Constraint-based Maintenance Scheduling on an Electric Power-Distribution Network

The exploitation of a power-distribution network involves the scheduling of multiple maintenance and unforeseen repair tasks. The main resource is a network subject to topological, economical and electric constraints. A line section being maintained needs to be isolated from the rest of the network by opening all surrounding switches. This, in turn, would leave other areas of the network de-energized, which is unacceptable in most cases. Hence, these areas have to get their supply via some alternative way, i.e., service needs being restored closing switches connecting to an energized part of the network, taking into account overloading of branches, energy losses, and the cost of the necessary switching operations. In case tasks are carried out in the same area, switching operations might be shared among them. In some cases a valid network reconfiguration might not even exist. Finally, typical scheduling constraints have to be met: resources of limited availability (manpower, vehicles, etc.), due dates, priority relations, etc. We present PLANETS, a prototype scheduler based on CHIP, generating nearoptimal schedules for the tasks to be carried out in one week making sure that, at any moment in time, the network is appropriately reconfigured to guarantee power supply to all consumers. The minimization criterion involves lateness, priority and total cost of the necessary switching operations.