I am using LUMPY/Discreate investments to model the expansion of new large hydropower projects in Ecuador. However when In include ILED for new hydropower projects, this seems to affect the 'lumpyness" of investments and installs sized that should not be allowed according to the NCAP_DISC constraints.
Has anyone encountered this type of problem?
I am attaching the a screenshot of the the NewCAP results for both scnearios with and withouth ILED (Lumpy investments activated for both). I also attach the results for hydropower expansion (different colors are different river basins). Results by 2060 are almost similar, but as I said, the ILED scenario overrides the discrete investment options for early years of the time horizon.
I am afraid that answering your question would require some more information.
It seems that it is the process ELCRNHYDS02, for which the capacity results looks unexpected?
But is it the VAR_NCAP results that look strange to you, or the VAR_CAP results, or both?
In addition, to better see what is involved, could you please also:
State the TIMES version you are using, and the full period definition of the model;
From VEDA-FE, show the relevant parameters for the process in question (at least NCAP_ILED, NCAP_DISC, NCAP_TLIFE) with all dimensions shown (from Browse -> TIMES View)
Form VEDA-BE, please show both the VAR_NCAP and the VAR_CAP results for this process (again all dimensions shown (attribute, scenario, region, process, period, and also vintage for VAR_CAP)
With this additional information I would most probably be able to answer the question.
Thanks for your response and I am pretty sure you will be able to spot the problem.
The process ELCRNHYDS02 is the one with unexpected visible results since it is a very large 3.6GW hydropower project. However other smaller hydro technologies also present the problem. Your questions does make sense to me. The VAR_NCAP and the VAR_CAP do not match when I introduce the ILED attribute. In the Var_NCAP table it shows that it istalls capacity in 2020, but in the VAR_CAP the new capacities really apperar in future periods (Is the Var_NCAP showing me when the investment for some capacity is made?)
Regarding your requests, I answer somethings here and I am emailing you some model spreadsheets too.
- I am using TIMES version 4.5.3, I am sending you the syssettings spread sheet so you can see the period definitions.
- From VEDA-FE and for the ELCRNHYDS02 process I am sending you all the relevant parameters for it.
- From VEDA-BW I am sending you the VAR_NCAP and VAR_CAP talbes with and without ILED.
09-06-2017, 12:07 AM (This post was last modified: 09-06-2017, 02:16 AM by Antti-L.)
Ok, the additional information appears to clarify things out.
The process P=ELCRNHYDS02 has NCAP_START=2020, and so the model can rightfully invest into it in the period 2030, as it does in your results. The BE results show that VAR_NCAP(2030)=3.6, which is fully consistent with the unit size specified by NCAP_DISC(1)=3.6.
The issue appears to be related to the amount of available capacity in the model periods 2030/2035. The BE results show that the amount of new capacity is zero in 2030, which is quite correct, because NCAP_ILED=8, and therefore the new capacity, invested in the period 2030, becomes available in the year B(2030)+NCAP_ILED(2030,P).
However, it is not fully clear which period definitions you have actually used in the model run. However, assuming that B(2030)=2028, the capacity would become available in 2036. Then, assuming also that B(2035)=2033 and E(2035)=2037, the length of the period 2035 would be five years, D(2035)=5. The question is then, how much of the capacity commissioned in 2036 should be assumed available in the period 2035?
You seem to suggest that there should be no capacity available in the period 2035, because you noted earlier that "This CAP shouldn't be allowed", referring to the 2035 capacity. However, that would not be economically consistent, because the new unit was fully paid during the eight-year construction period 2028–2035, and the capacity should thus be considered available in 2036–2037, i.e. during the two last years of period 2035. In the standard formulation, TIMES credits for that by setting the fraction of available capacity according to the fraction of the years it is available, in this case 2/5=0.4. But when using the OBLONG switch, the capacity credit is set more accurately, taking into account the discount rate. Therefore, TIMES sets the fraction of the available capacity in this case to 0.371, i.e. VAR_CAP(2035)=0.371*VAR_NCAP(2030) = 1.3356. This is exactly the value of VAR_CAP shown in the VEDA-BE results. Therefore, the assumptions of the period definitions I made above were in fact very likely correct. If not, I would be interested to see the model more closely! But note that I was also able to reproduce exactly the same capacity value using a test model with those period definitions.
Anyway, if you don't like to see such fractions of the installed capacity available in the period the capacity is coming on-line, I suggest that you use negative NCAP_ILEDs instead of positive values. That will result in the construction period being [B(T)+NCAP_ILED,B(T)–1], instead of [B(T),B(T)+NCAP_ILED–1]. I prefer using negative ILEDs myself in my models.
This is great. I appreciate you taking the time to look at this and for the thorough explanation. You estimation about period definitions was correct. I am using negative ILED times and that fixed everything.
