Understanding Performance Variability in Mega-Joule Class Inertial Confinement Fusion Implosions

Recent high-yield experiments at the NIF have have made performance variability a key question for the ICF community; How reliably can we produce MJ-class yields? What drives the observed variability, and how can we reduce it?

To answer these questions, multiple repeats of the MJ-class hybrid-E platform have been performed which provide explicit sampling of performance variability. We are interpreting this series of experiments using a new statistical model to learn a `global’ probability distribution that describes shot-to-shot variability. Our method uses ensembles of multiphysics simulations to link observed variations to interpretable physics degradations, which can then be linked with detailed analysis of experimental data. We use our results to explore the importance of specific degradations, and to validate our conclusions by linking to high fidelity post-shot simulations and independent analyses of experimental data. In this talk, we will demonstrate how our statistical model can be used to explore performance sensitivities in hybrid-E implosions and in new implosion designs.