Memory usage in myopic electricity-only Spain (8-node) scenario 2025–2050

[Alicia Mortera Canga]

Hi all,

I’m building a Spain-only PyPSA-Eur model clustered to 8 nodes, using a myopic approach (one year at a time). The goal is to optimize the different components under different policy/real-world deployment scenarios.

If I try to optimize every single year from 2025 to 2050 in one go, I hit memory limits and the run crashes. If instead I advance in 5-year steps (2025→2030→2035…), it works. I’ve simplified parts of the code, and noticed memory is very sensitive to the number/type of components included and how constrained the model is.

Environment: PyPSA-Eur 0.8.1, PyPSA 0.25.2, Python 3.10.12, solver Gurobi, 32 GB RAM, Windows + WSL Ubuntu.

Has PyPSA-Eur introduced a myopic workflow that distinguishes between “fixed/inherited” capacity (coming from previous years) and “new/optimizable” capacity, similar to MyPyPSA-Ger? Are there any additional recommendations to help address this issue? I’m thinking of upgrading the RAM on my machine.