tsam
Time series aggregation for faster energy system models.
I use tsam in my day job, and flixopt uses it under the hood. Long-horizon energy models get expensive fast, and tsam is one of the standard tools for cutting them down to size — it picks a small set of representative periods (typical days or weeks) from large input time series so optimization runs in tractable time. The codebase had grown organically out of academic research, and modernizing it is what pulled me in.
Originally developed at Forschungszentrum Jülich (IEK-3) and used across the energy systems community. I’m a heavy contributor — modernizing the code, working on a full rewrite, with the xarray-native interop landing in tsam_xarray.