Efficient prediction of turbulent flow quantities using a Bayesian hierarchical multifidelity model

High-fidelity scale-resolving simulations of turbulent flows quickly become prohibitively expensive, especially at high Reynolds numbers. As a remedy, we may use multifidelity models (MFM) to construct predictive for flow quantities interest (QoIs), with the purpose uncertainty quantification, data fusion and optimization. For numerical simulation turbulence, there is hierarchy methodologies ranked by accuracy cost, which include several numerical/modeling parameters that control robustness resulting outputs. Compatible these specifications, present hierarchical MFM strategy allows simultaneous calibration fidelity-specific in Bayesian framework as developed Goh et al. 2013. The model provide an improved prediction combining lower higher fidelity optimal way any number levels; even providing confidence intervals QoI. capabilities our are first demonstrated on illustrative toy problem, it then applied three realistic cases relevant engineering flows. latter friction different numbers channel flow, aerodynamic coefficients range angles attack standard airfoil, propagation sensitivity analysis separation bubble over periodic hills subject geometrical uncertainties. In all cases, based only few high-fidelity samples (typically direct simulations, DNS), leads accurate predictions QoIs accompanied estimate confidence. result UQ analyses also found be compared ground truth each case.