Time-Dependent DIII-D Heat Transport Simulations Using Neural-Network Models

The neural network transport model BRAINFUSE has been developed to produce transport fluxes based on local parameters [1]. The BRAIN-FUSE model has been integrated into the transport modeling framework ONETWO [2,3] in order to develop time dependent solutions and has been validated by artificially varying the input neutral beam power and comparing the output to DIII-D scans. These efforts have led to the development of a time-dependent workflow within the OMFIT integrated modeling framework. The new work flow can evolve the electron and ion temperatures as a function of time dependent sources and equilibria. The effects of different engineering parameters can be explored and optimized in support of DIII-D operations. The efficiency of this workflow enables planning plasma operations of next-day experiments, as will be required for ITER.\par \vskip6pt \noindent [1] O.~Meneghini et al., Phys.\ Plasmas {\bf 21}. 060702 (2014).\par \noindent [2] W.W.\ Pfeiffer et al., General Atomics Report GA-A16178 (2980).\par \noindent [3] O.~Meneghini et al., Bull.\ Am.\ Phys.\ Soc.\ {\bf 58}, 109 (2014).