The method of successively refining the grid to obtain results on a fine grid with much less CPU time than would otherwise be required for the same problem has been studied. It is found that by using this method, the CPU time savings are as high as a factor of 50 for certain problems but it is disadvantageous for others.