it is an index to show how the mesh quality influence the calculation results. You can use the different dimension mesh to calculate, then analyse the differences. The best mesh is that it has the same calculation resultion as the smaller mesh and these mesh bigger than its have worse resultion than it.
Dear Prof. Xiangyang and Walid Adel, since I have a quite similar question, so may I know is it possible to achieve the best mesh which provide the same calculation result as the smaller mesh? Or maybe there is some tolerance or acceptable range of results? Some say that 10% tolerance is acceptable, and some say that 5% is acceptable. But I didn't found any references on this matter? Any comment?
Dear Fadzli, I think it's very difficult to find a general acceptable tolerance. It must based on your mesh and the required accuracy, so you must make many runs with different number of elements until you reach to the equilibrium with the minimum number of elements.