06-03-2024, 09:54 PM (This post was last modified: 06-03-2024, 09:56 PM by Louis.)
Hello,
I am trying to compute levelized costs using veda's outputs, and compare this computed value with VAR_NcapR (the aim is first to understand correctly how Var_NcapR is computed, and second to be able to decompose VAR_NcapR into its different contributions).
To do so, i use the following formula (for a period of 10 years, 2046 to 2055, and the base year is 2005):
LC = [Cost_Inv*10/(1+r)^41+Cost_Fom*Time_NPV+sum(ts,Var_FIn(ts)*EQ_combalM(ts)]/[VAR_FOut*Time_NPV].
If i compare the levelized cost i compute with VAR_NcapR, the values are almost similar, but i am upset by the "almost". Below is a print screen of the vlaues corresponding to my case. I would appreciate some help on what i am doing wrong!!
06-03-2024, 11:44 PM (This post was last modified: 07-03-2024, 02:40 PM by Antti-L.)
I tried to use your numbers and arrived at the same value as you showed TIMES reporting:
I did not understand your investment cost term [Cost_Inv*10/(1+r)^41], which is not correct, and you missed the INV+ part of it, that's it. One can calculate the same value from Cost_Inv as well, but because the INV+ was missing, I used the Cost_NPV values (which cannot be used in general, because they are summed over all vintages, but in your case only the 2050 vintage seemed to be active? Unfortunately, important indexes were hidden in your picture).
07-03-2024, 03:17 PM (This post was last modified: 07-03-2024, 03:19 PM by Antti-L.)
You can see the formula for levelized cost calculation here.
As you can see there, one should include the full investment expenditure (as a lump sum payment) in (the beginning of) year t, discounted to the base year. However, you did not include that amount in your calculation, but instead, you multiplied by 10 the annualized investment costs without the risk premium. And that is not consistent with the LEC formula.
(06-03-2024, 11:44 PM)Antti-L Wrote: I tried to use your numbers and arrived at the same value as you showed TIMES reporting:
I did not understand your investment cost term [Cost_Inv*10/(1+r)^41], which is not correct, and you missed the INV+ part of it, that's it. One can calculate the same value from Cost_Inv as well, but because the INV+ was missing, I used the Cost_NPV values (which cannot be used in general, because they are summed over all vintages, but in your case only the 2050 vintage seemed to be active? Unfortunately, important indexes were hidden in your picture).
Hi Antti-L,
Thanks a lot for your answer, things are much more clear now.
(06-03-2024, 11:44 PM)Antti-L Wrote: I tried to use your numbers and arrived at the same value as you showed TIMES reporting:
I did not understand your investment cost term [Cost_Inv*10/(1+r)^41], which is not correct, and you missed the INV+ part of it, that's it. One can calculate the same value from Cost_Inv as well, but because the INV+ was missing, I used the Cost_NPV values (which cannot be used in general, because they are summed over all vintages, but in your case only the 2050 vintage seemed to be active? Unfortunately, important indexes were hidden in your picture).
Hi Antti-L,
Thanks a lot for your answer, things are much more clear now.
Hi Antti-L,
I have a follow up question regarding the computation of the levelized cost. Thanks to your answer, i am now able to compute the levelized costs and decompose them into their different components for many processes. However, in some cases my computation is wrong: it gives almost the good levelized cost, but not exactly the correct value. I can try to image this considering two different processes of my model: HNGA105, for which my computation works; and EGEOT205, for which it doesn't. I am attaching to this post the excel file with the times outputs and my computations for both processes.
For HNGA105, when i multiply cost_inv by time_npv, i obtain exactlty the reported value of cost_npv for inv (and the same for cost_inv+, and cost_fom). This is not the case for EGEOT205, and i can't understand why. I have to add that in the case of EGEOT205, there are some NCAP_ICOM (CUELC and STEEL). If necessary, i can give you the parameters of the two processes.
Additionnal question, that may be related to the first one: for HNGAS105, cost_inv, cost_inv+ and cost_fom are proportional to var_ncap. This is not the case for EGEOT205 (see the column "2040/2050" in the excel file, where i compute the ratio of var_ncap, cost_inv, cost_onv+ and cost_fom between 2040 and 2050).
I hope that you could help me on this!
Best regards,
Louis
• You seem to have somewhat misunderstood the TIMES levelized cost calculation:
o Levelized cost is calculated over the full lifetime of the capacity installation, as defined in literature.
o In TIMES, levelized cost is thus vintage-specific (not calculated for each period, but for each vintage)
o You seem to be calculating a levelized cost separately for each period.
• The data on your spreadsheet is not fully sufficient for replicating the TIMES levelized cost calculation.
• I am sure I would be able to replicate the TIMES calculation in Excel if full process data and full data for the reporting parameters related to the process were available.
• The reported Cost_NPV values and those calculated by you appear to differ, the source for the difference being unclear, indicating a potential hidden source of inconsistency in your calculation.
• Note that in the TIMES levelized cost calculation only flows explicitly in the topology are taken into account.
In summary, it seems to me that your definition of levelized costs differs from that commonly used in literature and employed in TIMES. However, I think your calculation is nonetheless reasonably consistent as such, and so you could well use your own approach for your own purposes, for deriving a closely resembling cost measure by cost component and by period.
However, if you would still like me to show a replication of the TIMES levelized cost calculation, please provide me with the full process data e.g for EGEOT205 (all data from VEDA Item details) and full report parameter data for the process (all capacity, activity and flow result data, all related commodity marginals, and all cost parameters, Var_NcapR, Time_NPV, and Cap_New) directly copied either from the VD file or from VEDA result tables (with all dimensions expanded).
(14-03-2024, 09:38 PM)Antti-L Wrote: I can make the following comments:
• You seem to have somewhat misunderstood the TIMES levelized cost calculation:
o Levelized cost is calculated over the full lifetime of the capacity installation, as defined in literature.
o In TIMES, levelized cost is thus vintage-specific (not calculated for each period, but for each vintage)
o You seem to be calculating a levelized cost separately for each period.
• The data on your spreadsheet is not fully sufficient for replicating the TIMES levelized cost calculation.
• I am sure I would be able to replicate the TIMES calculation in Excel if full process data and full data for the reporting parameters related to the process were available.
• The reported Cost_NPV values and those calculated by you appear to differ, the source for the difference being unclear, indicating a potential hidden source of inconsistency in your calculation.
• Note that in the TIMES levelized cost calculation only flows explicitly in the topology are taken into account.
In summary, it seems to me that your definition of levelized costs differs from that commonly used in literature and employed in TIMES. However, I think your calculation is nonetheless reasonably consistent as such, and so you could well use your own approach for your own purposes, for deriving a closely resembling cost measure by cost component and by period.
However, if you would still like me to show a replication of the TIMES levelized cost calculation, please provide me with the full process data e.g for EGEOT205 (all data from VEDA Item details) and full report parameter data for the process (all capacity, activity and flow result data, all related commodity marginals, and all cost parameters, Var_NcapR, Time_NPV, and Cap_New) directly copied either from the VD file or from VEDA result tables (with all dimensions expanded).
Hello Antti-L,
Thank you again for your answer. Sorry, my spreadsheet was unclear, but i indeed computed the levelized cost for each vintage: in my excel file, for EGEOT205, the cell D31 is the one to consider for the levelized cost of vintage 2040. Cell F31 is confusing and is not a levelized cost, i should have deleted it (however, it has allowed me to realize something that is not clear for me: why are cost_inv and cost inv+ not proportionnal to VAR_Ncap in the case of EGEOT205, while it is for HNGA105).
I would be very pleased if you could indeed replicate the TIMES levelized cost calculation. I attach to this post the item details for EGEOT205, and the results from the VD file (for the USA). I hope i forgot nothing!
Unfortunately, it turned out that you apparently have used some options that I could not easily figure out (I should have asked for them as well), and so I decided to make a small test model with your model data, such that it reproduces all the capacities, flows, Cost_Inv, Cost_Fom, Cost_Act, and the commodity marginals, and then did the levelized cost calculation from the result data. Indeed, I obtained fully replicated values in Excel.
Please see my LEC calculations replicating the values TIMES reported in the attached.
My relevant RUN options are also listed there. It seemed to me that perhaps you did not use the levelized annual cost option, which is basically necessary for obtaining fully self-consistent cost reporting. Even without it, the differences should be quite small, but my objective was to fully reproduce the values.
(20-03-2024, 12:09 AM)Antti-L Wrote: Thanks for the data.
Unfortunately, it turned out that you apparently have used some options that I could not easily figure out (I should have asked for them as well), and so I decided to make a small test model with your model data, such that it reproduces all the capacities, flows, Cost_Inv, Cost_Fom, Cost_Act, and the commodity marginals, and then did the levelized cost calculation from the result data. Indeed, I obtained fully replicated values in Excel.
Please see my LEC calculations replicating the values TIMES reported in the attached.
My relevant RUN options are also listed there. It seemed to me that perhaps you did not use the levelized annual cost option, which is basically necessary for obtaining fully self-consistent cost reporting. Even without it, the differences should be quite small, but my objective was to fully reproduce the values.
I hope it can be of some help.
Thank a lot Antti-L for your answer. I will read your excel file to have a better understanding of the calculations.
(20-03-2024, 12:09 AM)Antti-L Wrote: Thanks for the data.
Unfortunately, it turned out that you apparently have used some options that I could not easily figure out (I should have asked for them as well), and so I decided to make a small test model with your model data, such that it reproduces all the capacities, flows, Cost_Inv, Cost_Fom, Cost_Act, and the commodity marginals, and then did the levelized cost calculation from the result data. Indeed, I obtained fully replicated values in Excel.
Please see my LEC calculations replicating the values TIMES reported in the attached.
My relevant RUN options are also listed there. It seemed to me that perhaps you did not use the levelized annual cost option, which is basically necessary for obtaining fully self-consistent cost reporting. Even without it, the differences should be quite small, but my objective was to fully reproduce the values.
I hope it can be of some help.
Thank a lot Antti-L for your answer. I will read your excel file to have a better understanding of the calculations.
Hello Antti-L,
I have read the results you have shared, and when i apply your values to my excel file with my own formulas, i can compute correclty the value of Var_NcapR. Which is good! However, i still can't undertand why this doesn't work when i apply these same formulas to the case of EGEOT205 in the USA. I tried to run the model with the same options as yours (see the "RUN OPTIONS" sheet in the attached excel file, but it looks like the ANNCOST LEV is not activated. How to activate it from the VEDA interface?), and the results are quite similar to the previous run ( EGEOT205 sheet, the inputs from the vd file in green, and the computation in orange). If i only focus on the cost_inv values, it looks like my computations give values that are 0.3% lower that what is found in the vd file (except for cost_inv (ACT) which is correctly computed): for example, the sum of the discounted values for cost_inv gives around 6410, while the cost_NPV (INV) reported in the vd file is 6428. I would be happy to provide you with any additionnal detail that is missing in my post to help you clarify this situation. (Note: CUELC and STEEL are implicit commodities for EGEOT205).
Best regards
>for example, the sum of the discounted values for cost_inv gives around 6410, while the cost_NPV (INV) reported in the vd file is 6428.
Yes, that small difference (<0.3%) was seen in your earlier results as well. I am not quite sure where it comes from but I suspected it being due to not using levelized annual costs. If you can provide the full model input files (*.DD, *.RUN), I could of course check. Levelized annual cost can be activated by $SET ANNCOST LEV in the RUN file. See modifying-run-files
22-03-2024, 07:18 PM (This post was last modified: 22-03-2024, 07:21 PM by Louis.)
(22-03-2024, 03:57 PM)Antti-L Wrote: >for example, the sum of the discounted values for cost_inv gives around 6410, while the cost_NPV (INV) reported in the vd file is 6428.
Yes, that small difference (<0.3%) was seen in your earlier results as well. I am not quite sure where it comes from but I suspected it being due to not using levelized annual costs. If you can provide the full model input files (*.DD, *.RUN), I could of course check. Levelized annual cost can be activated by $SET ANNCOST LEV in the RUN file. See modifying-run-files
Thank you Antti-L. I will run th model with ANNCOST LEV activated (see attached tfm_ins.png) and let you know if this works better. Also, i have one other question: In VEDA, the "levelized costs" option is set to "ExcludeEmissions" (see attached lce_option.png). Do we have to set this option in VEDA in addition to set RPT_OPT(‘NCAP’,’1’) = –1 in the ~TFM_INS table?
(22-03-2024, 03:57 PM)Antti-L Wrote: >for example, the sum of the discounted values for cost_inv gives around 6410, while the cost_NPV (INV) reported in the vd file is 6428.
Yes, that small difference (<0.3%) was seen in your earlier results as well. I am not quite sure where it comes from but I suspected it being due to not using levelized annual costs. If you can provide the full model input files (*.DD, *.RUN), I could of course check. Levelized annual cost can be activated by $SET ANNCOST LEV in the RUN file. See modifying-run-files
Thank you Antti-L. I will run th model with ANNCOST LEV activated (see attached tfm_ins.png) and let you know if this works better. Also, i have one other question: In VEDA, the "levelized costs" option is set to "ExcludeEmissions" (see attached lce_option.png). Do we have to set this option in VEDA in addition to set RPT_OPT(‘NCAP’,’1’) = –1 in the ~TFM_INS table?
Good news: with the ANNCOST LEV activated, the 0.3% difference disappears and i can therefore compute correctly the levelized costs. Thanks a lot Antti-L. However, the objective function has also changed (2% lower, while the cost_inv are 0.3% higher), which is not clear for me. Shouldn't the objective function remain constant, since the only modification in the model was the activation of ANNCOST LEV? Also, on a related topic, could you explain me how to compute the cost_inv values? It looks like it's not staightforward from NCAP_Cost and VAR_Ncap
22-03-2024, 11:19 PM (This post was last modified: 25-03-2024, 04:19 PM by Antti-L.)
I cannot see how the the switch ANNCOST LEV could affect the objective function, other than by the constant term related to NCAP_PASTI capacity investments. If you have defined NCAP_PASTI investments with investment costs, then there can indeed be a cosmetic difference, but otherwise there should not be any difference. In other words, because the constant term does not affect the solution, there should be no genuine difference.
The constant term may be changed, because levelized annual costs imply that all investment costs should be allocated to the model periods, and for the past investments some of the costs would usually fall outside the horizon, and those parts of the costs are therefore excluded when using levelized annual costs. The objective function can be fully reproduced from the annual costs when using levelized annual costs, and using it does not genuinely affect the objective function (or model solution), but only the cost reporting.
>Also, on a related topic, could you explain me how to compute the cost_inv values? It looks like it's not staightforward from NCAP_Cost and VAR_Ncap
With the levelized annual costs, Cost_inv is calculated by levelizing the investment cost to the operating periods in such a way that by multiplying all Cost_inv with Time_NPV you can reproduce the objective function term for investment costs (INV). Without the levelized annual costs option, Cost_inv simply represents the annualized investment payment at each Milestone year (see the documentation).
