Surrogate models are an efficient method to expedite the process design by superstructure optimization. For their application in biorefineries’ process design, several surrogate models are benchmarked in a case study regarding their validation metrics and their performance in a reference superstructure optimization problem. Despite good validation metrics for most surrogate models, their prediction quality in the superstructure optimization does not reflect this. For the use of surrogate models in superstructure optimization, the need for a profound assessment of options, and the possible use of dynamic sampling strategies become evident.
|Conference||31st European Symposium on Computer Aided Process Engineering |
|Period||06/06/2021 → 09/06/2021|
|Series||Computer Aided Chemical Engineering|
- Process Design
- Superstructure Optimization
- Surrogate Modelling
- Machine Learning