On the multi-mode, multi-skill resource constrained project scheduling problem : computational results

Abstract

This paper is concerned with an extension of the Resource-Constrained Project Scheduling Problem (RCPSP) which belongs to the class of the optimization scheduling problems with multi-level (or multi-mode) activities. We developed a practical tool, useful to represent multi-mode projects, and to find a solution for the problem on hand – select the best mode for each resource in each activity in order to minimize the total cost, considering the resource cost, a penalty for tardiness and a bonus for early completion. We implemented an adaptation of a filtered beam search (FBS) algorithm to this problem, using the C# programming language. A “filtered beam” search is a heuristic Branch and Bound (BaB) procedure that uses breadth first search but only the top “best” nodes are kept. We give some of the most important solution details and we report on further computational results, by testing the application for different problem sizes.