Non-intrusive semi-analytical uncertainty quantification using Bayesian quadrature with application to CFD simulations

To improve the safety, reliability, and performance of complex engineering systems, it is crucial to understand quantify uncertainties. This paper presents a framework non-intrusively semi-analytically parametric uncertainty within CFD simulations using Bayesian quadrature (BQ). An in-house quantification (UQ) code based upon this mathematical developed. The then validated by applying due varying parameter in simple analytical test function. mean variance obtained BQ are compared with those from solution stochastic simulation Latin hypercube sampling (LHS) method. validation case shows that outperforms LHS approach terms computational efficiency accuracy. UQ utilised characterise (due unknown inlet flow profile) predicted operating parameters an industrial scale butterfly valve, as well uncertainties high-wavenumber damping factor Cs) SAS-SST simulated bluff-body flow. It found entry profile non-ignorable effects on valve parameters. Meanwhile, changes opening. For flow, large variances properties exist region where separate shear layer dominates because varyingCs. Moreover, effect Cs more significant turbulence quantities, acts generation turbulent eddies directly.