Require technology progression

[mb537]

I'm sure there's a way to implement what we're trying to achieve, but I haven't yet found documentation describing it yet. 

The objective is to build a novel technology pathway that, as capacity for the commodity flowing through it increases, will be able to "upgrade" to a more efficient and lower cost technology. 

Lets say we have the three below transmission technologies:
TransmissionA $$$
TransmissionB $$
TransmissionC $

Transmission A is the most expensive and least efficient, but because this is a novel technology we're starting with what we have. As capacity of this technology increases, we assume that there will be advancements that lead it to get cheaper and more efficient, advancing to TransmissionB, and finally TransmissionC. The three technologies all have the same purpose of moving a fuel commodity through them, but they will vary with respect to ncap_cost, fom, and fuel requirements. It seems to me that we'd need to describe three separate technologies and create a "threshold" for them coming in. 

One thought is to have the parameters that differ between the technologies change, but the only option I have found is to define, extrapolate, or interpolate between time periods. I have not found an attribute that allows for defining a parameter change based on capacity. Creating a minimum capacity or activity bound forces in a given technology, whereas I want it to act more like a "threshold" that it must reach before it's allowed to build. Is there a way to implement a threshold like this? Or is there a parameter that lets me set a value for the minimum amount of capacity a technology can be built at? For example, TransmissionA can be built out at any size, large or small, but TransmissionB and TransmissionC can only be built with a minimum capacities of 10 and 50 PJ? 

Alternatively, the TIMES documentation lays out some of the framework for endogenous technological learning (ETL), but I haven't been able to find examples of ETL implementation on the forum or in the documentation, so I'm not sure if its really applicable in this case or would be the right solution. 

Any assistance would be greatly appreciated!

[Antti-L]

What you are seeking is a non-convex condition, so not possible with a pure LP model. But yes, you could use the NCAP_SEMI(r,y,p) parameter for defining a minimum size for the new capacity installed for a process. "Semi" comes from semi-continuous. One should use NCAP_SEMI together with NCAP_BND(r,t,p,'UP') for defining the upper bound for newly installed capacity (if not defined, the upper bound will be set equal to NCAP_SEMI(r,t,p)). Note also the default I/E (=MIG).

Because of the non-convexity, using NCAP_SEMI will result in a MIP problem, and will thus take considerably longer to solve. One should also use the OPTCR tolerance for setting the MIP optimality tolerance.  Using ETL may indeed be another option, and will likewise result in a MIP problem (but still harder). I think it is well described in the docs (Part I & Part II), but I have never tried using it myself.

[mb537]

This has been extraordinarily helpful, thanks Antti!