Just guesswork on my part, but are you sure that you had the same settings in everything that could affect the ouptut? Is it possible that some results could have been with results averaged at each node and some not averaged, for example? Are the meshes exactly the same, as best as you can tell? I know that fillets in SolidWorks can come up with different results depending on the order in which you pick edges. Is there something comparable that can happen in Simulation?
The only things I changed between runs were the variable values. I didn't reorder features, and I checked the mesh settings to make sure they were consistent.
Can we see some pictures of the parts, fea setup and results?