Uncertainty Quantification by Adaptive Multifidelity Surrogates of Noisy CFD Data

The objective of the present work is to present an adaptive RBF-based N-fidelity (NF) surrogate for uncertainty quantification of complex industrial problems, fully exploiting the potential of simulation methods that naturally produce results spanning a range of fidelity levels: RANS (Reynolds-Averaged Navier-Stokes) simulations with adaptive grid refinement, and/or multi-grid resolution. The NF method is further advanced to reduce the effects of the noise in the CFD outputs through regression and in-the-loop optimization of the model.