Description
Injection of electron cyclotron (EC) waves can initiate plasma breakdown in tokamaks even prior to the application of a loop voltage, offering a non-inductive start-up path for future reactor-scale devices. Extending EC breakdown to future devices requires a predictive framework capable of quantifying operational requirements such as minimum EC beam power, prefill pressure, and connection length in advance. Here we report the first experimental comparison of such a framework against dedicated KSTAR breakdown experiments. The framework employs BREAK, a 3D particle-in-cell code for tokamak plasma breakdown, into which localized EC wave--particle interaction and resultant energy-gain models have been incorporated in preceding work. The EC energy exchange is treated as a fast microscopic process, separated from the slower macroscopic evolution of the discharge, enabling quantitative prediction of the breakdown boundary. With fixed EC launch conditions on KSTAR, prefill pressure and connection length are scanned independently at two vertical-field settings by varying the prefill gas puff and the vertical magnetic field. The observed breakdown boundary is found to be consistent with simulation predictions for both settings, within the uncertainty of the inferred in-vessel neutral pressure. Applying the same framework to an ITER-like magnetic geometry, successful avalanches are predicted near and below 1 MW at a prefill pressure of approximately 2 mPa, indicating that EC breakdown remains accessible at modest beam power.