Mainz 2022 – wissenschaftliches Programm

Bereiche | Tage | Auswahl | Suche | Aktualisierungen | Downloads | Hilfe

P: Fachverband Plasmaphysik

P 11: Codes and Modelling

P 11.5: Vortrag

Mittwoch, 30. März 2022, 15:00–15:15, P-H11

Application of surrogate models for tokamak edge plasma simulations — •Stefan Dasbach and Sven Wiesen — Forschungszentrum Jülich GmbH, Institut für Energie-und Klimaforschung - Plasmaphysik, 52425 Jülich, Germany

The effect of operational parameters on the exhaust in a tokamak can be adequately simulated by plasma edge codes like SOLPS-ITER. These simulations suffer from two main limitations: a) due to their complexity these transport codes suffer from long convergence times, and b) each simulation yields only a result for a single tokamak scenario with fixed parameters. However for rapid design studies of future fusion power plants systems codes require fast simplified models for the exhaust in many different scenarios (machine size, field, heating, etc.). Promising candidates for such surrogate models are machine learning models trained on simulation data. The development of such surrogate models is however limited by the high computational requirements for creating a sufficient training database. This work discusses several different pathways of how this limitation might be overcome and shows first steps in their implementation. The approaches shown include the generation of a simulation database based on less numerically demanding fluid neutral simulations and first trials on a surrogate model only replacing the numerically more demanding kinetic neutral part of SOLPS-ITER.