Resource-Constrained Multi-Project Scheduling Problem (RCMPSP)

Generator and Instances

 

Generated via the methods described in the paper (P21) and analyzed in the paper (P23):

Here you can find the instances generated for the tests described in the paper P21 (section 5).  Please read the following information carefully.

This test bank contains 12,320 problems.  Each problem contains 3 projects, each with 20 activities, and 4 types of resources.  Each project is represented as a design structure matrix (DSM) rather than as a flowchart or precedence table.  The projects also contain feedback marks (cycles), but these can be ignored if desired.

The problems are organized as follows.  You will see 20 file archives (replications), each containing 616 problems.  Each of the 20 archives/replications contains the same basic spreadsheet structure and VBA code, albeit with different problems randomly generated to the same specifications.

Each archive/replication contains 8 Excel workbooks (files), each with 77 problems on separate worksheets and a summary worksheet at the front.  Each workbook also contains the RCMPSP generator code, and the summary worksheet in each file has a "button" embedded on the worksheet which, when pressed (and if macros are enabled), will regenerate the problems to the specifications indicated.  You will NOT need to do this if you simply want to use the existing problem instances.

Each of the 8 Excel files differs in complexity settings (4 levels:  "HHH", "HHL", "HLL", or "LLL") and MAUF variance (2 levels:  0 or 0.25).  The 77 problems in each file differ in NARLF and MAUF levels, with NARLF in seven integer levels over [−3, 3] and MAUF in eleven 0.1 increment levels over [0.6, 1.6].

Of course, all four of these factors can be specified by the user, but the current formatting of the workbooks lends itself to our particular experiment.  We apologize for not making the platform more flexible, but we are posting it in the hopes that others will find the examples and the code useful.  We welcome others to build upon what we've done and/or port the code to other platforms.  We apologize if you find the spreadsheet platform unconventional or inconvenient, but it was expedient for us.

We regret that we do not have time to provide technical support for these materials.  We offer them up for those who have the know-how to make use of them.

Each archive is approximately 8MB in size.

Rep1, Rep2, Rep3, Rep4, Rep5, Rep6, Rep7, Rep8, Rep9, Rep10, Rep11, Rep12, Rep13, Rep14, Rep15, Rep16, Rep17, Rep18, Rep19, Rep20

 

[Back to publications]