Comprehensive Uncertainty Estimation for CFD

Computational fluid dynamics (CFD) is the primary tool of aerodynamic design, and the ability to reduce and quantify uncertainties in CFD results is essential. Efforts throughout the CFD community are aimed at improving the accuracy of CFD methods. The proposed work will develop computational algorithms and software that yield comprehensive estimates of uncertainty in CFD results. The methods involve error modeling, in which specific sources of error, such as truncation error, and error in the transition and turbulence modeling, are related to the errors and uncertainty in the solution via solutions of an error equation. These methods will be relatively comprehensive, accounting for the dominant sources of uncertainty in typical CFD analyses, and yielding the uncertainties in all data predicted by the CFD solution. For example, the uncertainty in distributed aerodynamic heating can be computed. The techniques to be used in the work have been demonstrated in predictions of uncertainty due to truncation error in structured-grid CFD data. The proposed work will extend these methods to unstructured-grid CFD and to transition and turbulence modeling errors.