CP-SAT - Modelling alternate resources in RCPSP

Dan G

unread,

Jan 15, 2025, 2:19:44 AM1/15/25

to or-tools-discuss

Hello everyone,

Happy new year to all of you and your loved ones. I wish that this year all of you reach optimality as fast as possible.

I have a scheduling model that relies on :

Time constraint such as before_task_start + task_duration <= after_task_start...
Cumulative to ensure that the resource are not exceeded

For some of my tasks, I would like to specify precisely with resource is used. For instance, instead of having a capacity of 3 machines, I would like to know which task is performed on which machine (A, B or C), knowing that the machine can process the task at the different consumption of resource.

My question is about the modelling part. For now, my first approach would be to create a cumulative constraint for each machine and a interval for each possible task on the machine and use the constraint add_exactly_one() to ensure that only one machine is used.

Since I am not used to all the concepts of scheduling and how to solve it in the best way, I am wondering if there is a more efficient way to model it.

Thank you in advance for your answer.

Best Regards,

Dan

Laurent Perron

unread,

Jan 15, 2025, 2:54:06 AM1/15/25

to or-tools-discuss

There is a complete rcpsp_sat.py example.

--
You received this message because you are subscribed to the Google Groups "or-tools-discuss" group.
To unsubscribe from this group and stop receiving emails from it, send an email to or-tools-discu...@googlegroups.com.
To view this discussion visit https://groups.google.com/d/msgid/or-tools-discuss/e6553c6d-991f-4565-97e6-22705470d227n%40googlegroups.com.

Dan G

unread,

Jan 22, 2025, 4:58:31 AM1/22/25

to or-tools-discuss

Thank you.

With the example, I implemented additional Cumulative for the alternative resource and I used an add_exactly_one constraint to choose between the intervals.

Best Regards,

Dan

Laurent Perron

unread,

Jan 22, 2025, 5:08:37 AM1/22/25

to or-tools...@googlegroups.com

Yes, but this is already covered by the multi-mode rcpsp specification.

No need to change the code, just the data.

Laurent Perron | Operations Research | lpe...@google.com | (33) 1 42 68 53 00

To view this discussion visit https://groups.google.com/d/msgid/or-tools-discuss/10874f09-63c6-419e-81f2-df978cab2380n%40googlegroups.com.

Dan G

unread,

Jan 23, 2025, 4:23:50 AM1/23/25

to or-tools-discuss

Thank you Laurent. Your remark opened my mind and I am now reading about the multi-mode RCPSP and see how can I adapt my data for it.