The multi-project scheduling problem
The Resource-Constrained Multi-Project Scheduling Problem (RCMPSP) is an extension of the well known single-project RCPSP. It treats a portfolio of multiple projects that need to be scheduled simultaneously subject to precedence and resource constraints. As most companies have a portfolio of projects, this problem has gained more attention from academics in recent years. Several artificial benchmark datasets for this problem have been proposed. This manuscript proposes additional sets to complement the existing sets and compares them with each other.
A. Artificial datasets
1. MPLIB: (MPLIB.zip): Download our two libraries (MPLIB1 and MPLIB2) with one click!
In order to facilitate the comparison of different datasets, we converted all existing benchmark sets to the same format (.rcmp). The file format is explained in the file "RCMPSP_DataFormat" or "Blueprint RCMP.txt". Furthermore, each set is accompanied by a csv-file containing resource and network information about each of the instances.
New project portfolio data
Multi-project Library MPLIB1 (published in Journal of Scheduling)
The following datasets were designed by the Operations Research and Scheduling group in the first paper:
- 6 projects per instance (60 activities / project)
- 12 projects per instance (60 activities / project)
- 24 projects per instance (60 activities / project)
Reference: Van Eynde, R. and Vanhoucke, M., 2020, Resource-constrained multi-project scheduling: Benchmark datasets and decoupled scheduling. Journal of Scheduling, 23, 301–325.
Multi-project Library MPLIB2 (published in European Journal of Operational Research)
The following datasets were designed by the Operations Research and Scheduling group in a second paper:
- Set 1. Instances with no precedence relations between projects (MP= 0, MF= 1) and all projects use all resource types (CR=1)
- Set 2. Instances with interproject precedence relations and but all projects use all resource types (CR=1)
- Set 3. Instances with no precedence relations between projects, but CR, RD and PD are varied
- Set 4. Instances for which all parameters values vary within a range (all combinations)
Reference: Van Eynde, R. and Vanhoucke, M., “New summary measures and datasets for the multi-project scheduling problem”, European Journal of Operational Research, To Appear
Data from the following papers were converted to the .rcmp file format:
2. MPSPLIB: (MPSPLIB.zip)
Homberger, J. (2007). A multi-agent system for the decentralized resource-constrained multi-project scheduling problem. International Transactions in Operational Research, 14(6):565-589
Homberger, J. (2012). A ()-coordination mechanism for agent-based multi-project scheduling. OR spectrum, 34(1):107-132.
3. RCMPSPLIB: (RCMPSPLIB.zip)
Vázquez, E. P., Calvo, M. P., and Ordóñez, P. M. (2015). Learning process on priority rules to solve the rcmpsp. Journal of Intelligent Manufacturing, 26(1):123-138.
4. BY: (BY.zip)
Browning, T. R., and Yassine, A. A. (2010a). A random generator of resource-constrained multi-project network problems. Journal of scheduling, 13(2):143-161.
B. Solution procedures
Different solution procedures to solve the multi-project resource-constrained project scheduling problem have been compared and published in the following paper:
- Bredael, D. and Vanhoucke, M., 2023, “Multi-project scheduling: a benchmark analysis of metaheuristic algorithms on various optimisation criteria and due dates”, European Journal of Operational Research, To Appear (doi: 10.1016/j.ejor.2022.11.009)
You can now download the solutions of all these procedures: Excel file coming.