Numerical Uncertainty Quantification for Radiation Analysis Tools

Abstract

Recently a new emphasis has been placed on engineering applications of space radiation analyses and thus a systematic effort of Verification, Validation and Uncertainty Quantification (VV&UQ) of the tools commonly used for radiation analysis for vehicle design and mission planning has begun. There are two sources of uncertainty in geometric discretization addressed in this paper that need to be quantified in order to understand the total uncertainty in estimating space radiation exposures. One source of uncertainty is in ray tracing, as the number of rays increase the associated uncertainty decreases, but the computational expense increases. Thus, a cost benefit analysis optimizing computational time versus uncertainty is needed and is addressed in this paper.The second source of uncertainty results from the interpolation over the dose vs. depth curves that is needed to determine the radiation exposure. The question, then, is what is the number of thicknesses that is needed to get an accurate result. So convergence testing is performed to quantify the uncertainty associated with interpolating over different shield thickness spatial grids.

Utilization of CAM, CAF, MAX, and FAX for space radiation analyses using HZETRN

“…In this paper, we focus on the distribution of target points within the effective dose tissues and use the 968 ray distribution used by Anderson et al (2007). The distribution is depicted in Fig.…”

“…These two gaps (limited use of available human phantoms and target point selection) create potentially large uncertainties that need to be addressed. Anderson et al (2007) have provided one of the few uncertainty estimates for a portion of this process by examining spatial discretization in the interpolation databases and angular discretization in the ray distributions. It was found that 512 rays were sufficient for the lunar lander and habitat studied, and that for solar particle events (SPE), at least 20 approximately log-distributed spatial grid points were necessary to provide sufficient accuracy in the interpolation database.…”