consumption balance results

[meldoone]

I am using CBC as is is included in the base GAMS licence.

I have also tried to use GPLK, but it takes more time for model to be solved.

Is the CBC a normal solver for the TIAM model or would you recommend to purchase a CPlex solver?

[AKanudia]

I have no experience with this solver. Can someone else comment on this?

[meldoone]

I have tried to load ETRef scenario and solving it. I had the solver running for 14 hours again and been told that TIME LIMIT exceeded.

I was using GLPK and CBC solver and VEDA-FE->Case manager->Control panel->GAMS options->ITERLIM=999999.

I will try to re-run the model with CBC solver with this parameter set to 999999999 but I am noy sure that this would help.

What would you recommend to do? Buy the CPlex solver probably?

[Antti-L]

It seems that you are perhaps not using the barrier method.

Check your options file for CBC (COINCBC.OPT), and try the option:
startalg barrier
This should perform better, if you have been using the default dual simplex method until now. If so, let us know how it goes.

[meldoone]

I don't have an *.opt file for CBC in VEDA directory nor in GAMS directory.

[Antti-L]

Ok, then I suggest that you create one in the VEDA work folder.  Without the options file you are running with the default options (i.e. dual simplex).

[meldoone]

Is it right uploads/93/COINCBC.rar ?

How would VEDA and GAMS know that they should use this options file (I choose "OTHER" solver type in VEDA-FE and choose CBC in gamsinst utility)?

Is there any "default option" file to modify?

[Antti-L]

Yes, that's fine for the start. Later, you can read about other options in the solver documentation (check your GAMS\docs subfolders).

The name of the options file is predefined by GAMS. TIMES advices GAMS that  the solver should read the options file if it exists. If it does not exist, you are running with the default solver options. The file should exist in the folder where you run the model, i.e. the VEDA work folder.

[meldoone]

Thanks, I will respond after a series of experimentations with the options file.

[meldoone]

I have a model solved for 5 hours.

Dummy imports are the same as in the ET_Ref_1303 scenario, but the objection function differs:

uploads/93/sys_cost.jpg

[AKanudia]

Congratulations to you and thanks to Antti for arriving at this point.

I am not sure which version of TIMES code was used to run the Ref that is posted on the www. I think it was before this change in the code that changed OBJ values. I can check.

Let us know how the energy/emission results line up.

Antti observed that my statement above was too vague. The change I referred to above applied only to cases using the OBLONG switch for OBJ function. But this switch is not active here, so this is not an issue.

[meldoone]

Primary energy and emissions results differs a little from the ref_1303 results.

uploads/93/ghg_emissions_2.jpg

uploads/93/pr_energy.jpg

[AKanudia]

There is more than the code version at play here.

Post an image of your solve form.

[meldoone]

Do you mean Case manager?

uploads/93/Case_manager.jpg

[AKanudia]

Good news is that I get the same solution as you using the model as it was uploaded. Bad news is that I don't yet know the source of this difference.

I suggest you go ahead with your work; I will investigate this with Antti's help.