Implement Kotzur method for representative days

Open fneum opened this issue 5 months ago • 1 comments

Theory

Kotzur et al. (2018)

https://arxiv.org/pdf/1710.07593

Other Implementations

Calliope: https://calliope.readthedocs.io/en/stable/user/advanced_features.html#time-resolution-adjustment
eTraGo in older PyPSA version: https://github.com/openego/eTraGo/blob/bc3447c2c9b2df77bbf3c66168ed67846eca38c7/etrago/tools/constraints.py#L2420-L3473
FINE: https://vsa-fine.readthedocs.io/en/master/mathematicalDocumentation/storageDoc.html

Motivation

While hourly consecutive modelling has become the norm in many energy system models, several types of problems may make it necessary to reduce the time dimension. One key example is capacity expansion with multiple investment periods (with piecewise linearised endogenous technological learning). See Zeyen et al. (2023): https://www.nature.com/articles/s41467-023-39397-2

Tips

Implement simplification in Appendix B. https://github.com/PyPSA/PyPSA/issues/36#issuecomment-704752151

Jul 10 '25 20:07 fneum

Not sure how helpful this is, but I did some work on this 4 years ago https://github.com/lisazeyen/learning_curve/blob/d9a5742a7d7ed784a9e37a51563dc90f32765990/scripts/pypsa_learning/temporal_clustering.py#L178

Oct 23 '25 08:10 lisazeyen