When we solved our simple problem we saw that the truncation error was O(x) We therefore expect that as the number of grid points is was O(x). We therefore expect that as the number of grid points isincreased (i.e. x decreases) the error should decrease and our solution get closer to the exact solution.Using N=8 and N=16 for example(and solving using MATLAB) we can see that the numerical solution get sincreasingly close to the exact solution.Thus CFD calcs should always be performed on grids of differing performed on grids of differing refinement to estimate "error" ineach case.