Comparative Analysis of Some Metaheuristics for Discrete-Continuous Project Scheduling with Activities of Identical Processing Rates
Discrete-continuous project scheduling problems with positive discounted cash flows and maximization of the net present value are considered. A class of these problems with an arbitrary number of discrete resources and one continuous, renewable resource is taken into account. Activities are nonpreemptable, and the processing rate of each activity is the same continuous, increasing, and concave function of the amount of the continuous resource allotted to the activity at a time. Three common payment models — lump sum payment, payments at activity completion times, and payments in equal time intervals are analyzed. Adaptations of three well-known metaheuristics — simulated annealing, tabu search, and genetic algorithm are described. The paper focuses on a comparative analysis of the metaheuristics. The algorithms are computationally compared on a basis of an extensive experiment. Some conclusions and directions for future research are pointed out.