Towards a Unified K-Epsilon Turbulence Model for the Practical Analysis of Aeropropulsive Flows

Abstract

While LES-based methods are starting to be used more routinely for analyzing complex aero-propulsive flows, their routine use for design/optimization and systems studies, where numerous solutions must be generated in a “production-line” manner, remains a future goal requiring more robust computer resources and advances in LES methodology. A Reynolds Averaged Navier-Stokes (RANS) approach currently provides the most viable option for performing fast-paced industrial/engineering studies, with a principal drawback being the accuracy and reliability of the turbulence models used, which often struggle to accurately capture the physics for more complex flow without resorting to highly specialized extensions. Progress made towards developing a unified RANS k-epsilon turbulence model, shown to accurately analyze a variety of high-speed aero-propulsive problems without problem specific model tuning or user intervention is described in this paper. This has entailed: including generalized corrections for compressibility and baroclinic torque; accounting for variations in turbulent Prandtl and Schmidt numbers; and employing an adaptive approach to locally implement non-linear modeling extensions. This paper will describe several simulations where turbulence model extensions have provided improvements in comparison to data, and will also discuss areas where improving the model’s performance requires further work. Improving the unified model will be guided to some extent by comparisons with data, but more so by comparisons with LES/DNS based solutions shown to adequately analyze problem sets for which the RANS model has had limited success. In a companion paper at this meeting, DES-based solutions being used to support upgrades to the RANS model, are described. The DES model uses this unified RANS model for wall bounded and interfacing regions and can thus deal with not only conventional turbulence model interfacing, but with the interfacing of scalar fluctuations.