I am simulating incompressible flow through small orifice by my own code. I met a very hard problem. When I simulates a larger orifice flow(beta=0.5), the code is very easy to converge. However,if the orifice becomes smaller, it will become harder and harder to converge. I doubt this is because of sudden geometry change. Does anybody have experience in this aspect? Thanks

Hi, I am not a CFD expert user or programmer but I have good experience in the type of difficulty you met. In the partial fulfillment required for master degree I am making a comparison study between ALGOR FEA and PHOENICS for a simple laminar flow between two plates. At the beginning I found that the results were fluctuating and relatively far from reality and experimental results although the problem I was handling is considered as analytically solvable. My supervisor asked me to change the grid specifications; I found the results changing extremely within the grid dimensions. Due to this I ran a computer experiment by changing the grid specification and then plot the results against the grid dimension. The resulting curve gave me a good indication about the best grid dimensions setting. When I used this setting excellent results were obtained. I think you have to check what is the most suitable grid specification for your case and may be you will get good results. For you as a code developer I think you will benefit very much if you read about the theory of numerical grid generation. Another useful idea is to try comparing your code results with a well-established CFD package like fluent, CFX, PHOENICS or ALGOR FEA. The comparison of the results may lead you to a new approach that you might not see before. Very glad if this modest contribution could help.