Model Infeasible Question

[VanessaDing]

Hello everyone, thanks for your time. 


I'm reaching out to ask a model infeasible question. Currently, I am trying to divide the power system of China in the model into power systems of different regions in a global model.


My idea is to add the new process that defines all the power related processes in different regions of China, eg: previously ewind105, now add ewind115, ewind125, ewind135 .... Since this is a global model, the old technology will be valid for other countries in the definition of tech selection  , and the new  process will be valid for China.


However after modifying and running it, the model always shows infeasibility about technologies that have nothing to do with the power sector and China, which I haven't modified it.

The .lst file has been attached. the current infeasible part is

Number of equations in the conflict: 2.
fixed: EQ_ACTFLO(UK,2006,2025,CH1NGA005,ANNUAL) = 0
fixed: EQE_CAPACT(UK,2006,2025,CH1NGA005,ANNUAL) = 0

Number of variables in the conflict: 1.
lower: VAR_NCAP(UK,2006,CH1NGA005) > 11.377

CPXPARAM Tune TimeLimit
160G6
CPXPARAM SolutionType
2
CPXPARAM Barrier Limits Iteration
999999
Infeasibility row 'EQE_CAPACT(IND.2006.2025.'CC1ELF005'.ANNUAL)':0 0.0432661.
Presolve time 0.20 sec.(124.49 ticks)
Barrier time 0.22 sec.(124.49 ticks)

I haven't changed the CH1NGA005 technique, so I don't know what's wrong with it, although it's in the same excel file as the power system technique I've changed.

Many Thanks.

[Antti-L]

It looks to me that this may be because of using FIXBOH with an older GDX file.
I suggest that you re-run the Baseline scenario, which you use for fixing the first periods.

However, you have also numerous domain violations reported in the listing file, even from Subres_b-newtechs but also from SysSettings, iis_oldtechconstrain, nearterm_calib, and uc_tra, which looks disturbing to me.  So, you seem to have also other serious problems in the current version of your model.  Perhaps starting from scratch (full re-import) might help...

[VanessaDing]

It looks to me that this may be because of using FIXBOH with an older GDX file.
I suggest that you re-run the Baseline scenario, which you use for fixing the first periods.

However, you have also numerous domain violations reported in the listing file, even from Subres_b-newtechs but also from SysSettings, iis_oldtechconstrain, nearterm_calib, and uc_tra, which looks disturbing to me.  So, you seem to have also other serious problems in the current version of your model.  Perhaps starting from scratch (full re-import) might help...

Hi Antti-L,

Thanks very much for your clarification. This was indeed the GDX reference problem. After fixing it and revising the domain violations, the model can normally run now.

Many thanks again!

[VanessaDing]

Hello Everyone, Sorry for the interruption again. I have another infeasible question.

Ran into a snag with my model after adding a nearterm_calib scenario to calibrate the capacity data. Everything was fine before this, but now it just won't solve. Thought it might be the constraints being too tight, so I ditched all the CSP capacity constraints. No dice. Now, it’s tripping up on something with offshore wind tech:

Number of equations in the conflict:  1.
fixed: EQE_UCRTS(UC_Offshore_Calib,JPN,2006,ANNUAL) = 0

Number of variables in the conflict:  1.
lower: VAR_UCRTS(UC_Offshore_Calib,JPN,2006,ANNUAL) > 0.001


Is it possible I’ve made the constraints too strict with this new scenario, or is there something else I’m missing?

Thanks for your time. The files have been attached.

[Antti-L]

Well, it is hard to make any firm conclusion without knowing much anything about this constraint, but one can see from the Cplex diagnostic that it reports the infeasibility in 2006 (likely the base year?) and that the infeasibility might be as small as 0.001. That may indicate that the constraint is not quite working for that model year and should thus perhaps just be removed from that year (by defining the RHS starting from the next model year).  But this is just guessing.