LAMINAR FLOW THROUGH A DOUBLE ELBOW GEOMETRY WITH LATTICE BOLTZMANN NUMERICAL METHOD N. M. Akmal, C. S. N. Azwadi and M. S. Zuhairi Faculty of Mechanical Engineering, Universiti Teknologi Malaysia, 81200 Skudai, Johor akmaln@uthm.edu.my azwadi@fkm.utm.my Abstract This paper presents a lattice Boltzmann method for the simulation of laminar flow through double elbow geometry. The lattice Boltzmann method was built up on the D2Q9 model and the single relaxation time called the lattice-BGK method. The dependence of reattachment length on the Reynolds number is determined. Results show that the flow was found to be strongly dependent on Reynolds number and two- dimensional behavior for Reynolds numbers below approximately 400. Comparable results were obtained between the present approach and those from a Navier-Stokes solver. Keywords: Double elbow geometry, laminar fluid flow, lattice Boltzmann method 1. Introduction The lattice Boltzmann model (LBM) has increasingly been accepted as a viable alternative approach to the well-known finite difference, finite element, and finite volume techniques for solving the Navier-Stokes equations [1, 2, 3]. Unlike any other well-known computational methods, LB scheme treats the fluid behavior at the microscopic level and brings together its information to predict the macroscopic behavior of fluid flow. This provides the opportunity to go deeper into the particles community and understand how the interaction between them would affect the macroscopic parameters of fluid flow. Another important improvement to enhance the computational efficiency is the implementation of the Bhatnagar Gross Krook collision operator (BGK) approximation (single relaxation time approximation) for the collision function [4]. Even under the simplifications provided using BGK, LBM has demonstrated its ability to simulate flows in porous media [1], immiscible fluids [5] and magneto-hydrodynamics [6]. Historically LBM was derived from the lattice gas automata (LGA) method [7]. Consequently, the LBM inherits some features from its precursor, the LGA method. The dynamics of distribution function evolving on a lattice space consists of two main steps; collision, particle at the same site collide according to a set of hard sphere particle collisions rules; and streaming, particle move to the nearest node in the direction of its velocities. However, instead of using Boolean representation of particle in LGA, LBM uses real numbers represent the local ensemble-averaged particle distribution function, and only kinetic equations for the distribution function are solved. The number of discrete velocities determines the lattice structure of LBM models. In other words, the discretization of physical space is coupled with the discretization of momentum space. As a result, computational in LBM is only restricted with uniform lattice structure and second order accuracy in space and time [8]. In this study, we used LBM to investigate the velocity profile and flow characteristics of a fluid flowing through double elbow geometry. This geometry is commonly found in piping systems, involving high accelerations and decelerations when the fluid is flowing at high speeds. The reattachment length after elbows is studied to determine the location of the fully developed profile after flow separation. Pressure gauges, flow meters, anemometers and other auxiliary devices are usually installed outside of the separated region to avoid the flow instabilities that could commonly occur. Pressure loss, low flow rate, 1 Proceedings of the International Graduate on Engineering and Science (IGCES'08) 23 - 24 December © 2008 IGCES Mechanical Engineering