An Empirical Analysis of a Set of Hybrid Heuristics for the Solution of the Resource Leveling Problem

Abstract

Consideration is given to the heuristic solution of the resource leveling problem (RLP) in project scheduling with limited resources. The objective is to minimize the changes in the level of resource usage from period to period over the planning horizon of the project while keeping the project duration fixed. First, we present two novel greedy schedule algorithms for the RLP solution. The performance of the proposed algorithms is investigated as low-level hybrids in the context of three famous population-based heuristics, namely differential evolution (DE), genetic algorithm (GA), and particle swarm optimization (PSO). Then, we additionally present two new high-level hybridization schemes (HS), referred to herein as parallel and serial HS, respectively, which combine DE, GA, and PSO in a single hybrid solution algorithm. Detailed experimentation over known complex datasets measures the efficiency of the new hybrids. Statistical analysis employed rank the hybrids according to their solution efficiency. Moreover, comparisons between the developed best hybrid and commercial project management software show a substantial higher performance for the former over real-world construction projects.