Application of hybrid Computational Fluid Dynamics turbulence model, STRUCT-[epsilon], on heated flow cases

Abstract

Computational Fluid Dynamics (CFD) modeling is a powerful numerical method that can be used to characterize fluid flow, pressure drop, and thermal transient behavior in complex flow geometries. However, with current CFD simulation techniques, the accurate modeling of turbulence structures is often prohibitively expensive or time-intensive. Therefore, a new hybrid turbulence model, STRUCT-[epsilon], was developed to more accurately and quickly resolve the formation and propagation of unsteady turbulence structures. STRUCT-[epsilon], introduces a source term that implicitly reduces fluid eddy viscosity, which in turn reduces Reynolds stresses which are traditionally over-predicted with two-equation models. Most notably, STRUCT-[epsilon]s method of implicit hybrid activation is uniquely simple to implement while remaining grid-size and inlet turbulence condition independent. STRUCT-[epsilon], has previously demonstrated improved accuracy in the prediction of flow topology and velocity compared to results produced by URANS models for several internal and external flow cases. This study seeks to extend understanding of STRUCT-[epsilon]s capability by benchmarking the model's performance with the experimental or Direct Numerical Simulation (DNS) results from a variety of flow cases, including an impinging jet [3], film cooling [4], and an infinite wire-wrapped nuclear fuel assembly [5]. For each case, computed parameters from each CFD simulation were evaluated and compared both numerically and qualitatively, through the computation of root mean square error and identification of characteristic flow features. The improved performance of the RANS turbulence model could have large implications on the practicality and applicability of CFD modeling in the design and qualification of numerous technologies.