Elevating zero-dimensional predictions of tokamak plasmas to self-consistent theory-based simulations

A new workflow in the OMFIT integrated modelling framework (STEP-0D) has been developed to make theory-based prediction of tokamak scenarios starting from zero-dimensional (0D) quantities. The workflow starts with the PRO-create (profiles creator) module, which generates physically plausible plasma profiles and a consistent equilibrium from the same 0D quantities as the H98Y2 confinement scaling. These results form the starting point for the STEP (Stability, Transport, Equilibrium, and Pedestal) module, which then iterates between state-of-the-art theory-based physics models for the equilibrium, sources, core transport, and pedestal to obtain a self-consistent solution. A systematic validation against the H98Y2 database and the global confinement database (DB5) demonstrated that on average STEP-0D is capable of predicting the energy confinement time with a mean relative error (MRE) <18%. A neural network-based energy confinement predictor, TaueNN (EPED-NN, TGLF-NN), was found to be faster but less accurate with an MRE of 23%. Lastly, STEP-0D was used to predict the performance of negative triangularity DIII-D experiments, and derive a theory-based confinement time scaling law for negative triangularity plasmas.