Speaker
Description
As the number of objects in LEO increases, Design for Demise (D4D) becomes an ever-more important philosophy for space engineering which relies upon robust modelling of re-entry processes. Uncertainty Quantification (UQ) is a vital part of ensuring this robustness but is limited by the significant computational costs associated with high-fidelity re-entry simulation and as such most UQ approaches only consider low-fidelity object-oriented codes.Recent developments in information fusion for Monte Carlo estimators enable multifidelity-based methods for UQ which have significant potential for D4D applications. Estimator algorithms have been developed which can leverage the correlation between high and low fidelity models to enhance the convergence rate of the estimator through variance reduction. In this work the multifidelity capabilities of the TransatmospherIc flighT simulAtioN tool (TITAN) are used as models within the estimator framework developed by Schaden and Ullman, named the Multi-Level Best Linear Unbiased Estimator (MLBLUE), in order to perform UQ on a re-entry test case. The open-source MLBLUE implementation and extension developed by Croci, Willcox and Wright, BLUEST, enables the automatic selection of models from the multiple fidelity options provided by TITAN and allocation of samples in order to minimise error for a given computational budget via Semi-Definite Programming. This enables computationally feasible estimation of the mean values of landing location whilst still incorporating high fidelity information. This results in an optimal convergence rate with increasing budget. The test case provides initial conditions derived from a Two/Three Line Element file which have inherent uncertainty in translational information. This uncertainty is propagated by the MLBLUE estimator into a landing distribution. This distribution can be used to better quantify the performance of deterministic simulation processes by using statistical distance metrics in addition to the broader contexts of application in risk analysis in the definitions of declared and safe re-entry areas.
