Uncertainty Quantification in Crater Formation for Gas-Granular Flows due to Plume Surface Interaction

摘要

The liberation of dust and debris particles caused by rocket plume flow from spacecraft landing on the unprepared regolith of the Moon, Mars, and other extra-terrestrial destinations poses a high risk for robotic and human exploration activities. The regolith particle flow induced by a spacecraft landing on the unprepared surfaces of the Moon, Mars, and other celestial bodies occurs in a combination of “extreme environments” that combine low gravity, little or no atmosphere, rocket exhaust gas flow that is supersonic and partially rarefied, and the unusual mechanical properties of regolith. Of these environmental factors, characterizing the regolith granular material fluidic behavior and gas-granular interactions is the most complex and least developed. In this work, CFD Research implemented two methodologies to quantify underlying uncertainties in physics-based models into the highly scalable Eulerian-Eulerian gas-granular flow solver Loci/GGFS. Intrusive and non-intrusive methodologies for uncertainty quantification allow for the entire range of input sensitivities to be passed through the physics-based submodels, giving the analyst a unique and powerful capability for assessing their impacts on spacecraft design. The intrusive Forward Automatic Differentiation (FAD) was implemented into Loci/GGFS. FAD directly computes sensitivity derivatives through operator overloading and the chain-rule of each function and model in the CFD solver. FAD is used to perform sensitivity analysis in only as many simulations as input sensitivities. Two noninstrusive methodologies, Morris-One-at-a-Time (MOAT) and Sobol Indicies, were utilized through useage of a UQ framework, the Sensitivity Quantification of Uncertainty Analysis Toolkit (SQUAT). FAD was verified against results from SQUAT and off-design simulations of Loci/GGFS for an important analog gas-granular case and a key PSI validation case, showing the power of the method.