Session: 16-01-01: Government Agency Student Poster Competition

Paper Number: 150167

150167 - Evaluation of Turbulence Models in Simulating Rayleigh-Bénard Convection for Cloud Formation 

Cloud formation occurs through several forms, with Rayleigh-Bénard convection being one of the prominent processes. Rayleigh-Bénard convection involves warming a fluid from the bottom while cooling it from the top, resulting in a temperature gradient that may cause buoyancy-driven stability (laminar flow) or instability (turbulent flow). Turbulence is of particular concern in cloud formation since clouds are naturally turbulent and evolve over time. 

However, turbulence modeling in Computational Fluid Dynamics (CFD), although important, is complicated by the chaotic nature of turbulent flows and their sensitivity to initial and boundary conditions. As such, different models have been developed to address various turbulence scenarios, each with its strengths and limitations. Therefore, it is important to examine different turbulence models for Rayleigh-Bénard convection and cloud formation as it helps identify the most appropriate model for accurately simulating this process. 

This study analyzed three different turbulent models: Reynolds-averaged Navier-Stokes (RANS) k-epsilon, RANS k-omega, and the Large Eddy Simulation (LES) model. Since the LES model is intrinsically unsteady, transient analysis was conducted for each of the models mentioned above to ensure an appropriate comparison. The analysis was performed on a 3D cylindrical geometry of 3*pi m³ volume in ANSYS Fluent R22 commercial software. The governing equations were the conservation of mass,  momentum, and energy, each with their respective turbulent terms. The top and bottom walls had isothermal boundary conditions such that the temperature difference established was 10 K. The Boussinesq approximation was also used to ensure simplicity in dealing with the density term in the governing equations. Temperature, velocity, vorticity, turbulent kinetic energy, and turbulent dissipation rates were analyzed to compare the models. Quantitative and qualitative comparisons were conducted using volume-averaged results and contour plots, respectively. 

The LES model displayed detailed turbulent structures and small-scale vortices, showing a higher-resolution view of the flow. In contrast, the RANS k-epsilon and k-omega models showed smoother and more averaged flow structures. The RANS k-omega model, in particular, excelled in resolving near-wall boundary layers, whereas the RANS k-epsilon model showed broad and averaged vorticity patterns. 

Although the LES model offered a much more detailed representation of flow characteristics, it required significantly more time and computational resources compared to the RANS models, making it computationally expensive. The RANS models, despite being less detailed, provided a good general overview of the convection current within the enclosure while using fewer computational resources. Thus, for scenarios where a general understanding of the flow is sufficient, the RANS k-epsilon and k-omega models are adequate. However, for cases involving the inclusion of aerosols, the detailed resolution provided by the LES model may be more appropriate. 

The results from this study highlight the strengths and limitations of each turbulence model, enabling better predictions of cloud dynamics and informing the development of more reliable atmospheric models. These insights contribute to improved climate research and offer valuable guidance for selecting the appropriate turbulence model based on specific research needs.

Presenting Author: Ivana Barley Southern University and A&M College

Presenting Author Biography: Ivana Barley is a graduate student at Southern University and A&M College pursuing her Master’s in Mechanical Engineering. Her fields of interest are computational fluid dynamics, thermal and fluids sciences, and computer-aided design. She aspires to be a leading mechanical engineer in the near future.

Authors:

Ivana Barley Southern University and A&M College
Stephen Akwaboa Southern University and A&M College
Patrick Mensah Southern University and A&M College