Article ID: | iaor20021137 |
Country: | Netherlands |
Volume: | 102 |
Issue: | 1 |
Start Page Number: | 263 |
End Page Number: | 286 |
Publication Date: | Feb 2001 |
Journal: | Annals of Operations Research |
Authors: | Lova Antonio, Tormos Pilar |
Keywords: | scheduling |
Frequently, the availablity of resources assigned to a project is limited and not sufficient to execute all the concurrent activities. In this situation, decision making about their schedule is necessary. Many times this schedule supposes an increase in the project completion time. Additionally, companies commonly manage various projects simultaneously, sharing a pool of renewable resources. Given these resource constraints, we often can only apply heuristic methods to solve the scheduling problem. In this work the effect of the schedule generation schemes – serial or parallel – and priority rules – MINLFT, MINSLK, MAXTWK, SASP or FCFS – with two approaches – multi-project and single-project – are analysed. The time criteria considered are the mean project delay and the multiproject duration increase. Through an extensive computational study, results show that with the parallel scheduling generation scheme and the multi-project approach the project manager can obtain a good multiproject schedule with the time criterion selected: minimising mean project delay or minimising multiproject duration increase. New heuristics – based on priority rules with a two-phase approach – that outperform classical ones are proposed to minimise mean project delay with a multi-project approach. Finally, the best heuristics analysed are evaluated together with a representative sample of commercial project management software.