in my work, i figured out that when i change the mesh size, and than when i submit the job, it gives totally different results under same circumstances, so why this happening to me, and how can i solve this problem ?
You should perform mesh study. Run the analysis with different mesh setup (corse to fine). Observe your interest result for every mesh. Choose the mesh that give minimum error between the analysis.
Let say, mesh 2 and mesh 1 shows 10% error. Mesh 2 and mesh 3 result shows 3% error. Mesh 4 and mesh 3% shows 2% and so on. When you plot your interest result against the mesh setting, you will see which mesh is good for your analysis. In other words, the result show insignificant change regardless of mesh setup.