Load Shifting Storage Processes

[DOlih]

Hello Everyone,

I would like to know if there is any demo about Load Shifting Storage Processes in VEDA-TIMES or if anyone that has any example that could share with me. 

Thanks in advance!

DOlih

[Antti-L]

Well, there is an example included at least in the experimental storage Subres included in the VEDA Advanced Demo. 
See the process RCASIFTELC1 on sheet RCA.

Note: I have not tested that model, and so I don't know how well it works, but the example as such should be valid.

[DOlih]

Well, there is an example included at least in the experimental storage Subres included in the VEDA Advanced Demo. 
See the process RCASIFTELC1 on sheet RCA.

Note: I have not tested that model, and so I don't know how well it works, but the example as such should be valid.

Great! Thank you very much Antti, I will run it. 

Best Regard!

[DOlih]

Well, there is an example included at least in the experimental storage Subres included in the VEDA Advanced Demo. 
See the process RCASIFTELC1 on sheet RCA.

Note: I have not tested that model, and so I don't know how well it works, but the example as such should be valid.

Great! Thank you very much Antti, I will run it. 

Best Regard!

Just to let know that in this example the parameter STG_SIFT(r,t,p,c,s) is spelled wrongly as STG_SFIT, so TIMES does not generate any data related to it.

[Antti-L]

Not seeing it.    I can only see it correctly spelled there in the Subres Excel file: STG_SIFT. 
Do you mean it is incorrectly spelled inside VEDA?

[DOlih]

Not seeing it.    I can only see it correctly spelled there in the Subres Excel file: STG_SIFT. 
Do you mean it is incorrectly spelled inside VEDA?

I dowloaded it again and checked; it was correct. Hahahaha I do not know what happened, maybe it was my PC or even myself.   

My bad about the false alert Antti! Thanks!

[DOlih]

Hello Everyone, again

Talking about the Storage Processes to Load Shifting in TIMES I have the following question:

I understand that with STG_SIFT we can define the proportion in repect to the total demanda that can be shifted, what means that we can impose a limit for the deviation from the original demand loads. I can also define it to a specific season if I use STG_SIFT~ACT. If I wanna define it for a specific timeslice, I can just put it like: STG_SIFT~ACT~ts ? Will this work ?

Another thing is that I could not find a way to define a constraint to limit the amount of energy I can increase or decrease for a specific timeslices, like for timeslice 01 the demand can go up 30% of the original load and go down like 5% at this timeslice. To do that will I need to define like a user Constraint ? or If I put STG_SIFT~UP~ts I could do it ?

Thanks

[Antti-L]

Yes, you can put STG_SIFT~ACT~ts for a specific timeslice (where ts is some seasonal timeslice), and STG_SIFT~COM~ts (here ts will be levelized to the commodity timeslices), to bound the proportion of demand for that can be shifted in timeslice ts. Shifted means here either reducing or increasing the demand in that timeslice, and increasing or reducing it in some other timeslices, respectively.

There is currently no way to bound asymmetrically the proportion of demand that can be shifted, such that the proportion that can be increased would have a different limit than the proportion that can be decreased in a given timeslice.

[DOlih]

Yes, you can put STG_SIFT~ACT~ts for a specific timeslice (where ts is some seasonal timeslice), and STG_SIFT~COM~ts (here ts will be levelized to the commodity timeslices), to bound the proportion of demand for that can be shifted in timeslice ts. Shifted means here either reducing or increasing the demand in that timeslice, and increasing or reducing it in some other timeslices, respectively.

There is currently no way to bound asymmetrically the proportion of demand that can be shifted, such that the proportion that can be increased would have a different limit than the proportion that can be decreased in a given timeslice.

 Thank you for your explanation Antti. It helped me a lot. 

Right now, I have the following question:

The model I am working on has representative days (one day with 24 hours per season) and would check that everything is correctly defined. I understand that when we talk about storage processes, they are more sensible to the timeslice configuration due to the daynight cycle to account charging and discharging. And, which timeslice will represent some days. For example, Wh01 (YRFR) [Winter hour 01] = 0,010273973, corresponds to 3,75 days or 90 hours. 

In my Load Shifting process I have define a ACT_TIME(FX)=3, so I should observed that the shifted loads are met within at most 3 hours of advance or delay. However, in some results I observed that is shifting ocurring more than 3 hours in a sequence:

          VAR_Fin
Wh01
Wh02
Wh03
Wh04 0,049153054
Wh05 0,048829243
Wh06
Wh07 0,056845732
Wh08
Wh09
Wh10
Wh11 0,172213099
Wh12 0,178137085
Wh13 0,143340148
Wh14 0,094983998
Wh15 0,175993925
Wh16 0,171247474
Wh17 0,170855785
Wh18
Wh19
Wh20 0,015364134
Wh22 0,190557271
Wh23 0,153092721
Wh24 0,155762875

¿How is the fact that the hours "connected" in these? Because if a timeslice corresponds to 3,75 days it is difficult to observe the shifting in detail, right ? Or I am confused?

I am very concerned by the possibility of moddeling demand response. Will the maximum shift in hours be respected, if the timeslice is taken as representative of 3.74 days? (even if it is one hour, where the hour repeats itself around 90 hours)

[Antti-L]

The load shifting is of course designed to respect the specified time limit. However, unless you have specified also costs on the shifting (as in the example), the model might easily want to choose chaining the elementary load shiftings in such a way that even if one shift is 3 hours, then a second shift is chained with it, resulting in a combined 6 hours' shift.  In a convex model, it is not possible to prohibit such chained shifts, but if you define costs on the load shifting time, it would become more expensive to do so. In a MIP model it would be possible to strictly prohibit chained shifts.

As you showed only the VAR_Fin results, I don't see how you could verify that the shifted loads are met within some time.  Can you explain in detail, how you concluded that "in some results I observed that is shifting occurring more than 3 hours in a sequence", on the basis of those VAR_Fin values?

[DOlih]

The load shifting is of course designed to respect the specified time limit. However, unless you have specified also costs on the shifting (as in the example), the model might easily want to choose chaining the elementary load shiftings in such a way that even if one shift is 3 hours, then a second shift is chained with it, resulting in a combined 6 hours' shift.  In a convex model, it is not possible to prohibit such chained shifts, but if you define costs on the load shifting time, it would become more expensive to do so.  In a MIP model it would be possible to strictly prohibit chained shifts.

As you showed only the VAR_Fin results, I don't see how you could verify that the shifted loads are met within some time.  Can you explain in detail, how you concluded that "in some results I observed that is shifting occurring more than 3 hours in a sequence", on the basis of those VAR_Fin values?

Thank you very much for your explanation. I was looking for a way to restrict such behavior by User Constraint, but ... 

For these results the ACT_TIME was defined as 3 hours. But we can observe that from Wh11 to Wh17 there is shift ocurring upwards for 7 time-slices in sequence. So, I understood that in this situation the time interval is not respected. 

[Antti-L]

Well, there could be chained shifts (each at most 3 hours), but I still think you must look at both the input and the output flows before jumping into conclusions.  And as mentioned, using a cost on the load shifting time is recommended, to make chained shifts less attractive.

[DOlih]

Well, there could be chained shifts (each at most 3 hours), but I still think you must look at both the input and the output flows before jumping into conclusions.  And as mentioned, using a cost on the load shifting time is recommended, to make chained shifts less attractive.

I see. I only saw the chained shifts where the EQ_CombalM for the electricity presents small values, but even though with in the case where there is costs associated, the model increased as much as possible the demand on the time interval where prices are smaller. What I expected to be done... Do you have any recomendation that I should take in account that if even though with costs associated the model still chaining shifts ? 

I send you the results for the case with STG_SIFT = 0.24 and ACT_TIME = 6, the costs are equal to 5 eur/MWh. 

https://1drv.ms/x/s!AqI_LxQ79HCWlWgUs3ty...r?e=CegVfo

Thank you very much Antti!