Uncertainty Quantification of Velocimetry-Based Load Current Inferences for 100 ns Multi-Mega-Amp Pulsed Power Experiments

Accurate determination of the electric current delivered to the load region is critical to the success of various multi-mega-amp pulsed power experiments. Magnetic inductance (B-dot) probes often fail over ten mega-amps, and when successful can only infer current centimeters from the load region, before major current loss mechanisms kick in. This has prompted interest in velocimetry-based load current inferences. The present work is part of a systematic uncertainty quantification effort. Here we apply Bayesian statistics to a series of synthetic experiments. Tens of thousands of multiphysics simulations were run with perturbed drive currents to determine correlations between input current and output surface velocity (our experimental measurable). We find that for very short pulses (100 ns rise time) relevant to cylindrical platforms significant uncertainties, not typical of planar dynamic materials experiments, can be introduced around peak current if the velocimetry measurement is not well matched to the current pulse. All regions of enhanced uncertainty are explained physically, and compensating strategies are developed in order to increase confidence and certainty in load current inferences.