Computational fluid dynamic of single and two phase flow in microchannels

Abstract

A numerical study of single and two-phase fluid flow and heat transfer in single microchannels was conducted using the computational software package ANSYS Fluent. The geometry construction and mesh generation were performed using the hexa meshing option in the ICEM mesh-generator. In the single-phase flow simulations, two different configurations were studied: (1) a single channel with a hydraulic diameter of 0.561 mm and (2) a multichannel with inlet and outlet manifolds, comprising 25 channels with a hydraulic diameter of 0.409 mm. In the single channel simulations, the effect of axial conduction and geometrical parameters, i.e. aspect ratio and hydraulic diameter, on the fluid flow and heat transfer were investigated. Four numerical models, using 2D thin-wall, 3D thin-wall (heated from the bottom), 3D thin-wall (three-side heated) and 3D fully conjugated geometries were employed in order to study the effect of axial conditions on heat transfer rate. The simulations were conducted for Reynolds numbers ranging from Re = 100 to Re=3000 and water was used as the working fluid. The setup of the 3D thin-wall model simulation used thermal boundary conditions that were very similar to those assumed in the accompanying experiments (uniform heat flux), resulting in the numerical friction factor and heat transfer results showing an excellent agreement with the existing experimental data. In contrast. On the contrary, the results of the 3D fully conjugated model demonstrated that there is a significant deviation (more than 50%) when compared to the 3D thin-wall and the experimental data. The results of the 3D fully conjugated model indicated that there is a significant conjugate effect and the heat flux is not uniformly distributed along the channel. The effect of aspect ratio and hydraulic diameter on fluid flow and heat transfer in a single microchannel was investigated by using the 3D thin-wall approach to avoid conjugate heat transfer effects. Two sets of simulations were conducted, where in the first set of simulation, the effect of hydraulic diameter was studied by varying the channel width and depth while keeping the aspect ratio constant. The range of hydraulic diameters was 0.1–1 mm and the aspect ratio was fixed at 1. In the second set of simulations, the aspect ratio ranged from 0.39 to 10, while the hydraulic diameter was kept constant at 0.56 mm. The simulations were conducted for a range of Reynolds numbers, Re = 100–2000, and water was used as the working fluid. The friction factor was found to decrease slightly with aspect ratio up to AR ≈ 2, after which it increased with increasing aspect ratio. The results demonstrated that the slope of the velocity profile at the channel wall changes significantly with aspect ratio for AR > 2. The effect of the aspect ratio and hydraulic diameter on the dimensionless hydrodynamic entry length is not significant. Also, the aspect ratio does not affect the heat transfer coefficient, while the dimensionless Nusselt number as well as the friction factor were found to increases with increasing hydraulic diameter. In the multichannel configuration, the flow distribution inside the twenty-five parallel channels, including the effect of conjugate heat transfer, was studied. The numerical study was carried out using R134a as a working fluid. The obtained friction factor is in agreement with experimental data provided by Fayyadh et al. ( 2015), while the value is under-predicted compared to Shah and London (1978) in the laminar region and Philips (1987) in the turbulent region. For the heat transfer results, the numerical Nusselt number results differed significantly from the experimental results by Fayyadh et al.( 2015) as well as the correlation provided by Shah and London (1978). In addition, based on the average velocity and temperature of the fluid inside the twenty-five channels, the flow was uniformly distributed only at the channel located in the centre region. This could be due to the occurrence of flow recirculation because of the the sharp edges in the inlet manifold leading to a reduction of flow rate in the channels located at the lateral edges of the parallel microchannels. Therefore, the under-prediction of numerical heat transfer results could be attributed to the conjugate effects and flow mal-distribution. Subsequently, the effect of various parameters such as, inlet/outlet area manifold, inlet/outlet flow arrangement and number of channels (N=9, 13, 17, 25, 33) on the flow distribution inside channels was examined. The dimensionless channel flow ratio and flow maldistribution factor were introduced to quantify the flow distribution inside individual channels and the uniformity of this flow distribution. A uniform flow distribution is achieved when the MF value approaches 0. In the present study, the area of the inlet and outlet manifolds were varied by varying the length of manifolds while the width of the manifold remained constant. The inlet and outlet manifold areas were showing a significant effect on flow distribution, where longer inlet and outlet manifolds were observed to have a better flow distribution. In addition, three different types of flow arrangements, namely, U-type, I-type and Z-type flow arrangements with two different inlet and outlet manifold areas were used. Based on the flow distribution results, the Z-type flow arrangement was showing the worst flow uniformity compared to the other flow arrangement types, irrespective of the inlet and outlet manifold area used. However, in the study of the effect of the number of channels on the flow distribution, two different phenomena were observed as the number of channel increases. Firstly, if the number of channels is less than twenty, the maldistribution value was found to increase as the number of channels increases. In contrast, if the number of channels exceeds twenty, the maldistribution value reduces with increasing number of channels. Finally, a newly design for the inlet manifold was proposed in this study, where edges with a curved shape were suggested in order to reduce the occurrence of flow recirculation at the sharp edges. This resulted in a better flow distribution over the parallel channels. In the two-phase study, a 1 mm circular microchannel with a vertical orientation was used. The setup of these numerical simulations, using R134a vapour-liquid as a working fluid, had two purposes. The first aim was to develop a numerical flow regime map in order to identify the slug flow boundaries and the dependence on the annular nozzle configuration. In these simulations two dimensional, axi-symmetry was assumed in order to save computational time and effort. The second aim was to study the topology of the hydrodynamic flow and heat transfer distribution of slug flow using three-dimensional flow simulations and related these to the operational condition, i.e. the gas superficial velocity. Based on the numerical flow pattern maps developed in this study, four basic flow patterns, including bubbly flow, slug flow, slug-annular, annular flow, were observed. The present results were verified through comparison with the visualization results reported in the experiments of Chen et al. (2006) and were shown to agree well. The study also showed that Computational Fluid Dynamic can be used to obtain a reliable two-phase flow pattern map. v The slug formation in the R134a vapour-liquid, using the annular (concentric) nozzles configuration of Shao, et al. (2008), was studied. A three dimensional computational domain of 1 mm vertical circular channels was used. A mechanism that consist of expanding, contracting and necking processes was found to occur at the lower end of the bubble. Close to the nozzle, as similar pattern of behaviour as shown by Shao et al. (2008) was observed. In addition, slug cells with a dry zone and irregular shapes were obtained in the present 3-D numerical work. Previously, most researchers assumed the flow to remain axisymmetric in order to save computational time. This assumption lead to identical bubble slugs. However, the process of bubble generation using the axisymmetric assumption is not realistic and a fully three dimensional simulation is required. Asadolahi et al. (2011). In Gupta et al. (2010) concluded that dry-out is one of the problems in numerical work in and it can be avoided by properly constructing numerical grids and algorithms However, Talimi et al. (2012) suggested that the appearance of a dry-out condition, in numerical work is not because of a poor mesh resolution, which could leads to nonphysical results. To settle this dispute, further study of this phenomenon is needed. Therefore, several individual cells with different behaviours and shapes were carefully selected in order to understand the characteristics of the hydrodynamic and thermal behaviour. In the final part of the two-phase flow study, the effect of superficial gas velocity (with constant superficial liquid velocity) on two phase slug flow was studied. Three different superficial gas velocities were employed in this study. As the mixture velocity increases, the void fraction, β, the capillary number, Ca, and the Reynolds number, Re, are also increasing. In addition, changes in bubble shape, bubble length, liquid film thickness, velocity and temperature profile, pressure drop and heat transfer at different operations were discussed.