- Multiscale-modelling of the electrochemical reactions inside a PEM fuel cell
- Model- and machine learning based investigation of material changes during conditioning
- Computer-aided image analysis for validation of material parameters
- Sensitivity analysis to isolate parasitic influence of each parameter
- Goal: Minimize the duration of the conditioning process
Especially in the automotive sector, Fuel Cells provide weight and range benefits compared to large battery packs. However, for the moment the long duration of the conditioning process hinders a large production volume and increases cost: After production fuel cell usually only achieve a fraction of their maximal power, only after a break-in procedure, also called conditioning, the desired power is achieved and assembly into the vehicle is possible. The conditioning process currently takes several hours for completion, thus massively hinders mass production for the automotive sector.
The aim of this work is to understand the material changes during conditioning and identify the main contributors that prolong the conditioning procedure. This generated knowledge can then be used to minimize the conditioning needed. For this, model-based as well as machine-learning techniques are used, in order to reduce costly generation of data, increase data independence, but also be able to approximate effects that can hardly be modelled. Model-order reduction techniques are used to create a realistic computational model of the electrochemical processes during conditioning while reducing the computational load.