Verification, validation, and uncertainty in computational fluids dynamics

In this paper, we discuss issues in verification, validation, and uncertainty prediction in computational fluid dynamics (CFD). Mesh adaptation is used to perform the systematic grid refinement studies required for code and simulation verification. We demonstrate the method of manufactured solutions (MMS) for code verification. Examples of successful and failed validation show that reliable prediction can only occur with an appropriate mathematical model of physics along with accurate numerical solution of the differential equations. The paper emphasizes good CFD practices to achieve systematic verification and put validation studies on solid grounds. Finally, the sensitivity equation method (SEM) is used to cascade input data uncertainty through the CFD code for estimating the uncertainty of the flow response. Key words: adaptive finite elements, verification, validation, uncertainty, Navier-Stokes. On discute des enjeux de verification, validation et de prediction d'incertitudes en CFD (<< computational fluid dynamics >>). On utilise des maillages adaptatifs lors des etudes systematiques de raffinement de maillages requises pour la verification de code et de simulation. On presente la methodes des solutions manufacturees pour la verification de code. Des exemples de verification reussie et echouee demontrent que des predictions fiables s'obtiennent uniquement avec un modele mathematique approprie et des solutions numeriques precises des equations differentielles. L'article met l'accent sur les bonnes pratiques en CFD pour assurer une verification systematique et la rigueur de la validation. Finalement, on utilise la methode de l'equation des sensibilites pour obtenir des estimes des incertitudes des predictions induites par celles des inputs. Mots-cles: elements finis adaptatifs, verification, validation, incertitude, Navier-Stokes. [Traduit par la Redaction]
1. Introduction The accurate and reliable prediction of turbulent flows has been the subject of much research by the computational fluid dynamics (CFD) community for many years. A review of [...]