Veda2.0 Released!


Failure to reproduce a previous run from built model under Veda1.4
#1
Hello
I am running Veda 1.4.5.838

I have an existing version of TIMES/VEDA model. It was previously built using the “start from scratch” tool and run successfully. I will refer to this build and run as the “default”. I am certain that the files have not been changed since the default build and run.
 
However, I have now encountered an issue of reproducibility that I am struggling to understand and solve.
 
On my existing machine, all the model files appear “Consistent”. However, if I sync and run the model, the model build changes and the result differs from the default run. I will refer to this as the “sync” run.
 
I have shared the model (including the built databases) with a colleague. When they open the model in VEDA (with all versions and settings identical to mine), SysSettings is flagged as inconsistent. If the model is synced, it produces the same output as my “sync” run.
 
In addition, while no other files appear as inconsistent, if either I or my colleague choose the “start from scratch”, we get a different build and solution to both of the default and sync runs.
 
I have created a GDX_diff files for "run_default~Dataxxxxx.GDX" and "run_sync~Dataxxxxx.GDX" (xxxx is the date-time stamp for the run), and this indicates a difference in UC_RHSRT for three constraints which point to a difference in inter-/extrapolation.
 
I attach the SysSettings that appears as consistent on my machine, but Inconsistent on my colleague’s machine.

All runs were done using GAMS 31.2, TIMES source code v442 (although the same issue seems to arise with v436). 
 
I am at a loss as to how to proceed to troubleshoot the issue to allow me to get the same output as my "default" run. I would appreciate any input as to how to go about this.


Attached Files
.xlsx   SysSettings.xlsx (Size: 38.46 KB / Downloads: 1)
Reply
#2
Finding out why you have the differences between the data GDX files is the key thing to look at. So, if the only relevant differences are in the UC_RHSRT parameters for three constraints, I would suggest investigating where you have defined these parameters and why they can possibly be output differently by VEDA under your two installations, or after a re-sync, as it seems.
Reply


Possibly Related Threads…
Thread Author Replies Views Last Post
  Model Import Yan Yan 8 2,442 30-01-2024, 05:45 PM
Last Post: Yan Yan
  problem with (re)SYNC of previous working models/DB after reinstall of VEDA Koen Smekens 9 8,417 29-12-2020, 08:01 PM
Last Post: Antti-L
  Problems to sync a small model with 8736 time-slices Pernille.S 21 37,575 28-05-2020, 12:01 AM
Last Post: AKanudia
  How can I model discharge times for two DAYNITE processes? ach 3 5,269 26-04-2020, 02:52 AM
Last Post: Antti-L
  Infeasibility in Model Runs Ismail Kimuli 0 2,111 24-02-2020, 10:49 AM
Last Post: Ismail Kimuli
  Model infeasible regardless of CO2 limit- not sure why? ach 2 5,007 10-12-2019, 04:39 AM
Last Post: Antti-L
  Understanding region emission cap in global model lppnog 4 8,565 15-02-2019, 07:49 PM
Last Post: lppnog
  Reproducing Model Run Errors O.Broad 14 25,152 10-01-2019, 05:45 PM
Last Post: AKanudia
  Demand projection in multi region model Lukas 6 11,975 15-11-2018, 03:55 PM
Last Post: Lukas
  Appropriate solve parameter for large model Kannan 18 37,675 26-04-2018, 03:02 AM
Last Post: Antti-L

Forum Jump:


Users browsing this thread: 1 Guest(s)