Fast Uncertainty Quantification in Engine Nacelle Inlet Design Using a Reduced Dimensional Polynomial Chaos Approach

Combining A Reduced Polynomial Chaos Expansion Approach with Universal Kriging for Uncertainty Quantification

“…We used the same computational configuration for the engine nacelle inlet case as that in our previous work. 1 This allows a direct comparison between the results for the RPC and RPC-K methods. For convenience, the schematic of the engine nacelle geometry is shown in Fig.…”

“…The boundary conditions can be referred to our previous study. 1 The design of experiments includes seven parameters, n = 7, including the freestream Ma ∞ , AOA, the corrected mass flow rate (ṁ), the ratio of the turbulent eddy viscosity to the molecular viscosity (µ t /µ), the diameter of the nacelle inlet, the length of the nacelle, and the air density. We are interested in their influence on the flow separation along the nacelle lip and inlet distortion.…”

“…We have developed a dimension reduction based PC approach. 1 Although such an expansion contains a small subset of the overall PC coefficients, it is comparable in accuracy to the full-dimensional expansion since only the least impactful dimensions are removed.…”

“…Thereafter, a new and sparse PC expansion can be efficiently generated from the reduced dimensional random subspace. The RPC method has been tested by Gao et al 1 and will be used in conjunction with the universal Kriging method in the present study.…”

“…Consider that there are N sets of samples for the design parameter space, ω = ω (1) , ..., ω (N ) with corresponding CFD output values U = U (1) , ..., U (N ) . Following Schöbi et al 4 and Dubourg, 10 based on the generalized least-squares solution to the maximum likelihood function, the metamodel parameters β, σ 2 can be obtained by…”