The Impacts of Model Uncertainty on RANS CFD Simulations of a High-Speed Craft

Abstract

Computational scientists frequently publish papers discussing various sources of uncertainty in numerical methods for computational fluid dynamics. The frequently discussed sources of uncertainty are round off error, discretization error, iterative error, and mathematical model uncertainty (i.e. uncertainty in turbulence modeling). While all of these sources of uncertainty are real and impact the results of a simulation, the authors have found through experience that the most critical element to achieving accurate simulation results for high-speed craft is the generation of a high quality mesh on which the numerical methods are solved. There are, in general, two requirements for a high quality mesh; refinement regions must be applied in regions where the physics of the flow are most significant, and the three dimensional model of the hull form must be appropriately defined and fair. The latter appears to be a major contributing factor to uncertainty in CFD simulations that is not often discussed in the literature. Further, in practice there are numerous sources of geometric uncertainty between a prescribed CAD geometry and physically constructed model. In this paper, simulations are performed on two models of GPPH, with and without an edge radius on the transom to evaluate the impact of including fine geometric details (like tooling radii) in a RANS CFD model. The results of the simulations show that the inclusion of the rounded edge leads to large simulation errors in resistance and running attitude. This work has concluded that inclusion of fine geometric details in a planing hull CFD model is not beneficial to the overall accuracy of the simulations relative to necessary design accuracy.