I have completed a duct analysis using the uniform mesh at there different mesh sizes for the same duct (0.01, 0.005 and 0.0035) as the process of mesh dependency analysis. Surprisingly all the results are very different from each other. I have used 2013 version for this analysis. Any comment on why the results are so different?
This would suggest that your mesh was possibly too coarse originally and that the refinement brought you closer to the real result.
Did you reach convergence each time?
Are you confident that the setup is correct? Incompressible flow etc?