Brandon Michael Medina

Verification and benchmark problems allow code developers to assess numerical accuracy and increase confidence that specific sets of model physics were implemented correctly in the code for application to real world problems. In this work, we compare a benchmark test of a relativistic (500keV) electron beam propagating through a pressurized (0.1 mbar) Ar gas cell and present the results between the general plasma code EMPIRE and the hybrid code GAZEL. EMPIRE can be run as a fully kinetic Particle-In-Cell (PIC) problem with Direct Simulation Monte Carlo (DSMC) collisions or as a hybrid problem with both fluid and PIC charged species which collide with a background neutral fluid via fluid-fluid rate-based interactions and PIC-fluid Monte Carlo Collisions (MCC) that produce a charged fluid of low-energy secondary plasma. GAZEL is a hybrid code that utilizes deterministic collisions between the fluid and beam particles via a pseudo-fluid, which is generated from spatial averages of the beam computational particles. We compare both fully kinetic EMPIRE results with hybrid EMPIRE and hybrid GAZEL results and find reasonably good agreement until late times when presently numerical instabilities result in the beam breaking up in the EMPIRE simulations.

Laser Doppler velocimetry, also known as photonic Doppler velocimetry (PDV), is a diagnostic commonly used in shock compression experiments. To quantitatively compare experimental results to computational results, the uncertainty in the velocimetry measurements must be well characterized. That is, we must know which features in the experimental data are statistically significant and which are not. In this contribution, we will present recent work on understanding the resolution, precision, and accuracy of PDV measurements. We will discuss how the uncertainty is a function of signal strength, noise, and signal digitization, and how the uncertainty is influenced by the analysis method. We will show how knowledge of uncertainty can be used to rigorously compare experiment with computational predictions.