differencing schemes for 3-D Convection-diffusion problems

Bob Roach

But you still didn't answer Andy's question as to the effective Reynolds number. If it is low, you may be able to get away with a scheme that is more central. If you can stand it, a fine grid will reduce the cell Reynolds number to the point where central difference would be the preferred stencil. Apart from that, John Chien's advice about the varying coefficients being included in the differencing is extremely important if you want to keep from introducing non-conservation type errors which definitely delay convergence. Good lucks.