Towards Digital Twins for Power-to-X: Comparing Surrogate Models for a Catalytic CO2 Methanation Reactor

by Luisa Peterson, Ali Forootani, Edgar Ivan Sanchez Medina, Ion Victor Gosea, Kai Sundmacher, and Peter Benner

Contains

Python implementation to reproduce the results in [1]. The dependencies are given in each folder as different Python environments are used for all subroutines.

• 01_First principle model: code to reproduce the mechanistic data
• 02_Full order model data: data from the mechanistic model that is used for the surrogate models
• 03a_GNN: code to build up the GNN surrogate model
• 03b_GNN+POD: code to build up the GNN surrogate model on reduced data
• 04_OpInf: code to build up the OpInf surrogate model
• 05_GN-SINDy: code to build up the GN-SINDy surrogate model

Runtime

We will report the runtime of each code inside the respective folder. To obtain the times, we used the same computer system with the characteristics given below.

Computational Resources:

• Processor: 12th Gen Intel Core i5-12600K
• Memory: 32.0 GB
• Graphics: NVIDIA Corporation/Mesa Intel Graphics (ADL-S GT1)
• OS: Ubuntu 20.04.6 LTS (64-bit)
• Storage: 1.3 TB

License

See the LICENSE file for license rights and limitations (MIT).

References

[1] Luisa Peterson, Ali Forootani, Edgar Ivan Sanchez Medina, et al. Towards Digital Twins for Power-to-X: Comparing Surrogate Models for a Catalytic CO2 Methanation Reactor. TechRxiv. August 02, 2024. DOI: 10.36227/techrxiv.172263007.76668955/v1