CCS retrofit transition

[ejin]

Hi Antti and community,

I am simulating the decarbonization technologies for the ammonia production. The carbon emission is constrained to be decreased granularly from 2020 to 2050. The result of low-carbon technologies transition look good between 2020 to 2040. In 2045, the major technologies for ammonia production include ATR , ATR with CCS retrofit, ATR coupled with CCS (where the ATR plants built with CCS at the same time), and pyrolysis. However, in 2050, the technology share of ATR coupled with CCS increases and ATR decreases rather than the ATR with CCS retrofit increases and ATR decreases. It means the system would rather build new ATR coupled CCS plants than retrofitting the existing ATR plants with CCS. This result doesn't make sense because all cost parameters of building the new ATR coupled with CCS plants are larger than those of ATR with CCS retrofit. 

I used PCR_REFIT to fulfill this simulation and enable the retirement function. I am not sure if I simulate it correctly and don't know why I got the unexpected result. One reason I guess would be the total cost of building the new ATR coupled with CCS plants is cheaper than the retrofitting options based on the cost calculation in TIMEs, but still not clear. 

Looking forward to relies.

Thanks!!

[Antti-L]

Interesting question. You say:

>  It means the system would rather build new ATR coupled CCS plants than retrofitting the existing ATR plants with CCS. This result doesn't make sense because all cost parameters of building the new ATR coupled with CCS plants are larger than those of ATR with CCS retrofit.

First of all, make sure again that you have read the documentation, in particular the following:

  ● Investing into a retrofit option will never extend the original lifetime of the host process, but investing into a lifetime extension option will extend its lifetime in accordance with the full technical lifetime of the LE option.
  ● Retrofit options never have any salvage value, but lifetime extension options will have a salvage value whenever the lifetime of the LE option would extend beyond the model horizon, and is accounted according to the standard salvage value accounting in TIMES.

I tried to verify your conclusion about "doesn't make sense". As far as I can see, your NCAP_FOM (fixed O&M cost) and ACT_COST are in fact the same for both options. For NCAP_COST, I think you have the values 1224.5 and 410. Using these values, I calculated the effective capital cost for the (assumed) last model period in Excel, assuming a 5% discount rate, and I got a capital cost smaller for the ATRNH3_CCS_New process than for the retrofit ATRNH3_CCS_retrofit process. Therefore, to me it would seem to make sense. 

Could you therefore either show your calculation for comparing the overall costs between ATRNH3_CCS_New and ATRNH3_CCS_retrofit, for the 2050 period, so that I could see where I am making my mistake, or otherwise confirm that you had actually not taken into account the differences between a retrofit investment and a new investment when you calculated the cost impact?

If you would rather like that the investment cost of a retrofit option should be proportional to the remaining lifetime, you could convert those costs to a fixed O&M cost. Or, if you want it to have a salvage value, you could use the LE option instead. Moreover, if you would conversely rather like that the CCS_new technology would not have a salvage value (even though salvage value accounting is standard in TIMES and designed by highly qualified economists), that's also possible, by using NCAP_FDR.

[ejin]

Interesting question. You say:

>  It means the system would rather build new ATR coupled CCS plants than retrofitting the existing ATR plants with CCS. This result doesn't make sense because all cost parameters of building the new ATR coupled with CCS plants are larger than those of ATR with CCS retrofit.

First of all, make sure again that you have read the documentation, in particular the following:

  ● Investing into a retrofit option will never extend the original lifetime of the host process, but investing into a lifetime extension option will extend its lifetime in accordance with the full technical lifetime of the LE option.
  ● Retrofit options never have any salvage value, but lifetime extension options will have a salvage value whenever the lifetime of the LE option would extend beyond the model horizon, and is accounted according to the standard salvage value accounting in TIMES.

I tried to verify your conclusion about "doesn't make sense". As far as I can see, your NCAP_FOM (fixed O&M cost) and ACT_COST are in fact the same for both options. For NCAP_COST, I think you have the values 1224.5 and 410. Using these values, I calculated the effective capital cost for the (assumed) last model period in Excel, assuming a 5% discount rate, and I got a capital cost smaller for the ATRNH3_CCS_New process than for the retrofit ATRNH3_CCS_retrofit process. Therefore, to me it would seem to make sense. 

Could you therefore either show your calculation for comparing the overall costs between ATRNH3_CCS_New and ATRNH3_CCS_retrofit, for the 2050 period, so that I could see where I am making my mistake, or otherwise confirm that you had actually not taken into account the differences between a retrofit investment and a new investment when you calculated the cost impact?

If you would rather like that the investment cost of a retrofit option should be proportional to the remaining lifetime, you could convert those costs to a fixed O&M cost. Or, if you want it to have a salvage value, you could use the LE option instead. Moreover, if you would conversely rather like that the CCS_new technology would not have a salvage value (even though salvage value accounting is standard in TIMES and designed by highly qualified economists), that's also possible, by using NCAP_FDR.

Hi Antti, 

Thanks for your explanation and your suggestions. I noticed that the salvage value is not considered when it has a retrofit option but I don't expect it has a big impact on the effective capital cost. The NCAP_COST values you used are correct, can you show your calculations for the effective capital costs of two technologies? I don't know what formula (that would be really helpful if you can provide it) to calculate it, and also I don't calculate the overall costs for those two technologies, but I attached the total cost calculated by TIMEs itself. Again, I considered the investment differences between retrofit and new plant options, but just don't expect the significant impact of salvage value.

I tried to use the LE option instead, but again the results are the same with considering the salvage value of CCS retrofit. If the effective capital cost of ATRNH3_CCS_New is cheaper than the retrofit, why the ATR_CCS_retrofit technology increase between 2040 and 2045? 

I attached the total salvage values and also the levelized cost for each technology (under LE option) in case you need more information.

Thanks,

Enze

[Antti-L]

> I don't know what formula (that would be really helpful if you can provide it) to calculate it

Have you not been reading the documentation? All the cost accounting formulas related to the objective function are given in detail in Part II of the documentation, see Documentation_for_the_TIMES_Model-Part-II.pdf

Using the formulas, and assuming the length of the last period is 5 years and the discount rate is 5% (I don't know if these are exactly correct, as you did not provide that information), you should immediately see that the salvage value of ATRNH3_CCS_New is 73.56% of the investment cost. What this means is that you would get  1224.5 × (1−0.7356) = 323.768 as the effective capital cost. And then, by comparing this with 410 you can immediately see that it is cheaper to invest in the ATRNH3_CCS_New technology than into the retrofit.

Could you please explain why you think that you don't "expect the significant impact of salvage value". The technology has a lifetime of 35 years, and the length of the last period is (I assume) only 5 years. So, most of the investment costs will obviously be credited back as the salvage value.

[ejin]

> I don't know what formula (that would be really helpful if you can provide it) to calculate it

Have you not been reading the documentation? All the cost accounting formulas related to the objective function are given in detail in Part II of the documentation, see Documentation_for_the_TIMES_Model-Part-II.pdf

Using the formulas, and assuming the length of the last period is 5 years and the discount rate is 5% (I don't know if these are exactly correct, as you did not provide that information), you should immediately see that the salvage value of ATRNH3_CCS_New is 73.56% of the investment cost. What this means is that you would get  1224.5 × (1−0.7356) = 323.768 as the effective capital cost. And then, by comparing this with 410 you can immediately see that it is cheaper to invest in the ATRNH3_CCS_New technology than into the retrofit.

Could you please explain why you think that you don't "expect the significant impact of salvage value". The technology has a lifetime of 35 years, and the length of the last period is (I assume) only 5 years. So, most of the investment costs will obviously be credited back as the salvage value.

Thanks for your reply.

I will find the formula in the documentation then. 

I thing I misunderstood the salvage value for the investment happened in the last 5 years, which I thought it was still 35 years. Your explanation makes sense to me. Thanks!!

[ejin]

> I don't know what formula (that would be really helpful if you can provide it) to calculate it

Have you not been reading the documentation? All the cost accounting formulas related to the objective function are given in detail in Part II of the documentation, see Documentation_for_the_TIMES_Model-Part-II.pdf

Using the formulas, and assuming the length of the last period is 5 years and the discount rate is 5% (I don't know if these are exactly correct, as you did not provide that information), you should immediately see that the salvage value of ATRNH3_CCS_New is 73.56% of the investment cost. What this means is that you would get  1224.5 × (1−0.7356) = 323.768 as the effective capital cost. And then, by comparing this with 410 you can immediately see that it is cheaper to invest in the ATRNH3_CCS_New technology than into the retrofit.

Could you please explain why you think that you don't "expect the significant impact of salvage value". The technology has a lifetime of 35 years, and the length of the last period is (I assume) only 5 years. So, most of the investment costs will obviously be credited back as the salvage value.

Thanks for your reply.

I will find the formula in the documentation then. 

I thing I misunderstood the salvage value for the investment happened in the last 5 years, which I thought it was still 35 years. Your explanation makes sense to me. Thanks!!

Hi Annti,

I used the formula to calculate the salvage value and got the same value as yours (assuming 5% discount rate). So does it mean if the CCS_retrofit option has a lower investment than 323.768, the CCS_retrofit will be used? If yes, I changed it to 200, but the results are still the same. In addition, I mentioned that I changed retrofit option to LE option, and the results don't change at all. Any thoughts?

Thanks!!

[Antti-L]

No, I have no idea why your results do not change, without seeing the model.

But I am sure I would have an idea, if I had the model files, and I could reproduce your findings.  If you would like me to investigate it, please provide me with the *.DD and *.RUN files (all of these from the VEDA work folder) for the case that you wish the investment decision to be explained (e.g. with NCAP_COST=200). If you would like that, a download link via Dropbox, Google drive or equivalent would be fine (you could send the link e.g. in a private message).

[Antti-L]

Ok, while you were not responding, I made a test model trying to simulate a similar model behaviour. And I  was able to identify the apparent reason for the seemingly unexpected result, which you raised:

According to the design, a retrofitted process will not be credited by any salvage value.  Therefore, when retrofitting a process at the end of the model horizon, one will not only lose any salvage value of the retrofit investment but also the salvage value of the host process. The latter would indeed be credited, if the process would not be retrofitted. Therefore, a retrofit at the end of the model horizon implies a cost disadvantage arising from these two factors, while our earlier comparison (323.768 vs. 410) only took into account one of these factors (the lack of salvage value from the retrofit investment).

I also notice now that the reporting of the salvage values is currently incorrectly including salvage values also for retired capacities (e.g. for a retrofitted host process, where the original capacity is internally considered to be retired when the retrofit is installed and replaces it). This is a minor reporting bug, which will be fixed in the next version of TIMES, but otherwise it has no impact on the model solution or the value of the objective function.

It may, of course, be debatable whether a retrofitted process should be credited by some salvage value or not, but for now, the implementation follows the design, which assumes that no salvage value is credited.

You may want to consider adding one additional period to the model to reduce this kind of end-of-horizon effect.  For example, adding a 15-year period at the end (with 2060 as a milestone) would probably largely eliminate such.

[ejin]

Ok, while you were not responding, I made a test model trying to simulate a similar model behaviour. And I  was able to identify the apparent reason for the seemingly unexpected result, which you raised:

According to the design, a retrofitted process will not be credited by any salvage value.  Therefore, when retrofitting a process at the end of the model horizon, one will not only lose any salvage value of the retrofit investment but also the salvage value of the host process. The latter would indeed be credited, if the process would not be retrofitted. Therefore, a retrofit at the end of the model horizon implies a cost disadvantage arising from these two factors, while our earlier comparison (323.768 vs. 410) only took into account one of these factors (the lack of salvage value from the retrofit investment).

I also notice now that the reporting of the salvage values is currently incorrectly including salvage values also for retired capacities (e.g. for a retrofitted host process, where the original capacity is internally considered to be retired when the retrofit is installed and replaces it). This is a minor reporting bug, which will be fixed in the next version of TIMES, but otherwise it has no impact on the model solution or the value of the objective function.

It may, of course, be debatable whether a retrofitted process should be credited by some salvage value or not, but for now, the implementation follows the design, which assumes that no salvage value is credited.

You may want to consider adding one additional period to the model to reduce this kind of end-of-horizon effect.  For example, adding a 15-year period at the and (with 2060 as a milestone) would probably largely eliminate such.

Hi Annti,

Sorry for the late reply. I will PM you the Dropbox link for downloading the files and appreciate that you can run and test it to identify the solution for me.  Let me know if you need anything else.

Thanks you so much.

[Antti-L]

Thanks, but under your download link, there was only a single *.DD file, us_nh3_base_ts.dd, with the following contents:
---------
SET ALL_TS
/
ANNUAL
/
----------
That obviously cannot be the full input data of your TIMES model.
However, the RUN file you included (us_nh3_base.run) shows that the model input *.DD files should actually comprise of the following six *.DD files:
----------
us_nh3_base_ts.dd
base.dd
syssettings.dd
baselineco2reduction.dd
ccs_retrofit.dd
mp_cap_constrain.dd
----------

So, as you did you not include these six necessary *.DD files, I am sorry but I cannot reproduce your run, nor investigate it in any way.

However, as I tried to explain, I did already identify the cause for the  seemingly unexpected result, which you raised.  I explained it in my previous post above. Do you think there is still something to be investigated?  If so, could you please provide me with all the *.DD and *.RUN files for the relevant case, thank you.

[ejin]

Thanks, but under your download link, there was only a single *.DD file, us_nh3_base_ts.dd, with the following contents:
---------
SET ALL_TS
/
ANNUAL
/
----------
That obviously cannot be the full input data of your TIMES model.
However, the RUN file you included (us_nh3_base.run) shows that the model input *.DD files should actually comprise of the following six *.DD files:
----------
us_nh3_base_ts.dd
base.dd
syssettings.dd
baselineco2reduction.dd
ccs_retrofit.dd
mp_cap_constrain.dd
----------

So, as you did you not include these six necessary *.DD files, I am sorry but I cannot reproduce your run, nor investigate it in any way.

However, as I tried to explain, I did already identify the cause for the  seemingly unexpected result, which you raised.  I explained it in my previous post above. Do you think there is still something to be investigated?  If so, could you please provide me with all the *.DD and *.RUN files for the relevant case, thank you.

Hi Antti,

Thanks for your reply. Your explanation makes sense to me. But we definitely don't want to extend our horizon to 2060, but I will try to see if it will solve the problem. In addition, I mentioned before that I changed RF to LE option, but results are the same. It means although the salvage value is considered for CCS_retrofit, the salvage value of replaced host process is not considered, right? Maybe you can run the model and help me figure out the solution without extending the timeline to 2060. I've uploaded the files you need to the same link. Let me know if you need anything else.

Thanks a lot for your help!!

[Antti-L]

Ok, thanks, I got the *.DD files now.  I also tried running the model case, and that went fine.

Indeed, the host process will lose its salvage value also when it is replaced by an LE option, and only the latter will then have a salvage value.

> Maybe you can run the model and help me figure out the solution without extending the timeline to 2060.

Sure, but I am on holidays now, and so it may take a few days for me to look at your model in detail. But for suggesting a solution, please first describe what kind of a solution you would need to have, and why?  It seems that for you don't like that the ATRNH3 is still producing in 2050?  Do you think the retrofitting option should be given salvage credits?

[ejin]

Ok, thanks, I got the *.DD files now.  I also tried running the model case, and that went fine.

Indeed, the host process will lose its salvage value also when it is replaced by an LE option, and only the latter will then have a salvage value.

> Maybe you can run the model and help me figure out the solution without extending the timeline to 2060.

Sure, but I am on holidays now, and so it may take a few days for me to look at your model in detail. But for suggesting a solution, please first describe what kind of a solution you would need to have, and why?  It seems that for you don't like that the ATRNH3 is still producing in 2050?  Do you think the retrofitting option should be given salvage credits?

Hi Antti,

No worries. 

But for suggesting a solution, please first describe what kind of a solution you would need to have, and why?  It seems that for you don't like that the ATRNH3 is still producing in 2050?

The ATRNH3 that is still producing in 2050 makes sense for me. But the ATRNH3_CCS_new plants are invested to replace a portion of ATRNH3 instead of just retrofitting (ATRNH3_CCS-retrofit) the same portion of existing ATRNH3, which doesn't make sense to me intuitively. As you explained, I know the reason. So the solution I want to get or is it possible to consider the salvage value of the replaced host processes (ATRNH3) while we are using PCR_REFIT attribute? Or are there other ways to simulate the CCS retrofit option for ATRNH3 without using PCR_REFIT?

Thanks!!

[Antti-L]

Thanks for sharing this small test model, which proved to be useful for making some improvements into the Retrofits and Lifetime extension feature of TIMES.

Although simulating retrofits with user constraints would also be to some extent possible, it would be hard to make it working accurately by process vintage, and thus working well in general. In the general case, modeling retrofits and lifetime extensions requires relatively complex considerations, and therefore, I think the built-in feature is useful and handy, and I have now improved several aspects of it:

  •  There were some reporting bugs related to retrofits which have been fixed;
  •  The option of forcing retrofits to be equal to retirements has been made more consistent;
  •  When using the option of forcing retrofits to be equal to retirements, one can now also enable salvage credits for the host process (because it may well be assumed to retain its value when retrofitted), by specifying NCAP_FDR for the host process.

These improvements are available in TIMES v4.6.4.

A further enhancement would in principal also be possible, by allowing salvage credits for the retrofitted capacity also when that is less than the retired capacity of the host process. However, this is not possible in the current implementation approach, because it would require a somewhat different approach with some more variables and equations. Therefore, such a more general option is still not available in TIMES v4.6.4.

However, I think the improvements in v4.6.4 will offer a satisfactory solution to the problem you raised:  With 1) forcing retrofits to be equal to retirements, and 2) instructing TIMES that salvage values are to be credited, the ATRNH3 is now getting retrofitted also at the end of the model horizon, as you had expected. The figure below illustrates the solution using these options: 
   

The new release of TIMES should be available tonight: TIMES release v4.6.4
I hope you can download the new version, and take a look if that solves the problem for you.

[ejin]

Thanks for sharing this small test model, which proved to be useful for making some improvements into the Retrofits and Lifetime extension feature of TIMES.

Although simulating retrofits with user constraints would also be to some extent be possible, it would be hard to make it working accurately by process vintage, and thus working well in general. In the general case, modeling retrofits and lifetime extensions require relatively complex considerations, and therefore, I think the built-in feature is useful and handy, and I have now improved several aspects of it:

  •  There were some reporting bugs related to retrofits which have been fixed;
  •  The option of forcing retrofits to be equal to retirements has been made more consistent;
  •  When using the option of forcing retrofits to be equal to retirements, one can now also enable salvage credits for the host process (because it may well be assumed to retain its value when retrofitted), by specifying NCAP_FDR for the host process.

These improvements are available in TIMES v4.6.4.

A further enhancement would in principal also be possible, by allowing salvage credits for the retrofitted capacity also when that is less than the retired capacity of the host process. However, this is not possible in the current implementation approach, because it would require a somewhat different approach with some more variables and equations. Therefore, such a more general option is still not available in TIMES v4.6.4.

However, I think the improvements in v4.6.4 will offer a satisfactory solution to the problem you raised:  With 1) forcing retrofits to be equal to retirements, and 2) instructing TIMES that salvage values are to be credited, the ATRNH3 is now getting retrofitted also at the end of the model horizon, as you had expected. The figure below illustrates the solution using these options: 


The new release of TIMES should be available tonight. I will add a link here when it is ready.
I hope you can download the new version, and take a look if that solves the problem for you.

Hi Antti,

Thanks for your information. I will try the new version of TIMES and see if I can get the expected results as yours.