A kriging-PDD surrogate model for low-cost sensitivity analysis

In this work, a metamodeling technique, which couples universal Kriging interpolation and Polynomial Dimensional Decomposition, is proposed. It can be usefully used in the context of Uncertainty Quantification, in particular for the construction of metamodels to accelerate forward and backward uncertainty propagation or robust optimization and to perform sensitivity analysis. The proposed method is firstly validated on a battery of test cases well-known in literature. It is then applied to the uncertainty propagation for the hypersonic high temperature atmospheric entry flow of the EXPERT vehicle. The aim in this case is to perform the sensitivity analysis of some quantities of interest, namely the pressure and the heat flux at the stagnation point.