SBIR/STTR Award attributes
All NASA modeling and simulation activities are mandated to provide uncertainty characterization and quantification of the underlying physics submodels and their propagation towards the simulation output metrics. Recently developed simulation tools to predict Plume-Surface Interaction (PSI) effects such as dust lofting, obscuration, debris transport, and surface cratering lack practical uncertainty assessment capability. The multi-physics, multi-phase, gas-granular media interaction modeling relies on complex algorithms and numerous physics submodels derived from experimental datasets with limited fidelity and considerable uncertainties. Model and algorithmic complexity and the frequent immaturity and sparsity of the fundamental physics submodels elevates the urgency of sensitivity analysis capability for PSI simulations. This project proposes development of an efficient Forward Automatic Differentiation (FAD) based sensitivity derivatives in conjunction with non-intrusive UQ methodologies for gas-granular flow solver Loci/GGFS. The FAD will enable run-time sensitivity analysis and propagation of underlying sub-model uncertainties through the overall PSI simulation model towards uncertainty quantification of simulation output metrics. The approach is efficient, especially for large parameter spaces and requires a limited number of simulations compared to sampling methods. Sensitivities analysis allows identification of dominant sub-model contributors of uncertainty, guide improvements, and provide a rapid propagation of critical uncertainties to the simulation output metrics.nbsp; The resulting tools will be delivered to NASA for ready application for Lunar and Martian landers, including the Human Lander System, to aid in quantifying and identifying uncertainties and deficiencies in current simulations.
