Veda2.0 Released!


Preventing early retirements with RCAP_BND
#1
Hello. I'm seeing early retirement of vintaged processes in my TIMES model. I added investment costs, O&M costs hoping that costs associated with capacity deployment will prevent early retirement, but it didn't help.
I have not enabled early retirement. I tried to fix the problem by defining a column RCAP_BND in the FT_T table but upon synchronizing I get the "alien element ignored" error. What's the correct way to use RCAP_BND?

Thanks


Attached Files
.xls   SubRES_NewTechs.xls (Size: 31 KB / Downloads: 14)
Reply
#2
There should be no early retirements in TIMES unless you have activated them, on a process basis. Under VEDA, Using either RCAP_BND or PRC_RCAP activates early retirements for the process for which the bound is defined. Of course, setting a zero bound for all periods would again disable them, but the easiest way to disable them is to not define any RCAP_BND or PRC_RCAP for the process.
Reply
#3
Here's a result I've obtained.

Around 2050, nuclear and natural gas appear for a small time and then they disappear. It seems that is not from newly deployed capacity from the processes in the NewTechs file, but from old processes defined in the primary VT_REG_PRI_ file. I have explicitly defined the stock of nuclear and natural gas there to be non-zero until 2060. However, instead of being used until 2060, they are mostly eliminated until 2050 (possibly because the CO2 constraint specified for 2050 is very restrictive). If it's not early retirement, could you please help me understand why my model is behaving like this?

Also, the production seems to exceed the demand for some years, and I'm not sure why.


Attached Files Thumbnail(s)
   
Reply
#4
Well, the figure does look a bit strange to me, but you cannot expect that I would be able to explain your model results without knowing in detail what your model looks like.

Anyway, you can easily verify whether or not there are any early retirements from the Var_Cap results: If there are any early retirements, they are marked by a special indicator "¤" in the vintage dimension, and the capacity value is negative. The original capacity is reported as a positive value, with the normal '-' or '0' indicator in the vintage dimension ('0' meaning past investments and '-' new capacity investments), and so the total capacity in each period can be obtained as the sum of all Var_Cap values for that period.
Reply
#5
Thanks. It turns out it's not from early retirement. Here's the model. Apologies for not including it earlier.


Attached Files
.zip   model.zip (Size: 1.38 MB / Downloads: 5)
Reply
#6
Ok, thanks for confirming that there were no unexpected early retirements.

If you need assistance with the VEDA templates, I hope some of the VEDA experts can help you with your model.

(If you would like me to look at some problem with your model, you would need to provide me with the TIMES model input files, i.e. the *.RUN and *.DD files, which would reproduce the problem you would like me to look at, and the listing file from the model run, *.LST).
Reply
#7
Ok, for curiosity I tried running the model, and got the following results:


As you can see, I cannot reproduce your results. In my results, there is no overproduction, but only the constant demand of about 700 TWh is satisfied. However, using the existing nuclear plants around 2050 still appears to be part of the solution.

Note that because this MIP model seemed to take a rather long time to solve, I did not wait it till the end.  The solution I obtained was thus not quite proven optimal, but the optimality gap remained at 0.1%. But as the gap was already so small, I would not expect that the proven optimal solution would be much different.


Attached Files Thumbnail(s)
   
Reply
#8
Out of curiosity, I decided to try running the model also with semi-continuous new capacities, instead of your predefined discrete block sizes.

So, for example, new wind(onshore/offshore) can grow only in capacity block-sizes of:
 – [2..4] GW between 2017–2020,  instead of {2,4} GW
 – [1..20] GW from 2021 onwards,  instead of {1,5,10,15,20} GW from 2021 onwards

I got the following results with the semi-continuous capacities (proven optimal solution):


As you can see, the proven optimal semi-continuous solution is almost identical to the one with discrete capacity block sizes.  But the solution time was only 2 seconds, proven to optimality, whereas with the discrete capacity block sizes the solution time of my (interrupted) test run was 3200 seconds, although not yet proven to optimality. It might even have taken several hours to prove the optimality in the first case (I did not want to wait so long...). I think this demonstrates the potential value of the semi-continuous option.


Attached Files Thumbnail(s)
   
Reply
#9
Thanks! Is there a smart way to define semi-continuous capacities, instead of explicitly defining capacities spreadsheet cell-by-cell?
Reply
#10
Also, the nuclear that pops up from old capacity at 2046...I explicitly defined STOCK values for it, hoping it would retire gradually, interpolating between these points. Why is this strange behaviour happening? Is there any other way except explicitly defining input commodity's quantity (as I have done for hydro)?
Reply
#11
I was unable to find the all  DD files, but the RUN files are attached. The directory with DD files is polluted with multiple DD files that seem to match multiple models, and I picked the most obviously relevant one. Please let me know if this isn't adequate.


Attached Files
.zip   veda files.zip (Size: 616.69 KB / Downloads: 2)
Reply
#12
(17-04-2019, 02:04 AM)ach Wrote: Thanks! Is there a smart way to define semi-continuous capacities, instead of explicitly defining capacities spreadsheet cell-by-cell?
I am sorry but I don't understand.  What do you mean by "explicitly defining capacities spreadsheet cell-by-cell"? Semi-continuous new capacities are simply activated by NCAP_SEMI=<lower bound>. The upper bound is defined by NCAP_BND(UP). That is all instead of using NCAP_DISC.
Reply
#13
(17-04-2019, 02:13 AM)ach Wrote: Also, the nuclear that pops up from old capacity at 2046...I explicitly defined STOCK values for it, hoping it would retire gradually, interpolating between these points. Why is this strange behaviour happening? Is there any other way except explicitly defining input commodity's quantity (as I have done for hydro)?
I am sorry but I don't understand.  The nuclear capacity retires exactly as you have specified, interpolating between the data points.  The remaining capacity is 8 GW in 2050 according to your data. So, nuclear power can be produced around 2050 according to the remaining capacity you have specified. I cannot see anything strange in that, can you?
Reply
#14
(17-04-2019, 02:40 AM)Antti-L Wrote:
(17-04-2019, 02:04 AM)ach Wrote: Thanks! Is there a smart way to define semi-continuous capacities, instead of explicitly defining capacities spreadsheet cell-by-cell?
I am sorry but I don't understand.  What do you mean by "explicitly defining capacities spreadsheet cell-by-cell"? Semi-continuous new capacities are simply activated by NCAP_SEMI=. The upper bound is defined by NCAP_BND(UP). That is all instead of using NCAP_DISC.

Thanks! I didn't know about this variable since it isn't mentioned in the documentation (part II, IV) (EDIT: it is mentioned in part II, apologies) and I didn't think it would impact deployment.
Reply
#15
(17-04-2019, 02:47 AM)Antti-L Wrote:
(17-04-2019, 02:13 AM)ach Wrote: Also, the nuclear that pops up from old capacity at 2046...I explicitly defined STOCK values for it, hoping it would retire gradually, interpolating between these points. Why is this strange behaviour happening? Is there any other way except explicitly defining input commodity's quantity (as I have done for hydro)?
I am sorry but I don't understand.  The nuclear capacity retires exactly as you have specified, interpolating between the data points.  The remaining capacity is 8 GW in 2050 according to your data. So, nuclear power can be produced around 2050 according to the remaining capacity you have specified. I cannot see anything strange in that, can you?

Sorry, I should have been more clear. While the capacity does retire, I was hoping it would retire in a more sensible way. That is, producing a gradually declining amount of electricity from nuclear, for instance, instead of switching off nuclear for 20 years and turning it on for 3 years before switching it off again.  It seems a lot of my problems stem from my inexperience with VEDA.
Reply


Forum Jump:


Users browsing this thread: 1 Guest(s)