MINIMIZING THE TOTAL DISCOUNTED COST OF DISMANTLING A NUCLEAR POWER PLANT
Due to economical, technical or political reasons all over the world about 100 nuclear power plants have been disconnected until today. All these power stations are still waiting for their complete dismantling which, considering one reactor, causes cost of up to one Bil. Euros and lasts up to 15 years. In our contribution we present a resource-constrained project scheduling approach minimizing the total discounted cost of dismantling a nuclear power plant. A project of dismantling a nuclear power plant can be subdivided into a number of disassembling activities. The execution of these activities requires time and scarce resources like manpower, special equipment or storage facilities for the contaminated material arising from the dismantling. Moreover, we have to regard several minimum and maximum time lags (temporal constraints) between the start times of the different activities. Finally, each disassembling activity can be processed in two alternative execution modes, which lead to different disbursements and determine the resource requirements of the considered activity. The optimization problem is to determine a start time and an execution mode for each activity, such that the discounted cost of the project is minimum, and neither the temporal constraints are violated nor the activities' resource requirements exceed the availability of any scarce resource at any point in time. In our contribution we introduce an appropriate multi-mode project scheduling model with minimum and maximum time lags as well as renewable and cumulative resources for the described optimization problem. Furthermore, we show that the considered optimization problem is NP-hard in the strong sense. For small problem instances, optimal solutions can be gained from a relaxation based enumeration approach which is incorporated into a branch and bound algorithm. In order to be able to solve large problem instances, we also propose a truncated version of the devised branch and bound algorithm.