| Summary: | 碩士 === 國立成功大學 === 機械工程學系 === 102 === In this study, numerical calculations by single-phase and two-phase models of nanofluid turbulent forced convection in a three-dimensional wavy channel with uniform wall temperature are investigated. The elliptical, coupled, steady-state, three-dimensional governing partial differential equations for turbulent forced convection of nanofluids are solved numerically using the finite volume approach. The governing equations are solved with the standard turbulent model. The parameters studied include Reynolds number, nanoparticle volume concentration, wavy channel amplitude and wavy length. The numerical results are validated with the turbulent nanofluids in a smooth channel in the literature first, the maximum discrepancy within 2%, and then further extend to a wavy channel.
Two different types of wavy channels are considered and their average Nusselt number for a constant wall temperature is compared. The predicted average Nusselt number of the symmetric wavy channel shows better than that in-line wavy channel. The numerical results of the proposed models indicate the flow field and turbulent convective heat transfer characteristics have some differences for single and two-phase models. The thermal field predictions by the three two-phase models are essentially the same except the Eulerian model but very far from the numerical results of single-phase model. In the range of parameters in the study, the average Nusselt number of the wavy channel considered is found to improve with increase in Reynolds number, the wave amplitude, and the wavelength. In addition, after the comparisons of the numerical results with single and two-phase models, the multi-parameter constrained optimization procedure integrating the design of experiments (DOE), response surface methodology (RSM), genetic algorithm (GA) and computational fluid dynamics (CFD) is proposed to design the nanofluid turbulent convection of three-dimensional wavy channel. Three non-dimensional variables, namely wavy amplitude, wavy length, and volume concentration are chosen as design variables. The objective function E which is defined as thermal performance factor has developed a correlation function with three design parameters. The thermal performance factors predicted by regression function for in-line and symmetric channel cases are in good agreement with the numerical results of CFD by the difference within 6.3% and 3.3%, respectively. The combination of parameters is considered as the optimum solution.
|
|---|
