Adjoint-based surrogate modelling of Spalart-Allmaras turbulence model using gradient enhanced kriging

Bagheri, Amir K. and Da Ronch, Andrea (2020) Adjoint-based surrogate modelling of Spalart-Allmaras turbulence model using gradient enhanced kriging. In AIAA AVIATION 2020 FORUM. Aerospace Research Central.. (doi:10.2514/6.2020-2991).

Record type: Conference or Workshop Item (Paper)

Abstract

A Gradient Enhanced Kriging algorithm has been developed to find a surrogate model of a Reynolds-Averaged Navier-Stokes solver in low speed separated flow. The surrogate relies on gradients of the modelling quantities with respect to the design parameters. The gradients are obtained using a discrete adjoint solver and are used in the surrogate model to make more accurate predictions of the full order system. Adaptive sampling accelerates the generation of the surrogate model and ensures full system dynamics are captured in regions of high nonlinearity. It is proposed that, once the model has fully developed, it can be used to fine-tune the coefficients of the Spalart-Allmaras turbulence model to minimize the error between computational results and results from higher fidelity solvers or experiments, and therefore obtain parameters which are calibrated for solving separated flows.

This record has no associated files available for download.