Theses and Dissertations

ORCID

https://orcid.org/0000-0002-6734-5455

Issuing Body

Mississippi State University

Advisor

Bhushan, Shanti

Committee Member

Li, Like

Committee Member

Cho, Heejin

Committee Member

Knizley, Alta

Date of Degree

12-13-2024

Original embargo terms

Worldwide

Document Type

Dissertation - Open Access

Major

Engineering (Mechanical Engineering)

Degree Name

Doctor of Philosophy (Ph.D.)

College

James Worth Bagley College of Engineering

Department

Michael W. Hall School of Mechanical Engineering

Abstract

Fast breeder nuclear reactors use liquid metals such as Sodium (Na), Sodium-Potassium (Na-K), and Lead (Pb) as coolants since these liquids have high thermal conductivity, high thermal diffusivity, and lower heat capacity compared to water and air, thus involving low Prandtl numbers (Pr). However, liquid metals solidify at room temperature which poses challenges for experimental studies, making computational fluid dynamics (CFD) is considered a valuable analysis and design tool. Furthermore, the accurate modeling of turbulent heat transfer in low Pr flows remains one of the main challenges due to the gap between momentum and turbulent thermal diffusion. The study aims to address this challenge by enhancing the understanding of liquid metal coolant behavior and improving the accuracy of turbulence models in these types of fluids under different convective conditions. This research consists of two main parts, where the first one encompasses generating a DNS dataset for Reτ = 640, Pr = 0.004, 0.025, and 0.71, and Gr = 0 and 17.4×106 to supplement existing DNS databases, providing a more comprehensive foundation for turbulence models validation, and the second one envelop assessing the predictive capabilities of linear eddy viscosity-based Reynolds average Navier Stokes (RANS), Partially-average Navier stokes (PANS), and large eddy simulation (LES). The assessment covers four test cases ranging from canonical turbulent flow to more complex flow regimes involving separating and reattaching flows under different convective conditions for Reynolds numbers (Re) ranging from 640 to 40,341 and Pr varying from 0.004 up to 0.71. DNS results improve the understanding of Re, Pr, and buoyancy effects on both mean turbulent flows. Analysis reveals that buoyancy enhances heat transfer more significantly for lower Re and Pr. Furthermore, buoyancy alters flow and thermal structures by enhancing and reducing turbulence on both aiding and opposing sides, affecting heat transport. In addition, the assessment of different turbulence models demonstrates the superiority of LES compared to other models where the average of the prediction errors is 6% over all the cases aligning with findings from the excessive literature review.

Download

Share

COinS