Veda2.0 Released!


Modeling part loads
#1
Hi everyone,
I am using a model with 3 seasons, 24 hour timeslices and I have modeled part load operations using the following parameters
ACT_LOSPL~FX     0.075
ACT_CSTPL            89
ACT_LOSPL~LO     0.55
ACT~LOSPL~UP     0.9
ACT_MINLD             0.55
ACT_UPS~UP         0.6
ACT_UPS~LO         0.6
ACT_CSTUP           0.065
ACT_TIME~LO       24

For validation I took the inputs and outputs and calculated efficiency. Here, I am not getting an efficiency decline in part load operations. 
Please help. 

regards,
Pankaj


Attached Files Thumbnail(s)
           
Reply
#2
Quote:For validation I took the inputs and outputs and calculated efficiency. Here, I am not getting an efficiency decline in part load operations.

Thanks for the question. I assume that you are using the LP formulation, and not MIP?

Your statement above seems to imply that you have validated that you are getting partial load operations.  Can you post a screenshot showing the process activity by each timeslice in the season where you are seeing partial load operation occurring but no efficiency decline?
Reply
#3
Ok, as there was no response, I tried to replicate your model myself.

I thus created a small new model from scratch, and defined all the technologies and parameters that you have shown.

Indeed, I am seeing partial load operation for ELCTECOA446, in the following timeslices:
S1, S2, S3, S4, S5, S14, S15, S16, S17, S18
And indeed, I am also seeing partial load efficiencies in all of these timeslices. The full load efficiency is 0.26, and in all of the timelices mentioned above, the efficiency is lower than that. The efficiency is the lowest in the timeslices S1, S2, S3, S4, where it is 0.24186. This corresponds to an increase in the specific fuel consumption of 0.075, which is exactly as you defined for the minimum load. Indeed, the process load is at the minimum stable level in these timeslices.

In summary, as verified and validated by my test, using your model specification to the extent available, the partial load feature is working as expected, meaning that the efficiencies decline under partial load operation according to the input parameters specified.
Reply
#4
Sir,
Thank you for replying. I am using LP formulation for this. I have also observed that the plants are operating below minimum load levels which I have specified.
I am posting below the graphs for timeslice wise activity, efficiency and demand profile.


Attached Files Thumbnail(s)
   
Reply
#5
No, I think you are mistaken. What makes you claim that "plants are operating below minimum load levels which I have specified"? What is the online capacity in those timeslices? Surely you have understood that the unit size is infinitesimal when using the LP formulation?

If you want me to explain some details about your results, please express the questions clearly and post them here together with the full test model (which can be used to replicate your results). I will then answer the questions when I have been able to replicate the results.  

I am on holidays, and would thus hope that you help minimizing the extra time to be spent on this... Blush
Reply
#6
Sir,
I was referring to the hourly generation from my model and found that both the plants ELCTECOA446 and ELCTECOA447 operate at less than 55% loads. The online capacity is in the range of 2-5GW in the season S for both the plants of 15 and 10GW. 

My apologies for not specifying the problem clearly. 

I am attaching the test model along with the results file specifying VarFin and VarFout for processes. Perhaps I am mistaken somewhere and unable to demonstrate efficiency losses, increased fuel costs and higher emissions in timeslices with part load operations. It will be really great if you can help me out in it.


Attached Files
.xlsx   results.xlsx (Size: 51.41 KB / Downloads: 7)
Reply
#7
Ok, but let's make sure one can reproduce your results. I have now your VEDA templates, but having them does not guarantee that I can reproduce your model results.

So, to be sure that I can reproduce your results, could you please pack the model input files (*.DD, *.RUN), and also the *.LST file into a ZIP file and post that ZIP file here. All these files should be in your VEDA-FE work folder (usually GAMS_WrkTIMES).

Only by having the model input files I can be sure I can reproduce your results.  And the listing file will disclose me the TIMES version that you are using. Could you thus please do me the favour and provide these files?
Reply
#8
Yes sir. 
Please find attached the zip file with *.LST , *.DD and *.RUN files in attachment.


Attached Files
.zip   run files.zip (Size: 42.27 KB / Downloads: 2)
Reply
#9
The ZIP included only BASE.DD, no other DD files.

The following files are missing:
base_ts.dd
syssettings.dd
Reply
#10
(17-07-2018, 04:25 AM)Antti-L Wrote: The ZIP included only BASE.DD, no other DD files.

The following files are missing:
base_ts.dd
syssettings.dd

Sir,
Please find attached the updated folder with base_ts.dd and syssettings.dd


Attached Files
.zip   run files.zip (Size: 43.04 KB / Downloads: 8)
Reply
#11
Thanks for providing the model files.  

I was able to reproduce your results, and found out that the problem with your partial load efficiencies was caused by defining the ACT_EFF parameters on the PG flow, instead of the shadow side flows. The documentation says it clearly:

"…the endogenous modelling of partial load efficiencies requires that the process has its efficiency modelled through the ACT_EFF parameter (on the shadow side of the process)."

I corrected this modeling flaw (by using ACT_EFF(NRG)), and immediately got reasonable results:
 • The efficiencies decline as defined under partial load operation
 • The minimum stable load levels are respected in all cases

The results tor the M and the S season are illustrated in the Figures below.
       

I attach also the Excel file where I made these results processing calculations.
.xls   PankaPL.xls (Size: 391 KB / Downloads: 19)

Final remark:  You have defined very strange PRC_CAPACT coefficients.  PRC_CAPACT defines the unit conversion form the capacity unit to the activity unit.  As far as I can understand, in your model these units are GW and PJ.  Therefore, the correct unit conversion factor would be PRC_CAPACT=8.76*3.6=31.536.  But you have defined very different values for PRC_CAPACT, which do not make sense to me (they are even different for each process).  For example, according to the values defined, the 10 GW technology would be able to produce only 0.64 TWh in a year with full load, even though the correct amount is 87.6 TWh. Anyway, to derive the consistent load level in terms of capacity units, the PRC_CAPACT factors must be used for the unit conversion.
Reply
#12
Sir,
Thank you for spending so much time on the model and posting a very detailed explanation.
Indeed the part in the documentation got ignored and I should have modeled ACT_EFF on the shadow side of the process instead of primary group.
I will make suitable changes to efficiency & PRC_CAPACT and proceed ahead. Thank you for pointing it out.
Reply


Forum Jump:


Users browsing this thread: 1 Guest(s)