Assessing RANS-based turbulence models for mixed convection: predictions of flow stability and mixing length

This study provides a comparative evaluation of multiple turbulence models for predicting mixed convection within an axisymmetric exhaust column, where a hot jet rises through a cooler ambient region. Simulating mixed convection poses significant challenges due to the interplay of buoyancy forces, inertia-driven flow, and the complex mixing length region where natural and forced convection effects transition and interact. The simulations employ several RANS-based models—Realizable k–ε, k–ω SST, and the Reynolds Stress Model (RSM)—to analyze velocity profiles, temperature distribution, and turbulence characteristics. A structured mesh, refined near the walls, ensures accurate resolution of boundary-layer phenomena and thermal gradients. The Realizable k–ε model predicts significantly lower inlet and outlet mass flow rates (approximately ۱۱ and ۹ kg/s, respectively), along with a truncated mixing length that leads to reversed flow and undervalued outlet temperature (~۳۰۰ K). These results are inconsistent with physical expectations, highlighting the model's inability to capture the buoyancy and stability effects critical to mixed convection. In contrast, the RSM and k–ω SST models yield logical and physically consistent results, including sustained upward flow with outlet velocities around ۲–۲.۳ m/s and higher outlet temperatures (~۳۲۵ K), along with longer mixing lengths and minimal recirculation. These findings demonstrate that k–ω SST and RSM models are more reliable for predicting flow stability and thermal distribution in mixed convection scenarios, providing valuable insights for industrial applications.

Keywords:

Mixed convection , CFD , k–ε , k–ω , RSM