383376 Simulating Flow Field in Micro Channels Using Computational Fluid Dynamics

Wednesday, November 19, 2014: 2:00 PM
M304 (Marriott Marquis Atlanta)
José Carlos Gonçalves Peres and Ardson do S. Vianna Jr., Chemical Engineering, University of São Paulo, São Paulo, Brazil

Microreactors are promising equipments to perform reactions needing high heat and mass transfers rates, reactant conversion and dealing with explosive or synthesis of high value added products. Their simulation by means of computational fluid dynamics (CFD) is often time-consuming due to the complexity of the coupled fluid flow and reaction phenomena and the requirement of small element sizes. It is important to apply strategies that reduce the computational effort of simulations in order to achieve accurate predictions within reasonable time.

         This paper presents the simulation of the flow field in a microreactor with 76 channels in a serpentine arrangement using CFD. Initial characterization was conducted on a geometry consisting of 10 channels to reduce processing time. This computational domain was discretized by hexahedral and tetrahedral meshes, whose element edges were reduced until mesh independence was achieved. The geometric parameters of the meshes obtained in this part were then used to generate a mesh for the full reactor geometry. Fluid flow was estimated for flow rates between 12.5 and 1000 μL/min. The k-ε, RNG k-ε, k-ω, shear stress transport (SST), algebraic Reynolds stress (ASM) and Reynolds stress (RSM) turbulence models were also compared to laminar flow predictions.

         Solutions of the flow field in the simplified geometry (10 microchannels) did not change above 565,064 elements for the hexahedral meshes and 730,596 for the tetrahedral meshes. Besides requiring fewer elements, hexahedral meshes converged in almost half the time compared to tetrahedral ones and exhibited elements with smaller aspect ratios, which smooth the iterative process. The k-ω, SST, ASM and RSM models predicted coincident parabolic velocity profiles, with maximum values ca. 10-20% smaller than the laminar flow hypothesis depending on the position on the microchannel. The k-ε and RNG k-ε models showed a solution inconsistent with the other models, caused by problems with the e-equation in flows with low Reynolds number. Inlet flow rate affected the shape of the velocity profile in the curves of the microchannels: for high values, the profile was parabolic with maximum velocities were found in the center of the curves, while for small values they were shifted to ca. 30% of the curve radius.

Extended Abstract: File Uploaded
See more of this Session: Turbulent and Reactive Flows
See more of this Group/Topical: Engineering Sciences and Fundamentals