User constraint with trade process

[mresende]

I am facing some difficulties using a user constraint for limiting the capacity of trade processes. I have defined a unilateral trade process going from the region SUR to region CEN in the file ScenTrade_Transmission.xlsx, with a capacity of 508.5 PJ-year. Looking at the VEDA-BE, I can see that this capacity is being assigned to region SUR. TIMES decides to use 441 PJ in the first year of analysis (2013), which is less than the existing capacity (508.5 PJ-year) and so everything works. This scenario is named "TIMES_WithoutUC" in figure "ResultsUC" attached, which has a table generated in VEDA-BE.

Then, I added a user constraint to limit the new investments in this trade process (figure "userConstraintsGas" attached). As the capacity was placed in SUR, I used this region in the user constraints (this scenario was named "TIMES_WithUC"). Now, the maximum flow of this process along the horizon is 464 PJ, below the capacity of 58.5 PJ-year. One would conclude that TIMES does not need to invest any further in this process. However, TIMES is investing in the first year to create a new capacity of this process in the region CEN (which didn't exist previously) of 464 PJ-year, exactly the capacity that it is used in this new scenario. What I am concluding is that, for some reason, when I put the User Constraints, TIMES passes to consider the capacity installed in the region CEN, instead of the region SUR.

I would like to understand better how TIMES choose the region for the capacities of trade processes and if there is some way to write these user constraints in a way that TIMES does not invest in this trade process, an investment that is not necessary and may change the optimal solution.

Thank you

[AKanudia]

Before getting into the details, I would like to confirm if you have checked that the constraint appears as you would expect in the VEDA-FE Browse. Attach a screenshot of the browse view.

[mresende]

Here it is. It seems correct, except for the fact that capacities in the region CEN appear in the LHS of the equation

[Antti-L]

I would like to understand better how TIMES choose the region for the capacities of trade processes and if there is some way to write these user constraints in a way that TIMES does not invest in this trade process, an investment that is not necessary and may change the optimal solution.

Originally in TIMES, all processes always had capacity variables. However, since the model reduction option (REDUCE) was introduced in TIMES, when that option is used, capacity variables are created only when the user defines some attributes that are directly related to capacity variables.  

In your case, it seems that you have activated model reduction, and therefore capacity variables are created only for processes that have attributes directly related to capacity variables. Consequently, when you do not have any such attributes for the process TU_gas_dry_SUR_CEN_01, the model does not have capacity variables for that process in region CEN, but when you define UC_NCAP for that process in region CEN, the model does have capacity variables for it also in region CEN.

If you dislike seeing the capacities in region CEN, the easy solution would be not to define the superfluous UC_NCAP attributes in region CEN. However, I don't actually see any real harm caused by these capacity variables. I don't think they affect the solution in any way, as you obviously don't have any costs defined for the capacities in CEN.

It would, of course, be possible for TIMES to automatically delete all the superfluous UC_NCAP attributes that you have defined. Do you think that would be desirable?

[mresende]

Indeed, the solution is not affected. I thought that TIMES was using the INVCOST defined for region SUR aldo in region CEN, but after what you have just said, I looked in the BE and now I see that the investment cost associated with TU_gas_dry_SUR_CEN_01 in region CEN is 0.

However, I think that for reporting results, it would be desirable to delete these superfluous UC_NCAP. For, example, I have a table in BE with VAR_NCAP for all trade processes. When I see this table, I have a VAR_NCAP for this process (in the base year) that actually doesn't exist and therefore means nothing.

Thanks

[Antti-L]

Thanks for your response.

Very good, I am willing to implement the removal of all superfluous UC_NCAP, UC_CAP, UC_ACT at the same time in the next version, unless there are any objections from other users.

But then there is also the question whether TIMES should write QA check warnings to the QA_check.log file when these superfluous attributes are removed.

Anyone having an opinion about these matters, please comment here.

However, I think AKanudia also promised to "get into the details", probably in terms of defining the constraints in VEDA without generating any of the superfluous UC_NCAP attributes in the first place.