7 86.2 W27 29.2 75.5
U02 12 2.4 1 10 5 25.6753 2 W02 39.3 90.0 W28 34.1 81.6
U03 12 2.4 1 10 5 25.8027 2 W03 37.2 87.8 W29 33.2 80.1
U04 12 2.4 1 10 5 25.9318 2 W04 35.6 83.4 W30 38.1 88.0
U05 12 2.4 1 0.3 5 26.0611 2 W05 37.9 88.0 W31 25.2 72.2
U06 20 4 1 0.3 5 37.9637 2 W06 35.9 84.1 W32 30.9 77.6
U07 20 4 1 0.3 5 37.7770 2 W07 35.4 83.2 W33 33.1 80.0
U08 20 4 1 0.3 5 37.9637 2 W08 33.5 80.6 W34 26.0 72.9
U09 20 4 3 10 0.9 38.7770 2 W09 27.2 74.0 W35 25.8 72.6
U10 76 15.2 3 10 0.9 13.5073 2 W10 26.7 73.7 W36 25.0 70.5
U11 76 15.2 3 10 0.9 13.3272 2 W11 25.1 71.5 W37 31.1 78.0
U12 76 15.2 3 10 0.9 13.3538 2 W12 25.9 72.7 W38 23.8 69.5
U13 76 15.2 4 8.5 0.8 13.4073 2 W13 24.9 70.4 W39 25.6 72.4
U14 100 25 4 8.5 0.8 18.0000 3 W14 28.8 75.0 W40 25.5 72.4
U15 100 25 4 8.5 0.8 18.6000 3 W15 25.2 72.1 W41 27.7 74.3
U16 100 25 4 8.5 0.8 18.1000 3 W16 33.2 80.0 W42 27.9 74.4
U17 100 25 4 8.5 0.8 18.2800 3 W17 29.1 75.4 W43 33.1 80.0
U18 100 25 4 8.5 0.8 18.2000 3 W18 35.7 83.7 W44 38.1 88.1
U19 100 25 5 7 0.8 17.2800 3 W19 37.0 87.0 W45 38.2 88.5
U20 155 54.25 5 7 0.8 10.7367 4 W20 37.8 88.0 W46 40.3 90.9
U21 155 54.25 5 7 0.8 10.7154 4 W21 36.4 85.6 W47 42.2 94.0
U22 155 54.25 5 7 0.8 10.7367 4 W22 33.7 81.1 W48 38.7 89.0
U23 155 54.25 5 5 0.7 10.7583 4 W23 39.7 90.0 W49 42.8 94.2
U24 197 68.95 6 5 0.7 23.0000 4 W24 38.3 88.7 W50 45.0 97.0
U25 197 68.95 6 5 0.7 23.1000 4 W25 39.0 89.6 W51 58.0 100.0
U26 197 68.95 6 5 0.7 23.2000 4 W26 36.6 86.1 W52 43.4 95.5
U27 197 68.95 6 5 0.7 23.4000 4
U28 197 68.95 6 5 0.7 23.5000 4 - After the maintenance of the two 197-MW units is
U29 197 68.95 6 5 0.7 23.0400 4 finished, i.e. after 8 weeks, an obligatory rest period
U30 350 140 8 4.5 0.3 10.8416 5 is considered for the crews.
U31 400 100 8 5 0.3 7.49210 6
- The 400-MW units have also the same maintenance
crews and the rest is considered two weeks after the
U32 400 100 8 5 0.3 7.50310 6
overhaul of the first unit.
- The number of crews is 18.
III. SIMULATION RESULTS
The proposed MILP maintenance scheduling model has been The proposed problem has been modeled as a single-
simulated on the modified IEEE 24-bus test system. Table I objective optimization problem, aimed at maximizing the
represents the maintenance data of generating units including profit of GenCos from selling energy in the market taking into
the generation capacity, cost of energy generation, and also account the fuel cost of units, besides the variable and fixed
maintenance costs, besides the number of weeks and crews costs of units. The simulation has been done in GAMS by
required for the annual maintenance [16]. using the CPLEX solver installed on an Intel Corei7 Laptop
The weekly maintenance data for one year including the peak with 32 GB RAM. The simulation results have been plotted in
power and MCP based on the average annual cost are given Microsoft Excel by using an interface where the maintenance
in Table II. It is noteworthy that a year is comprised of 52 scheduling would be shown to the user within a Gantt chart
weeks and the energy price of a week is the average energy interface. Fig. 1 depicts the maintenance scheduling of all
price. The data represented in Tables I and II are used for the generating units of the modified IEEE 24-bus system. As
maintenance scheduling in the base case [17]. Furthermore, mentioned in the specific constraints of the problem, the
the assumptions made with respect to constraints (12)-(15) maintenance of 19-MW units are carried out pairwise and
are as follows: continuously in the maintenance scheduling.
- The 197-MW units, i.e. U24-U29 have fixed
maintenance crews. Hence, the maintenance of these
units cannot be carried out concurrently.

Authorized licensed use limited to: Polytechnic University of Bucharest. Downloaded on October 19,2022 at 14:10:09 UTC from IEEE Xplore. Restrictions apply.
 Maintenance Weeks
Plant Start Weeks Planned Weeks P/W
1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52
Unit 01 30 1 30 1 100%
Unit 02 27 1 27 1 100%
Unit 03 24 1 24 1 100%
Unit 04 33 1 33 1 100%
Unit 05 7 1 7 1 100%
Unit 06 3 1 3 1 100%
Unit 07 31 1 31 1 100%
Unit 08 31 1 31 1 100%
Unit 09 45 3 45 3 100%
Unit 10 34 3 34 3 100%
Unit 11 11 3 11 3 100%
Unit 12 34 3 34 3 100%
Unit 13 14 4 14 4 100%
Unit 14 15 4 15 4 100%
Unit 15 6 4 6 4 100%
Unit 16 24 4 24 4 100%
Unit 17 21 4 21 4 100%
Unit 18 35 4 35 4 100%
Unit 19 13 5 13 5 100%
Unit 20 38 5 38 5 100%
Unit 21 9 5 9 5 100%
Unit 22 17 5 17 5 100%
Unit 23 3 5 3 5 100%
Unit 24 8 6 8 6 100%
Unit 25 14 6 14 6 100%
Unit 26 21 6 21 6 100%
Unit 27 27 6 27 6 100%
Unit 28 34 6 34 6 100%
Unit 29 40 6 40 6 100%
Unit 30 8 8 8 8 100%
Unit 31 27 8 27 8 100%
Unit 32 37 8 37 8 100%

Fig. 1. Optimal Maintenace Scheduling of Power Generating Units in IEEE RTS

As can be shown, the rest time of crews after the overhaul of

the two units, i.e. one week, has well been addressed. Moreover, REFERENCES
the two-week rest time of the 400-MW units has also been [1] M. S. Javadi, M. Saniei, and H. Rajabi Mashhadi, “An
satisfied. The largest number of units decommitted for augmented NSGA-II technique with virtual database to solve the
composite generation and transmission expansion planning
maintenance occurs during weeks 9-15 when the load demand problem,” Journal of Experimental & Theoretical Artificial
is relatively low, which is quite economically justified due to the Intelligence, vol. 26, no. 2, pp. 211–234, Apr. 2014, doi:
low energy prices as well. The same thing happens during weeks 10.1080/0952813X.2013.815280.
38-42, during which one of the 400-MW units is decommitted [2] M. S. Javadi and A. Esmaeel Nezhad, “Multi-objective, multi-
year dynamic generation and transmission expansion planning-
for maintenance. This would not cause any difficulty to the renewable energy sources integration for Iran’s National Power
power system reliability considering the reserve requirements. Grid,” International Transactions on Electrical Energy Systems,
In addition, there would not be any maintenance during the vol. 29, no. 4, p. e2810, Apr. 2019, doi:
initial and final weeks of the year due to the high load demand https://doi.org/10.1002/etep.2810.
[3] M. Taherkhani, S. H. Hosseini, M. S. Javadi, and J. P. S. Catalão,
and high energy prices. Fig. 2 illustrates the weekly maintenance “Scenario-based probabilistic multi-stage optimization for
scheduling of generating units and the number of crews for the transmission expansion planning incorporating wind generation
maintenance each week. The total maintenance cost of the integration,” Electric Power Systems Research, vol. 189, Dec.
system for the studied year is M$ 2.723. 2020, doi: 10.1016/J.EPSR.2020.106601.
[4] C. G. Min, M. K. Kim, J. K. Park, and Y. T. Yoon, “Game-
theory-based generation maintenance scheduling in electricity
IV. CONCLUSION markets,” Energy, vol. 55, pp. 310–318, Jun. 2013, doi:
The maintenance scheduling problem of generating units is 10.1016/J.ENERGY.2013.03.060.
[5] D. Jia, H. Cheng, W. Zhang, Z. Hu, J. Yan, and M. Chen, “A new
known as one of the most vital problems of the power system. game theory-based solution methodology for generation
The aim beyond carrying out maintenance scheduling is to maintenance strategy,” European Transactions on Electrical
manage available resources to supply the load demand Power, vol. 19, no. 2, pp. 225–239, Mar. 2009, doi:
addressing the overhaul of generating units and mitigating the 10.1002/ETEP.208.
[6] M. Y. El-Sharkh and A. A. El-Keib, “An evolutionary
units’ trips. Accordingly, this paper presented an efficient programming-based solution methodology for power generation
MILP maintenance scheduling model, aimed at maximizing the and transmission maintenance scheduling,” Electric Power
profit of GenCos from selling energy in the market. In this Systems Research, vol. 65, no. 1, pp. 35–40, Apr. 2003, doi:
respect, the objective function included revenues of the units, 10.1016/S0378-7796(02)00215-8.
[7] M. Y. El-Sharkh, A. A. El-Keib, and H. Chen, “A fuzzy
the maintenance costs of units, and also a term addressing the evolutionary programming-based solution methodology for
continuity of the maintenance period. Furthermore, the techno- security-constrained generation maintenance scheduling,”
economic constraints, besides the maintenance scheduling Electric Power Systems Research, vol. 67, no. 1, pp. 67–72, Oct.
specific constraints relating to the time limitations and crews’ 2003, doi: 10.1016/S0378-7796(03)00076-2.
[8] M. A. Fotouhi Ghazvini, B. Canizes, Z. Vale, and H. Morais,
rest, were all modeled and satisfied. The developed model was “Stochastic short-term maintenance scheduling of GENCOs in
simulated using the modified IEEE 24-bus system. The an oligopolistic electricity market,” Applied Energy, vol. 101,
simulation results indicated that the presented maintenance pp. 667–677, Jan. 2013, doi:
scheduling model could effectively address the requirements of 10.1016/J.APENERGY.2012.07.009.
[9] J. T. Saraiva, M. L. Pereira, V. T. Mendes, and J. C. Sousa, “A
the problem and provided the user with a Gantt chart interface. Simulated Annealing based approach to solve the generator
maintenance scheduling problem,” Electric Power Systems
Research, vol. 81, no. 7, pp. 1283–1291, Jul. 2011, doi:
10.1016/J.EPSR.2011.01.013.
[10] G. Balaji, R. Balamurugan, and L. Lakshminarasimman,
“Mathematical approach assisted differential evolution for
generator maintenance scheduling,” International Journal of

Authorized licensed use limited to: Polytechnic University of Bucharest. Downloaded on October 19,2022 at 14:10:09 UTC from IEEE Xplore. Restrictions apply.
 Electrical Power & Energy Systems, vol. 82, pp. 508–518, Nov. [15] Y. Wang, D. S. Kirschen, H. Zhong, Q. Xia, and C. Kang,
2016, doi: 10.1016/J.IJEPES.2016.04.033. “Coordination of Generation Maintenance Scheduling in
[11] U. E. Ekpenyong, J. Zhang, and X. Xia, “An improved robust Electricity Markets,” IEEE Transactions on Power Systems, vol.
model for generator maintenance scheduling,” Electric Power 31, no. 6, pp. 4565–4574, Nov. 2016, doi:
Systems Research, vol. 92, pp. 29–36, Nov. 2012, doi: 10.1109/TPWRS.2016.2514527.
10.1016/J.EPSR.2012.03.016. [16] M. Estahbanati, “Hybrid probabilistic-harmony search algorithm
[12] M. Shabanzadeh and M. Fattahi, “Generation maintenance methodology in generation scheduling problem,” Journal of
scheduling via robust optimization,” ICEE 2015 - Proceedings Experimental and Theoretical Artificial Intelligence, vol. 26, pp.
of the 23rd Iranian Conference on Electrical Engineering, vol. 283-296, 2014, doi:
10, pp. 1504–1509, Jul. 2015, doi: https://doi.org/10.1080/0952813X.2013.861876.
10.1109/IRANIANCEE.2015.7146458. [17] M. Shahidehpour and M. Marwali, Maintenance scheduling in
[13] O. Sadeghian, A. M. Shotorbani, and B. Mohammadi-Ivatloo, restructured power systems. Springer Science & Business
“Generation maintenance scheduling in virtual power plants,” Media, 2012.
IET Generation, Transmission & Distribution, vol. 13, no. 12,
pp. 2584–2596, Jun. 2019, doi: 10.1049/IET-GTD.2018.6751.
[14] K. Suresh and N. Kumarappan, “Generation maintenance
scheduling using improved binary particle swarm optimisation
considering aging failures,” IET Generation, Transmission &
Distribution, vol. 7, no. 10, pp. 1072–1086, Oct. 2013, doi:
10.1049/IET-GTD.2012.0384.

U01 U02 U03 U04 U05 U06 U07 U08 U09 U10 U11 U12 U13 U14 U15 U16
U17 U18 U19 U20 U21 U22 U23 U24 U25 U26 U27 U28 U29 U30 U31 U32
20

18

16

14
Committed Crews (Person)

12

10

0
W01
W02
W03
W04
W05
W06
W07
W08
W09
W10
W11
W12
W13
W14
W15
W16
W17
W18
W19
W20
W21
W22
W23
W24
W25
W26
W27
W28
W29
W30
W31
W32
W33
W34
W35
W36
W37
W38
W39
W40
W41
W42
W43
W44
W45
W46
W47
W48
W49
W50
W51
W52
Maintenace Horizon (Week)

Fig. 2. Committed Crews for Optimal Maintenace Scheduling of Generating Units in the IEEE RTS.

Authorized licensed use limited to: Polytechnic University of Bucharest. Downloaded on October 19,2022 at 14:10:09 UTC from IEEE Xplore. Restrictions apply.