Early Retirement

[ebr]

Hi!

I want to investigate if it is economical to retire the district heating system (DHS) with large FOM costs in my model. I have tried to add the RCAP_BND variable to the DHS which is defined with a NCAP_PASTI. However, it is now choosing to retire the technology, even though this leads to a higher cost in the objective function. Does anyone has experience with this issue?

In BE, it also retires the DHS in every year, so that if RCAP_BND is changed to ~UP, the capacity changes throughout the modeling horizon.. 

---

NCAP_PASTI~2010 | EFF | INVCOST | LIFE |  FIXOM | RCAP_BND~LO
           15                   |0.841 |       0        |  100  |   833     |    15



Einar

[Antti-L]

Sure, if you put RCAP_BND(LO), it means that you define a Lower bound for the retirements. You are thus forcing retirement of the process, regardless of the costs.  It should not be surprising that forcing decision variables to have certain levels is most likely to cause increase in the overall system costs.

I don't fully understand what you mean by "In BE, it also retires the DHS in every year, so that if RCAP_BND is changed to ~UP, the capacity changes throughout the modeling horizon.. " Maybe you can elaborate further on it?

However, if you still see some problem with the early retirement functionality, I would be glad to help if you can explain in more detail what the problem seems to be, and preferably also provide a reproducible example of the problem.

Concerning RCAP_BND, note that is has been designed and documented to work as follows (it is not my design):

Limit on new retirement of capacity for every vintage of process p in region r in year y, for lim = LO/FX/UP or N.
The capacity retirement variable is vintaged but the interpolated / extrapolated bound is not, so the same bound is applied to every vintage of process p available in period t.

[ebr]

Sure, if you put RCAP_BND(LO), it means that you define a Lower bound for the retirements. You are thus forcing retirement of the process, regardless of the costs.  It should not be surprising that forcing decision variables to have certain levels is most likely to cause increase in the overall system costs.

I don't fully understand what you mean by "In BE, it also retires the DHS in every year, so that if RCAP_BND is changed to ~UP, the capacity changes throughout the modeling horizon.. "  Maybe you can elaborate further on it?

However, if you still see some problem with the early retirement functionality, I would be glad to help if you can explain in more detail what the problem seems to be, and preferably also provide a reproducible example of the problem.

Concerning RCAP_BND, note that is has been designed and documented to work as follows (it is not my design):

Limit on new retirement of capacity for every vintage of process p in region r in year y, for lim = LO/FX/UP or N.
The capacity retirement variable is vintaged but the interpolated / extrapolated bound is not, so the same bound is applied to every vintage of process p available in period t.

OK, I thought it allowed for retirement of 15 MW, and that it would optimize based on that. Then I belive everything is working as it should. 
My problem is that I need to force the capacity to be either 0 or 15 so that the operating costs are correct. I can't keep let's say 5 MW of the capacity and thus reduce FOM- since this is a transmission grid connected to all buildings in a small city. Is there a way to achieve this?


Post from BE with RCAP_BND~UP 
                                             2020 2025       2030 2035 2040    2045    2050
0| VAR_Cap | DH_SYSTEM  | 15     | 15       | 15     | 15  |  15    |   15   |  15 
¤| VAR_Cap| - DH_SYSTEM |-4.72 |  -5.35 | -5.98 | -6.3 |  -6.72|  -7.03|  -7.28
actual VAR_Cap |10.28| 9.65 | 9.02 | 8.64 | 8.27 |7.96| | 7.71



It seems like the PASTI is "restored" every year, even though it was retired last period. (NCAP_BND=0, so there is no new investments)

Einar

[Antti-L]

VAR_CAP reports the available capacity of a process in each period, split into past, new and retired capacity.  In your case there is only past and retired capacity (no new capacity is reported to be installed).  The retired capacity is reported as a negative term.

Fore example, in 2040 the remaining past capacity (ignoring retirements) is still 15, and the retried capacity is 6.72.  Thus, the net available remaining capacity is 15+(–6.72) = 8.28.  I cannot see any indication of any PASTI being "restored" every year, the past investment 15 just remains in place, just like it should, according to its lifetime. Could you explain what makes you claim that PASTI is being restored?

You can force the capacity to be zero in year y by setting CAP_BND(y,DH_SYSTEM,UP) = 0.  You cannot model that decision (choosing between zero or 15) endogenously, unless you use the MIP formulation (RETIRE=MIP, with RCAP_BLK).