An efficient parallel ADI algorithm for solving 3-D convection diffusion equations with Neumann boundary conditions

Convection diffusion equations are widely used to model various important phenomena and processes in science and engineering. The calculation of numerical solutions for three-dimensional models is very computation intensive. The alternate direction implicit (ADI) algorithm is very efficient for this kind of equations and suitable for parallel computing. However, when Neumann boundary conditions are involved in the equations, it is difficult to maintain the original order of accuracy. We discuss a method to deal with Neumann boundary conditions when the ADI algorithm is used. The new method maintains the second order accuracy and is very scalable on multiprocessor parallel computers.