I’m not 100% happy with how the model’s endogenous learning curves work today. There currently is no way to override them with a known (i.e. exogenous) pricing schedule, and they reflect learning from global deployment of the technology, which greatly limits the extent to which the modeled region's deployment levels can drive down costs (and hence, somewhat defeats the purpose of using an endogenous learning curve instead of an exogenous price schedule). Here are two features that could help:
Introduce a new variable that specifies the fraction of global or out-of-region deployment to take into account when making endogenous learning calculations. At the 100% setting, the model would work as it does today. At the 0% setting, the model would only consider in-region deployment (both historical and future) when calculating learning curves. The model could blend the two calculations in any ratio designated by the modeler’s choice of setting for this input variable.
Allow the endogenous learning calculations to be overridden with exogenous values, at least for power plants, if time-series data is provided for years beyond the start year for capital costs in elec/CCAMC. The model would check the data file, and if it sees a zero, that’s a flag for the model to use the endogenous learning curves. If it sees a cost number, it will use that.
I’m not 100% happy with how the model’s endogenous learning curves work today. There currently is no way to override them with a known (i.e. exogenous) pricing schedule, and they reflect learning from global deployment of the technology, which greatly limits the extent to which the modeled region's deployment levels can drive down costs (and hence, somewhat defeats the purpose of using an endogenous learning curve instead of an exogenous price schedule). Here are two features that could help:
Introduce a new variable that specifies the fraction of global or out-of-region deployment to take into account when making endogenous learning calculations. At the 100% setting, the model would work as it does today. At the 0% setting, the model would only consider in-region deployment (both historical and future) when calculating learning curves. The model could blend the two calculations in any ratio designated by the modeler’s choice of setting for this input variable.
Allow the endogenous learning calculations to be overridden with exogenous values, at least for power plants, if time-series data is provided for years beyond the start year for capital costs in elec/CCAMC. The model would check the data file, and if it sees a zero, that’s a flag for the model to use the endogenous learning curves. If it sees a cost number, it will use that.