A mixed integer-linear programming model for solving the hydroelectric unit maintenance scheduling problem

The unit maintenance scheduling problem is traditionally a challenging topic for electric utilities. In general, the system operators schedule the generating units for maintenance periodically. Research on the maintenance scheduling problem started in 1960's in the last century and has continued until today. Although significant contributions were made as a results of this continuous investigations, but many of those approaches were made in specific areas as different systems have their own special characteristics, such as system reliability considerations. According to the literature cited to-date, there is not one a successful case reported for maintenance scheduling of a pure hydro or hydro-dominated generating system. This thesis outlines an optimization modelling approach that is based on the linear programming techniques to solve the maintenance scheduling problem for hydro systems. This research focused on the potential use of mixed-integer/linear programming algorithms to solve the maintenance scheduling problem for a large scale hydroelectric system. The objective was to determine the best "timing" for each unit outage in the system. This thesis reformulated an existing hydro scheduling model and concentrated on the set of constraints that could effectively satisfy the maintenance scheduling requirements. The case studies carried out revealed that the most appropriate mixed-integer maintenance scheduling model would be formulated in such a way that it develops a maintenance schedule that ensures that the plants are operated in a way to maximize system efficiency and that head convergence have an influence on hydroelectric unit outage scheduling algorithm. Integer programming methods are usually considered suitable for maintenance scheduling problems, but in view of the complexities inherent in hydro models and in mixed-integer programming algorithms, it was found that it is very difficult to generalize one standardized approach to effectively handle this complex and large scale problem. Therefore, this research has focused on two main points. One focused on decreasing the number of possible binary variables in the model. The second focused on finding an appropriate algorithm that could approximate and reasonably simplify the computational process and transforms non-linear constraints into linear ones. It was also found that without specifically or delicately designing the maintenance scheduling constraints, the problem will be hard to solve even by the latest commercially available algorithms.Applied Science, Faculty ofCivil Engineering, Department ofGraduat