Abstract: In order to identify the appropriate priority rules for the multi-project scheduling problem with public and private resource constraints, a great number of computational experiments are conducted based on the benchmark MSPSLIB. Based on the relative deviation index, the performance of 25 classical priority rules are compared under three decision objectives which include minimizing maximum completion time, minimizing total tardiness and minimizing total weighted tardiness. Experimental results show that the performance of a priority rule is significantly correlated with the decision objective, number of tasks in a single project, number of parallel projects and resource utilization coefficient. Although the performance of priority rules is generally consistent with the existing literature, there are significant differences in the performance of MAXTWK and MINSLK. The conclusions obtained have guiding significance for practical multi-project scheduling decisions.

Key words: multi-project scheduling, public resources, private resources, parallel schedule generation scheme, priority rule