| Summary: | 博士 === 國立清華大學 === 動力機械工程學系 === 88 === The simulations of turbulent transport phenomena in ribbed channels with RANS (Reynolds-Averaging Navier-Stokes Equation) and LES (Large Eddy Simulation) were undertaken in this study. The enhanced heat transfer can be achieved through the use of obstructed or ribbed channels. The local turbulent fluid flow, heat transfer and thermal performance in a channel with various kinds of rib shape, rib permeability, and rib detachment have been investigated. The increase of thermal performance and alleviation of high temperature spots (hot spots) are the main concerns in the selection of these three parameters.
For RANS, a curvilinear nonothogonal body-fitted coordinate system was adopted. A two-layer k-εmodel, a KL k-εmodel and a RSM(Reynolds stress model) with wall function and wall-related pressure-strain treatment were used. For LES, the finite-volume technique was used to solve the time-dependent filtered compressible Navier-Stokes equations with a dynamic subgrid-scale turbulence model. Parallel processing was used to make three-dimensional time-dependent LES more feasible.
It''s desirable to examine the suitability of using two dimensional RANS with two-layer treatment and turbulence model modified to reduce turbulent kinetic energy at stagnation region for the predictions of turbulent fluid flow and heat transfer in a ribbed channel. The two-layer k-εmodel attains improvement over the k-εmodel with wall function in the prediction of flow around a single attached rib and multiple attached ribs of square cross-section. As to the simulation of detached rib flow, turbulence models with modifications to reduce turbulent kinetic energy at stagnation region (KL k-εmodel for single detached rib and RSM with wall-related pressure-strain model of Craft and Launder (CL) for multiple detached ribs) are essential to the prediction of vortex shedding flow.
In examing the effect of rib shape, calculations were conducted for channels with single or multiple rib pairs of four shapes of rib cross-section, which are square, triangular, semielliptic and semicircular. The lowest friction factor is obtained for the semicircular rib case for both single and multiple attached ribs. Triangular-shaped rib is favorable for a compromise between the thermal performance and possible presence of hot spots.
As to the effect of rib permeability, two kinds of permeable ribs were simulated in the present study, i.e., porous and perforated ribs. For the former case, a porous model of solving flow field inside and around the porous rib simultaneously was incorporated with k-εturbulence model to investigate the flow passing partly through and partly around a porous rib. The pore diameter and porosity are the main parameters investigated. The numerical results were in parallel trend with the experimental results performed in relevant geometrical conditions. The pressure drop and heat transfer enhancement decrease with increasing pore characteristic length and porosity. Among the parameters investigated, the thermal performance is the best when porosity is equal to 0.95 and pore diameter is equal to 4mm. For perforated ribs, a two dimensional slit configuration of the same rib open-area ratio (β) was used to simulate the spatially periodic turbulent fluid flows and heat transfer in a channel with perforated rectangular ribs mounted on one wall. To facilitate the simulation and as a first step towards a complex analysis, the perforated ribs with three dimensional discrete circular holes are replaced with the slit ribs with two dimensional slots. The main parameter investigated was the rib open-area ratio with values of 0%, 10%, 22%, 38% and 44%. Two critical ranges of β and three characteristic flow regimes are identified. The results also identify the dominant fluid dynamic factors responsible for the reported peak values of local Nusselt number distribution.
RANS and LES was used to study the complex shedding phenomena existing in the turbulent wake flow behind a detached rib with different distances from the wall. With the merit of resolving case dependent large eddy motion and modelling universal small eddy motion, the application of the 3D LES with dynamic SGS model to the simulation of detached rib flow is desirable and anticipated to provide a better prediction. The effect of gap-to-rib height ratio (C/H) on the turbulent flow field and heat transfer was investigated. The physical mechanism underlying the suppression of vortex shedding behind a square cylinder located at a distance smaller than the critical gap from a wall is revealed. The interaction between the cylinder and the wall is illustrated by the phase-averaged spanwise vorticity components and the vortex celerity of spanwise vortices. The mechanisms responsible for the augmentation of heat transfer through the spanwise and longitudinal vortices behind a square cylinder placed near a wall are documented. The time-averaged near-wall transverse velocity distribution is found to have the most profound influence on the nearby-wall Nusselt number distribution. The overall time-averaged Nusselt number along -1<X/H<9 is the highest for C/H=0.75. This is due to the weaker upwash flow away from the wall and stronger convective flow along the wall.
|
|---|
