TimesliceTroubleshoot

[[email protected]]

Hi, 

I split the daytime into 24 hours (H0, H1, H2...). However, the QA_CHECK.LOG implicates that:

*** User-provided G_YRFR values are not valid year fractions
*01 WARNING      - TS fractions normalized,  (R.TSL.S)=AL.ANNUAL.ANNUAL
*01 WARNING      - TS fractions normalized,  (R.TSL.S)=AL.SEASON.S
*01 WARNING      - TS fractions normalized,  (R.TSL.S)=AL.SEASON.F
*01 WARNING      - TS fractions normalized,  (R.TSL.S)=AL.SEASON.W
*01 WARNING      - TS fractions normalized,  (R.TSL.S)=AL.SEASON.R


Although I didn't find any error details (only warnings), the VEDA-run failed because of some-error, so I attached the files here, can you help give me some suggestion for debug?

Best,
Xiao

[Antti-L]

The timeslice fractions should sum up to 1, and yours do not.  Therefore, TIMES scales your G_YRFR fractions automatically so that the sum is 1.  As this kind of normalization would usually not cause any severe issues, only warnings are issued. Your model infeasibility is thus most likely caused by some other modeling error. Just look at what Cplex reports about it (should be shown in the log file, which you did not provide). BTW: I think your model is probably too big for solving it with the NEOS server.

[[email protected]]

The timeslice fractions should sum up to 1, and yours do not.  Therefore, TIMES scales your G_YRFR fractions automatically so that the sum is 1.  As this kind of normalization would usually not cause any severe issues, only warnings are issued. Your model infeasibility is thus most likely caused by some other modeling error.  Just look at what Cplex reports about it (should be shown in the log file, which you did not provide).  BTW: I think your model is probably too big for solving it with the NEOS server.

[Antti-L]

I think you should explain the purpose of your post: As far as I can see, the three new attachments in your latest post give no new information about your problem. The model is still proven infeasible, the reason for which is not shown in your attachments, but which I am sure Cplex does report. If VEDA Online does not give you the GAMS log file, you should use the GAMS OPTION SYSOUT=ON; to get the infeasibility report in the listing fle.

I attach a sample GAMS log file, so that you can see that it contains the full Cplex output. You have not yet shown the GAMS log file (only the TIMES QA log), and so the infeasibility report remains undisclosed.

Note:  The GAMS log file is sent to you via email (it is not in the ZIP file available in the dashboard).

[[email protected]]

I think you should explain the purpose of your post: As far as I can see, the three new attachments in your latest post give no new information about your problem. The model is still proven infeasible, the reason for which is not shown in your attachments, but which I am sure Cplex does report. If VEDA Online does not give you the GAMS log file, you should use the GAMS OPTION SYSOUT=ON; to get the infeasibility report in the listing fle.

Hi,

Thank you for the reply.

I'm trying to understand why the project is infeasible. The QA-LOG shows 0 ERROR(S), 14 WARNING(S), but with "Infeasible - No Solution" & "Solve status: Error unknown."

I've attached a GAMS log file (.lst.txt) in my last post, containing detailed execution information, including solver (CPLEX) messages about feasibility, the solution process, and errors.

Due to upload limitations, I can't attach .gdx files. If necessary, can I email them instead?

If the log isn't helpful, could you guide me on how to turn on the GAMS OPTION SYSOUT=ON in Vedaonline?

Best,
Xiao

[Antti-L]

> I've attached a GAMS log file (.lst.txt) in my last post

No, that's the listing file. The GAMS log file was sent to you by email when you ran the model (From: VedaOnline ).  The file is named *_CPLEX_run_log.txt. You should see the Cplex infeasibility diagnostic there...

[[email protected]]

> I've attached a GAMS log file (.lst.txt) in my last post

No, that's the listing file. The GAMS log file was sent to you by email when you ran the model (From: VedaOnline ).  The file is named *_CPLEX_run_log.txt. You should see the Cplex infeasibility diagnostic there...

Hi Antti, I have BA121-7_run_log.txt here.

Best,
Xiao

[Antti-L]

There you go!

Infeasibility row 'EQE_ACTEFF('BC',2035,2035,'EEHP_DistHeating',NRG,IN,SH0)':  0  = 0.139454.
Presolve time = 1.86 sec. (1549.32 ticks)

--- LP status (3): infeasible.
--- Cplex Time: 2.82sec (det. 1843.47 ticks)
--- Model has been proven infeasible

So, I hope it is clear now: You have a modeling error related to the process transformation of EEHP_DistHeating.  This might not be the only error you have, but it is the first one Cplex identifies for you to look at.

[[email protected]]

There you go!

Infeasibility row 'EQE_ACTEFF('BC',2035,2035,'EEHP_DistHeating',NRG,IN,SH0)':  0  = 0.139454.
Presolve time = 1.86 sec. (1549.32 ticks)

--- LP status (3): infeasible.
--- Cplex Time: 2.82sec (det. 1843.47 ticks)
--- Model has been proven infeasible

So, I hope it is clear now: You have a modeling error related to the process transformation of EEHP_DistHeating.  This might not be the only error you have, but it is the first one Cplex identifies for you to look at.

I deleted this process but still infeasible, can you help check it?

[[email protected]]

There you go!

Infeasibility row 'EQE_ACTEFF('BC',2035,2035,'EEHP_DistHeating',NRG,IN,SH0)':  0  = 0.139454.
Presolve time = 1.86 sec. (1549.32 ticks)

--- LP status (3): infeasible.
--- Cplex Time: 2.82sec (det. 1843.47 ticks)
--- Model has been proven infeasible

So, I hope it is clear now: You have a modeling error related to the process transformation of EEHP_DistHeating.  This might not be the only error you have, but it is the first one Cplex identifies for you to look at.

Hi Antti,

I reduced the model size, and can you help to find why the model is still infeasible?
--- Executing after solve: elapsed 0:37:22.089[LST:2322]
--- MODEL.gms(234291) 1465 Mb[FIL:"/var/lib/condor/execute/dir_3191413/gamsexec/MODEL.gms",234291,0]
--- TIMES Solve status: Intermediate infeasible
--- MODEL.gms(234303) 1465 Mb 1 Error[FIL:"/var/lib/condor/execute/dir_3191413/gamsexec/MODEL.gms",234303,0]
*** Error at line 234303: Execution halted: abort$1 '*** ERRORS DURING SOLUTION ***'[LST:2335]
--- MODEL.gms(234303) 1465 Mb 2 Errors[FIL:"/var/lib/condor/execute/dir_3191413/gamsexec/MODEL.gms",234303,0]
--- Putfile END_GAMS /var/lib/condor/execute/dir_3191413/gamsexec/END_GAMS[FIL:"/var/lib/condor/execute/dir_3191413/gamsexec/END_GAMS",1,1]
--- Putfile QLOG /var/lib/condor/execute/dir_3191413/gamsexec/QA_CHECK.LOG[FIL:"/var/lib/condor/execute/dir_3191413/gamsexec/QA_CHECK.LOG",1,1]
--- GDX File /var/lib/condor/execute/dir_3191413/gamsexec/out.gdx[FIL:"/var/lib/condor/execute/dir_3191413/gamsexec/out.gdx",0,0]
--- Profile Summary (2227 records processed)

Best,
Xiao

[Antti-L]

I could not see any indication of the model being infeasible in that run, but just hitting the iteration limit:

--- LP status (10): iteration limit exceeded.
--- Cplex Time: 2192.77sec (det. 1692125.05 ticks)

I think you should use barrier, but for some reason there is no cplex.opt file applied in the run, and so Cplex is using simplex. 
@Amit: Any idea why would that be so?

[[email protected]]

I could not see any indication of the model being infeasible in that run, but just hitting the iteration limit:

--- LP status (10): iteration limit exceeded.
--- Cplex Time: 2192.77sec (det. 1692125.05 ticks)

I think you should use barrier, but for some reason there is no cplex.opt file applied in the run, and so Cplex is using simplex. 
@Amit: Any idea why would that be so?

Hi Antti,

Thank you. No assistance is needed! I will try to reduce size further. 

Best,
Xiao

[rohit]

I could not see any indication of the model being infeasible in that run, but just hitting the iteration limit:

--- LP status (10): iteration limit exceeded.
--- Cplex Time: 2192.77sec (det. 1692125.05 ticks)

I think you should use barrier, but for some reason there is no cplex.opt file applied in the run, and so Cplex is using simplex. 
@Amit: Any idea why would that be so?

Dear Antti,

Currently, the solver option opt file is not being applied in the run on the Neos Server. We plan to include support for it in the next update.

[Ravinder]

I could not see any indication of the model being infeasible in that run, but just hitting the iteration limit:

--- LP status (10): iteration limit exceeded.
--- Cplex Time: 2192.77sec (det. 1692125.05 ticks)

I think you should use barrier, but for some reason there is no cplex.opt file applied in the run, and so Cplex is using simplex. 
@Amit: Any idea why would that be so?

Dear Antti,

Currently, the solver option opt file is not being applied in the run on the Neos Server. We plan to include support for it in the next update.

Dear Antti and Xiao Li,

This issue has been resolved in the latest update.