Modelling Gap Closure Within Pulsed Power Machines

Abstract

Gap closure is a process where expanding electrode plasmas close the anode-cathode gaps of the vacuum transmission lines used in TW pulsed-power machines, affecting power flow, and ultimately limiting operating performance [1]. Simulating the formation of plasma is therefore a critical modeling need for characterizing high-current diodes in machines such as the Saturn accelerator, and to assess the power delivered in Sandia's next-generation pulsed power (NGPP) machine; however, our ability to simulate this process is limited due to the enormous computational cost. Full PIC-DSMC simulations of gap closure are limited by huge variability in length and time scales necessary to resolve the physics. In order to preserve the solution stability, while accurately capturing the electron velocity distribution function and collisions, the time step needs to be less than a picosecond with grid spacing on the order of nanometers. Running a full simulation through the ~100 ns pulse thus requires tens of millions of time steps. One way to speed up the solution while keeping time steps and spatial grids unaffected is to use event splitting to allow for fewer macroparticles necessary for better statistics. This is done by taking the classic variable weight technique a step further by splitting a particle into multiple particles upon collision, with individual weights determined by the probability of each possible collision outcome. Event splitting enhances the ability of the DSMC technique to model low-probability processes, such as for the relatively rare ionization collisions that drive the formation of the plasma [2]. We will discuss the superiority of event splitting over standard 1D DSMC in terms of computational costs and level of stochastic noise [3]; this will be shown by comparing multiple simulations of gap closure run by both standard DSMC and the event splitting technique.