A Hybrid RANS-LES Computational Fluid Dynamics Simulation of an FDA Medical device benchmark

Abstract

Computational fluid dynamics (CFD) is a valuable tool that complements experimental data in the development of medical devices. The reliability of CFD still presents an issue and for that reason, no standardized approaches are currently available. The United States Food and Drug Administration (FDA) has initiated the development of a program for CFD validation, and has presented an idealized nozzle benchmark model. In this study, a nozzle flow with sudden expansion has been simulated using advanced RANS-LES turbulence models. Such models partially resolve the flow and are cheaper in computer resources and time in comparison to the Large Eddy Simulation (LES). Furthermore, they are more accurate than standard Reynolds-averaged Navier-Stokes (RANS) models.

A collection of hybrid turbulence models has been investigated: Detached Eddy Simulation (DES), Stress Blended Eddy Simulation (SBES), and Scale Adaptive Simulation (SAS), and compared to a standard RANS Shear Stress Transport (SST) model. Subsequently, all models were validated by experimental results already published by different research groups. Particle Image Velociometry (PIV) experiments were performed by inter-laboratory study, and the results are available online for numerical validation.  The flow conditions in this study are only restricted to a turbulence flow at a Reynolds number of Re =6500. Complementing the turbulence models investigation, two advection schemes were tested: high resolution (HR) and bounded central difference scheme (BCD). Among all advanced models the SBES model with BCD scheme has the best agreement with the experimental values.