Automatic Scheduling Method for Maintenance Outages Plan of Transmission System

The main objective of this paper is to develop a model for automatic scheduling of outages tasks for maintenances using Artificial Intelligent tool (Tabu Search). The developed method of TS is used to reduce maintenance outages for electrical power system with high standard of reliability. The developed method solved the maintenance outages with taking in consideration two main problems. The first one is power flow constraints and contingency studies constrains. The second problem is the number of the crews. The developed method for automatic scheduling of maintenance outage tasks using TS is systematically applied to reconfiguration of networks and work starting date scheduling. The works progress is maximized in the scheduling period while minimized the carry-over work to the next scheduling period. The developed method was used in 20-bus test system.


Introduction
Modern industrialized societies depend on the availability of a reliable supply of electricity to sustain their functions and standards of living.These expansions made the electric power systems and their operations the most complex systems of today's civilization, due to the highly nonlinear and computationally difficult environment involved (Furukawa et al., 1991).The planning and operation of these systems should therefore be as efficient, optimal as possible, with high reliability and with low costs.For these reasons, automatic scheduling method draws a lot a attentions.
Maintenance -scheduling optimization has been studied extensively in the open literature.Most of these studies and literature have been directly about or related to generation plants/units maintenance scheduling aiming to cut the fuel costs and obtaining good maintenance schedules.However, maintenance scheduling in transmission and distribution systems is occupying a small portion of the literature.This was the motivation to study this subject.
The objective of this paper is to develop a model for automatic scheduling of outages tasks for transmission and substation maintenance using Artificial Intelligent tool such as Tabu Search and to find the optimal solution for maintenance schedules to minimize the period of doing maintenance activity with preventing any carry-over of work activity to next week.Tabu search is chosen because of its ability to cross the boundaries of feasibility or local optimality, which were usually treated as barriers (Glover, Taillard, & Werra, 1993;Glover, 1989).
There are two main problem to find optimal solution for maintenance schedules.First problem is the power flow and contingency analysis.The second problem is the number of crew or the number of work groups which responsible to execute the work (Reihani, Davodi, Najjar, & Norouzizadeh, 2010;Alshaikh, & El-Amin, 2001).The number of crew (qualified crew) is very important because some time there is conflict between two or more activities in some area not because the continuity of the power (power flow) but because the crews cannot do both or all activities in same time.Step 2: En Tabu move Step 3: Go of Tabu lis Step 4: Re

Results
The 20-Bu   Power system are normally operated so that overload do not occur either in real-time or under any single outage (n-1) or complex outages (n-2).Contingency analysis (CA) is a "what if" scenario simulator that evaluates the impacts on an electric power system when problem occur (Stagg, & El-Abiad, 1968; Internet page, http://smartgrid.epri.com).For example, if line one taken out of the system then does it lead to overload.Furthermore, what impact this line on the system.This scenario called n-1 contingency.The n-1 contingency is initial loss of a singal transmission line.Also, the n-2 contingency is a sequence of events consisting of the initial loss of a single transmission line, following another loss of a single transmission line (and so on for n-3 and n-4 contingencies).In the flowing steps the process of getting contingency analysis list.First, the base case should be free from any over load.Second, one line taken out of the system then power flow study should be done on the system.If there is any over load then this line should be listed in the contingency list.All lines should be taken out separately.Third, the same done on n-2 except that the two lines take out simultaneity.For example, line one out with line two then power flow done after that line one and line three taken outs together then power flow done and so on.n-2 (contingency) ={ (line 1 ,line 2), (line 1, line 3),(line 1, line 4)…,(line 2, line 3,(line 2, 4),(line 2, line 5)…,(line 3, line 4),(line 3, line 5)…} Where (line out, line out) simultaneity.n-3= {(line 1, line 2, line3), (line 1, line 2, line 4), (line1, line 2, line5)… (line 1, line 3, line 4), (line 1, line 3, line4)…} Where {(line out, line out, line out)} simultaneity.
Figure 3 shows the base case of the system before doing any contingency on the system.Two cases of power flow study done (as an example) by using Power World simulator 8.0 edition.Figure 4 shows first case which representing the n-1 case after doing power flow analysis (the line located between bus 15 and bus18).The analysis shows clearly that taken out this line will not affect any branches by overloading.In conclusion, this line Some of cases will be addressed to explain the developed method under flowing conditions: 1.The program or the algorithm will find solution to the problem (over load or number of crew) when it occur.This means that if there is no problem in the initial schedule (requested schedule) the program will not do any change even though this schedule is not optimal solution.In other words, this developed method finds optimal solution to the problem when it occurs.
2. The relationship between the tabu search and algorithm is neighbor solutions so that it goes from one candidate solution to another one then choose the optimal solution from the candidates' solutions.

Maintenance scheduling for case 1:
The proposed maintenance schedule for case 1, given in table 2 is considered as the initial requested schedule by local maintenance centers.It has ten (10) maintenance outage tasks.The schedule proposes to maintain line 1 on Wednesday, line 2 on Friday, line 4 on Tuesday, line 5 on Friday, line 6 on Tuesday, Wednesday and Thursday, line 7 on Monday, line 8 on Saturday and Sunday, line 9 on Saturday, Sunday and Monday and line 10 on Tuesday.The program will generate the base configuration of the system.Also, it will save the given schedule as "the initial requested schedule" and will check for both constraints (overload and number of crews).Table 2 shows that the schedule at Tuesday has overload between tasks 4 and 7, since, taking out line 4 and 7 at same time will generate overload on the system.First step to solve the overload by moving of starting dates.Task no. 4 has higher priority than task no.7 so that task no.7 will move to resolve the overload problem.The program will search for optimal move between neighbor solutions.The solution (optimal solution) is given in table 3. The maintenance schedule of case 2 is given in table 4.There is overload between task number 3, 4 and 2. Tasks number 4 and 2 have a higher priority than task number 3 which means that task number 3 will be moved to remove the overload problem.In same manner as explained in case 1, case number 2 will proceed until the final solution is obtained by the TS program.Table 5 is the final solution for case 2. The initial requested maintenance schedule of case 4 is given in table 8.The schedule has no overload problems and crews conflict.Since the proposed schedule has no overload and has no crew's conflict, no moves of starting dates are required.Therefore, the output is the same as the initial requested schedule since no moves were encountered.The final solution of case 4 is given in table 9. Line 10 10

Conclusion
The developed scheduling methods were applied to 20-Bus test system and from the results, it shows the following: 1.It is confirmed that the developed methods are useful and practical as scheduling tools.
2. The works progress is maximized in the scheduling period while minimized the carry-over work to the next scheduling period.
3. The deviation of date work (starting of the work) which could be requested from local maintenance centers is minimized.
4. The results have been obtained in low execution times.
5. It shows a good performance for functions having small number of variables.
flow will be checked at the demand peak and off-peak time daily.

Figure 3 .
Figure 3. Base case of the test system Fig

Table 1 .
The prohibited lines

Table 2 .
Initial schedule for the 20 bus system-case no. 1

Table 3 .
Final schedule of case no. 1 obtained after TS program

Table 4 .
Initial schedule for the 20 bus system-case no. 2

Table 5 .
Final schedule of case no. 2 obtained after TS programThe proposed schedule, given in table 6, has overload problem in Saturday and Sunday (between task 4 and 8).Based on the priorities of tasks, task number 4 will be shifted to remove the overload.Moving these task, introduces a new overload in the system in Monday (between task 4 and 7).The program will try to solve this by move task number 4 which is less priority until this problem is solved.The staring date of task number 4 is Wednesday.The table 7 is the final solution of case 3.

Table 7 .
Final schedule of case no. 3 obtained after TS program

Table 8 .
Initial schedule for the 20 bus system-case no. 4

Table 9 .
Final schedule of case no. 4 obtained after TS program