Originally posted by: SystemAdmin

I’m working on large scale unit commitment (UC) models of the electricity market, formulated as mixed-integer linear programs (MILP). The model is generated in GAMS 24.0.2 and solved by the ILOG CPLEX 12.5 solver.

An issue concerning large scale MILP unit commitment problems is the long run time of these models. With respect to this issue, I wonder what the different solver options can contribute in reducing the run time. Currently, all CPLEX solver options are default, except the number of threads which is set to 4. The CPLEX solver manual shows different options for solving MIPs (node selection, branching, cutting planes, preprocessing, etc.).

Can someone give me some advice on how I should adjust these solver options in order to reduce calculation time? Does someone has experience with this?

Regards,

Kenneth
#CPLEXOptimizers
#DecisionOptimization

Originally posted by: SystemAdmin

The first thing to try is probably the CPLEX tuning tool. I don't know whether this is available through GAMS, but I guess you are able to convert your GAMS model and data into a CPLEX *.sav (or *.mps or *.lp) file.

Using this CPLEX file, load the model into the interactive CPLEX shell and call the tuning tool like this:

read model.sav
tune

Maybe, this is able to find some useful non-default settings.

Note that this can take a while, and it might be a good idea to do the tuning on smaller versions of your model.

Another approach would be (if you are allowed to share the data) to contact me at achterberg (at) de (dot) ibm (dot) com and send me your model. But please, if it is too large, do not send it by email. Just contact me and I will provide you with an FTP server address where you can upload it. But again: only do this if you are allowed to share the data.
Tobias
#CPLEXOptimizers
#DecisionOptimization