Scaling of ELM crash parameters

Introduction
Edge localized modes (ELMs) are periodically occurring instabilities that cause fast relaxations of the strong edge pressure gradient in the high-confinement regime (H-mode) of tokamak fusion plasmas [1]. These crashes induce intense heat fluxes towards the divertor tiles. This
is a major concern for future fusion devices like ITER [2]. To understand the nonlinear ELM dynamics, it is necessary to check the validity of the (potentially predictive) ELM models. This can be achieved by comparing modeling output to experimental results. One essential parameter for such a comparison is the structure of the ELM crash, i.e. the toroidal mode number n. Recent quantitative comparisons of n and other parameters of the ELM crash between the nonlinear code JOREK and results obtained on the ASDEX Upgrade tokamak (AUG) demonstrated the progress in understanding the ELM crash by nonlinear modeling [3]. Consequently, the next question that is tackled here is how the ELM characteristics change with peeling-ballooning critical parameters. We therefore introduce a database of 30 shots containing more than 2500 type-I ELM crashes on AUG and investigate how the structure size changes with plasma parameters, which enables a more detailed testing of the codes in the future.