Issue with Elastic Demand

[Hesam]

Hi everyone,

I'm running a case with elastic demand, following all the necessary instructions. However, I'm encountering a strange issue where the demand for all energy services drops to zero by a certain point, though it seems reasonable in the periods beyond that. I've attached the settings for the elastic demand-related parameters, and I’m puzzled as to why demand would drop to zero, especially with the COM_VOC parameter in place. Any insights or suggestions would be greatly appreciated!

[Antti-L]

Yes, such drop to zero should not be possible if the elastic demands are working correctly.
But if this this the JRC-EU-TIMES model, I am not overly surprised. I think that model is not able to produce sensible Base prices, due to the use of (and the apparent need of using) COM_TAXNET for the demands. How the demands could in fact be dropping to zero is, however, not clear to me even in that case.

Of course, some more insights could maybe given if you can attach the following files (zipped please):
  • The RUN file
  • The listing file (*.LST)
  • QA_Check.LOG
  • DEMAND_ElastPar.DD

I believe the COM_TAXNET has been used to avoid overproduction of demands.  But if the model is working well, there should be no overproduction. Of course, if there is overproduction, the demand prices would be zero, and not suitable to elastic demands at all. And with the COM_TAXNET, the prices are hardly usable at all (I guess they would be mostly negative).

[Hesam]

Dear Antti,


Thank you for your insights. The model appears to be functioning reasonably well, as the prices follow at least a rational trend. Moreover, even with prices nearing zero, the demand should theoretically increase if I’m interpreting it correctly.

To help resolve this issue, I’ve attached the requested files for a single node run, as I'm currently focusing on small model runs to follow troubleshooting.

Looking forward to your further thoughts.

Regards,

[Antti-L]

> The model appears to be functioning reasonably well, as the prices follow at least a rational trend.

Ok, if you say so, you are the expert of your model. 

> Moreover, even with prices nearing zero, the demand should theoretically increase if I’m interpreting it correctly.

I was referring to the Base prices. If the Base prices do not behave well, elastic demands will not either. Reproducing the Baseline with elastic demands should not affect the demand levels; the solution should basically remain unchanged (very small OBJ change may however appear). But elastic demands are primarily meant to be used in policy scenarios, where the prices of demands would change due to the policies.  Then, and only then should the demands levels change, either by decreasing with higher prices or by increasing with lower prices.

Thanks for showing the files. Unfortunately they did not reveal much more insight to your issue. Nonetheless, if you would still wish me to look at it deeper, please provide all the model files (*.DD, *.RUN, and fi-bau-hdic-v10_DP.GDX). The smaller the test model, the better. 

[Hesam]

Hi Antti,

Thank you for your initial review and insights! I appreciate your time and feedback.

As you suggested, I’ will send you the required files (*.DD, *.RUN, and fi-bau-hdic-v10_DP.GDX) for a deeper review. I hope these will help provide more clarity. 

By the way, my observation about the model behavior was based primarily on the general trend of marginal prices for key energy carriers and I understand that base prices are critical for elastic demand functionality, especially for policy scenarios.

Thanks again for your support, and I look forward to any further insights you might have.

Regards,

[Antti-L]

> I'm encountering a strange issue where the demand for all energy services drops to zero by a certain point

I thought you meant that demands would drop to zero after a certain period. Can you clarify this, because I am not seeing this happening?  So, what exactly is the issue you are seeing? Maybe give some example(s) as well?

[Hesam]

Hi Antti,

The demand for all energy services drops to zero for the periods leading up to 2020 milestone, as shown in the attached image. I hope this clarifies the issue I’m encountering.

Regards,

[Antti-L]

Ok, it was the other way round: Demands are zero up to (and including) 2020?

Well, that appears to be because you are using $SET FIXBOH 2020, without any load point (LPOINT).  So, fixing first periods up to 2020 to a previous solution, but without loading any previous solution. I tested with running the case first as a Baseline (without FIXBOH, without elastic demands), and then a "ReBase" trying to reproduce Base (with FIXBOH, with elastic demands).  It worked otherwise well (no zero demands), but not quite fully reproducing the Baseline. The small difference was in this case is apparently mainly caused by the FIXBOH, though.  Albeit small, it is still an unusually large difference for a FIXBOH effect (there is some relaxation done when fixing the initial periods).

Base:    **** OBJECTIVE VALUE          919802.1603
ReBase:  **** OBJECTIVE VALUE          919795.8452

So, if I understood you correctly, you had zero demands up to 2020 because of using FIXBOH without any previous solution.

[Hesam]

Hi Antti,

Thank you for the insight. I had noticed this feature in the GDX References/Use Solution window but was primarily focused on the elastic demand settings. Now that you’ve highlighted it, I will adjust the settings and hope this resolves the issue.

It would be really helpful if there was a hint or notification within the user interface to indicate when the "fix years up to..." option is activated without selecting a previous solution. That could prevent others from making the same mistake.

Thanks again for your support!

Best regards,

[Antti-L]

> It would be really helpful if there was a hint or notification within the user interface to indicate when the "fix years up to..." option is activated without selecting a previous solution.

Yes, agreed. You could suggest that to the VEDA developers.

I verified the difference being mostly due to the FIXBOH relaxation tolerances, which normally lead to much smaller differences. One might consider reducing them a bit, but there is a trade-off with numerical stability...

[Ravinder]

Hi Antti,

Thank you for the insight. I had noticed this feature in the GDX References/Use Solution window but was primarily focused on the elastic demand settings. Now that you’ve highlighted it, I will adjust the settings and hope this resolves the issue.

It would be really helpful if there was a hint or notification within the user interface to indicate when the "fix years up to..." option is activated without selecting a previous solution. That could prevent others from making the same mistake.

Thanks again for your support!

Best regards,

Dear Hesam,

We appreciate your suggestion and will take it into consideration. For now, here’s a workaround for your requirement:

The tooltip with GDX reference selection details will appear when you hover over the column if the GDX reference is set to true, without needing to open the case.

Please refer to the image below for your reference.