Acceleration of Computational Fluid Dynamics Simulations by Parallelization of the Linear Equation Solver Using Wavefront Technique

摘要

Computational fluid dynamics (CFD) uses numerical analysis methods and algorithms to solve and analyze problems that involve fluid flows. These algorithms require very high computational power that does not exist in traditional computers systems. This paper presents code accelerated methodology by parallelizing the linear equation solver and is developed to provide the numerical solution of the steady Navier-Stokes equations, using the Finite Volume Method (FVM) in primitive variables formulation. A case study of incompressible flow in a 2D lid-driven square cavity is investigated for $ext{Re}=1000$. Detailed second order spatially accurate results, where the tertiary vortices were observed, are presented. Co-located grid arrangements along with a uniform structured Cartesian grid up to ${701}^{st} 701$ were used. Strongly implicit procedure is used to solve the resulted linear algebraic equations. The parallelization of SIP solver is implemented by using wave-front technique and was implemented using OpenMP with multi-core processor. The solution time was reduced by 26%.