Stochastic exogenous electricity prices
#1

I wish to model the exogenous electricity prices in the external region stochastic. As I understand, this is currently not a direct model option in TIMES since it is not listed in the current set of stochastic parameters in the documentation.

Is there any way I can get around this by using the existing stochastic parameters as S_COM_FR and S_UC_RHS? Or is it challenging to include the stochastic parameter S_IRE_PRICE to the GAMS source code?All tips are highly appreciated  Smile.
Reply
#2
For defining uncertain variable costs for some process flows or activities, I would say that the easiest way is to use S_FLO_FUNC.

For example, to define uncertain activity costs in a stochastic scenario, you could define:
Code:
FLO_EMIS(r,y,p,'ACT','ACTCOST','ANNUAL') = 1; <br>FLO_COST(r,y,p,'ACTCOST','ANNUAL','CUR') = 1;<br>S_FLO_FUNC(r,y,p,'ACTCOST','ACTCOST',stage,sow) = X;
where ACTCOST is a new commodity of type ENV and LIM type 'N'). Here X represents the uncertain activity costs X per unit of activity, by stage and sow. The cost X can be positive, negative, or zero, and so you can model uncertain variations in either direction.
No changes are needed in the process topology (i.e. there is no need to add ACTCOST into the topology). Compared to using a dedicated attribute (e.g. S_ACT_COST), basically the only difference is the additional FLO_EMIS parameter that is needed here.
Note also that using the S_FLO_FUNC approach doesn't create any new variables or equations into the model. So, it is also in that sense equivalent to using a dedicated new attribute. The additional advantage is that you can use the same approach for defining uncertain variable costs on any process flows, not just activities. And the approach can be used for trade processes as well.

For trade processes, you can define FLO_EMIS(r,y,p,'ACT','ACTCOST','ANNUAL') as mentioned above, but for specific process flows you must use IRE_FLOSUM instead.
Reply
#3

Thank you very much for the informative reply!  

Reply
#4
I am struggling with using S_FLO_FUNC to model stochastic electricity prices in the external region.

I have included a dummy commodity (type ENV) that is duplicated each time electricity is traded by IRE_FLOSUM. This seems to work fine.

As I interpreted your reply, I should use the S_FLO_FUNC to include stochastic variations in the relationship between the electricity traded and the dummy commodity. Correct?

If this is the case I am not able to do it properly, the results are similar with S_FLO_FUNC (ELC, DUMMY) = 1 or with S_FLO_FUNC (ELC, DUMMY) = 0 for all scenario. The result does not seem to be affected by the S_FLO_FUNC parameter.

What am I missing here? Any tips on how I can succeed?

I am using the stochastic SPINES options in ANSWER-TIMES.
Reply
#5
Pernille:

I just tried myself the suggestions I gave earlier. First, I tested by setting:
FLO_EMIS(r,y,p,'ACT','ACTCOST','ANNUAL') = 1; 
FLO_COST(r,y,p,'ACTCOST','ANNUAL','CUR') = 1;
S_FLO_FUNC(r,y,p,'ACTCOST','ACTCOST',stage,sow) = X;
That worked as expected.  Then I tested replacing FLO_EMIS by IRE_FLOSUM, and it also worked as expected.  Did you actually try to set all the three parameters (IRE_FLOSUM, FLO_COST and S_FLO_FUNC) along the lines I suggested?

Of course, the different syntax you suggested (S_FLO_FUNC (ELC, DUMMY)) will not work for IRE processes. That syntax would work only for making FLO_FUNC uncertain, and that can be done only for normal processes.
Reply
#6
It works fine if I set S_FLO_FUNC (DUMMY, DUMMY). I apologize for overlooking this specification in your previous post. I thought it was logic to use S_FLO_FUNC (ELC, DUMMY) but I was clearly wrong for the trade process.

Thank you very much, this made my day!
Reply


Forum Jump:


Users browsing this thread: 1 Guest(s)