I have a question about early retirement: What is the best way to prevent sudden retirement of existing capacities and switch to new technologies?
Of course TIMES selects between different technologies based on the system's total cost while fulfills defined constraints, but sometimes over the considered time horizon we see sudden switch between technologies which is not practically favorable. For example a particular technology is implemented in 2020 (assume a technology lifetime of 30 years) but is fully retired early (e.g in 2030) and instead other technologies are used to supply demands. However, our concern is that this may not be practically desired to invest in a technology while it is retired soon, although from the modeling point of view it is fully feasible and cost-minimum. So, in order to prevent sudden switches between technologies (for practical considerations), does it make sense to constrain the level of retirement of any existing technology (e.g. maximum 30% of the technology A can be retired every 5 years). Should we define such a constraint based on the activity of each technology? I would be grateful to have your advise on what is the best way to define such a constraint.
17-06-2022, 04:53 PM (This post was last modified: 17-06-2022, 05:38 PM by Antti-L.)
>What is the best way to prevent sudden retirement of existing capacities and switch to new technologies?
The terminology ("retirement") may cause some confusion here: There are no early retirements, unless the modeller has explicitly enabled such, for each individual technology. By default they are all disabled. Early retirement means that the capacity is removed from the stock of available capacity.
Possibly you don't actually mean retirement, but that some technologies are no longer being operated, even though the capacity is still there, fully available for use? If that is the case, then you can just define a lower bound for the annual capacity factor (utilization factor), or even by timeslice. NCAP_AFA(r,y,p,'LO') is the parameter for defining lower bounds for capacity utilization rates.
However, it sounds a bit strange if you have investment into new capacity in 2020, with a 30 years' lifetime but used only for less than ten years. As you say, it sounds like a bad investment. Did the model actually decide on that investment, or is it a "real-world" investment that has taken place? If the latter, do you know the explanation for the seemingly poor decision? Are there some new policies implemented making it so adversely affected?
>However, it sounds a bit strange if you have investment into new capacity in 2020, with a 30 years' lifetime but used only for less than ten years. As you say, it sounds like a bad investment. Did the model actually decide on that investment, or is it a "real-world" investment that has taken place? If the latter, do you know the explanation for the seemingly poor decision? Are there some new policies implemented making it so adversely affected?
Actually the model decides on that investment: the reason for such a switch between technologies is the CO2 emission constraint which should reach to zero in 2050 compared to current unsustainable production. In attachments I provided figures which shows the share of technologies for producing ammonia from 2020 to 2050 considering emission reduction targets. For example we see that compared to 2040, in 2050 ATR technology disappears and is replaced by water electrolysis, because of reaching the net zero emission target (NZE). However, my point is that how we can have such a transition between technologies smoother? although the model predictions is based on the minimum cost while the emission constraint is satisfied but such a thing may not be very practically favorable. So, we would like to limit such a sudden switch (for practical considerations), although at the end we may have ammonia with higher production cost. I would be grateful to have your opinion for that.
I did ask you a question, but I could not see any answer. But ok, let's try again:
I think the results you showed should not even be possible if the following would hold:
1) you do NOT enable early retirements for these techs;
2) you define lower bounds for the utilization factors (e.g. on annual basis);
3) the lifetime of those techs is long enough (such as 30 years).
Why so? Because if you define lower bounds for the capacity utilization, the technologies will have to be used during their whole technical lifetime, at least at that minimum rate. And you could additionally also shape the minimum utilization factor by age, for some additional smoothing.
As you say yourself, your figure shows e.g. the ATR technology appearing and disappearing. So, that implies that you are either explicitly allowing early retirements, or you have not defined any lower bounds on capacity utilization, correct? So, why did you not force the transition between technologies to be smoother?
In your first reply you asked whether it is the model decision or it is the real world investment. I replied your question by mentioning that it is the model decision and just tried to elaborate on that by providing our example.
Anyway, based on your answer, now I assume that we will have smoother transition by considering a lower bound on utilization factor which we have not yet tried (we did not allow early retirement either).
Thanks for your recommendation, your follow up on that and the great discussion.
01-07-2022, 01:30 PM (This post was last modified: 02-07-2022, 02:57 AM by Antti-L.)
No, that was only one, less important question. To narrow down the possibilities, my most crucial question was:
AL>Possibly you don't actually mean retirement, but that some technologies are no longer being operated, even though the capacity is still there, fully available for use?
But thank you, now you finally answered this crucial question, which now clarifies that your issues are not at all related to the subject of this thread: "we did not allow early retirement".
Therefore, only now there would some point thinking about possible other reasons for the unexpected behavior you are seeing, which may also be related to modeling errors. I would thus suggest that you also check that these processes have in the results the investment costs you have defined (e.g. from Cap_NEW(LUMPINV), and the capacity variables VAR_CAP.L. In addition, a very large general discount rate might also produce such behaviour, and so check that the rate is in the correct range. For example, 5% should be written in the input data as 0.05, and not 5. 'Just in case'...
