Worst Case Analysis of a Launcher Vehicle Using Surrogate Models

Abstract Polynomial surrogate models of an industrial standard, closed-loop launcher model during an atmospheric flight phase are developed. A probabilistic collocation method is applied to derive surrogate models of varying complexity for wind perturbations occurring at different time instances during the atmospheric flight phase. These models are employed in optimisation-based worst-case analysis to significantly reduce the computational overhead while maintaining the accuracy of the computed worst-case uncertain parameter combinations. Comparisons of worst-case analysis results obtained using standard simulation models and their polynomial surrogate equivalents demonstrate significant potential of the proposed approach for radically reducing the computational burden associated with the industrial verification and validation process.