A review of uncertainty quantification methods for Computational Fluid Dynamics

With the rapid development of Computational Fluid Dynamics (CFD) algorithms and software, CFD has been playing an increasingly important role in various fields involving fluid mechanics. However, unavoidable uncertainties in CFD simulations have great influence on the numerical results. Therefore, developing CFD uncertainty quantification methods which characterizes all major uncertainty sources during the simulation and quantify their effects on the quantities of interest can not only satisfy the demands of CFD credibility evaluation in engineering practices, but also support the precise design of aircrafts. The development and application of uncertainty quantification methods in CFD are summarized in this paper. Firstly, the premise of CFD uncertainty quantification, i.e., uncertainty sources are introduced. Generally, uncertainties can be categorized into stochastic and epistemic ones. Their corresponding uncertainty quantification methods have been proposed. Then the development and application of uncertainty quantification methods are surveyed in typical CFD processes including the geometry generation, meshing, turbulence modeling, and inflow conditions. In the end, sensitivity analysis methods are introduced, which are used to make out the influence of different uncertainty factors on numerical results. Summarizing the previous work on uncertainty, suggestions for the future development of uncertainty quantification methods in CFD are given.