Uncertainty quantification of transport-coefficient datasets for magnetohydrodynamic simulations of pulsed power experiments

Plasma transport coefficients are necessary closures for magnetohydrodynamic codes used to simulate inertial confinement fusion experiments. These coefficients are typically tabulated into a dataset that spans a wide temperature and density range. During the tabulation process, a limited number of data points of varying fidelity are used to inform the dataset creation. However, the ability to include the uncertainty of these data during the dataset generation—which has been shown to be significant—has not been systematically explored. Moreover, the sensitivity of magnetohydrodynamic codes to these datasets has not been quantified. We present a framework for generating transport-coefficient datasets for uncertainty quantification studies with magnetohydrodynamic codes. Using Bayesian inference, the framework incorporates uncertainties in the data used to generate the datasets; through an optimal sampling approach, the framework generates a minimal number of datasets to perform uncertainty quantification with computationally expensive codes. We illustrate the utility of the framework by carrying out magnetohydrodynamic simulations of pulsed power experiments—although the framework is generic and application independent. We discover that a modest uncertainty in the data used to construct the datasets results in a significant uncertainty when compared to the resolution of experimental diagnostics. The framework highlights the region within the datasets where magnetohydrodynamic simulations are particularly sensitive; this prioritizes the temperature and density regimes where additional data will be maximally impactful. We discuss how the framework can be utilized in large-scale integrated magnetohydrodynamic simulations of inertial confinement fusion experiments underway on the Z Machine of Sandia National Laboratories. SNL is managed and operated by NTESS under DOE NNSA contract DE-NA0003525. SAND2024-08109A.