Simulation of channel flow by parallel implementation of thermal Lattice-Boltzmann method on GPUs

Lattice-Boltzmann method (LBM) is a new generation numerical technique used to solve fluid flow and heat transfer problems. It has certain advantages over the classical solvers based on Navier-Stokes equations. LBM is well suited to simulate single and multi-phase flows of single and multi-component fluids around complex geometries, and is suitable for parallel programming. One of the state-of-the-art technologies to perform parallel computing is the use of Graphics Processing Units (GPU). In this study, heat transfer analysis of single-phase laminar flow of a viscous fluid in a slit-channel with a constant wall temperature is presented using thermal Lattice–Boltzmann method. Both the flow field and the temperature field are determined by solving the associated Lattice–Boltzmann equations. Computing is performed on GPUs in parallel using JACKET which is an add-on package for MATLAB to handle GPU computing. Performance comparison with CPU-based and GPU–based computing is presented. NOMENCLATURE cs speed of sound f distribution function feq equilibrium distribution function g thermal distribution function geq thermal equilibrium distribution function w weight function ω collision frequency Ω collision operator τ relaxation time τt thermal relaxation time ξ lattice velocit