Dear Rayudu Nithin Manohar, FEM is not working effectively when problems with infinite domain and high gradient of parameters are solved. For example, in structural analysis, there is a contact problem for interaction of rectangular die and elastic half-space when high concentration of contact pressures take place. FEM analysis of local stress-strain state of micro-heterogeneous composite material is a difficult task too. In these cases we must choose analytical or semi-analytical (boundary element method first of all) solutions which are more effective than FEM.
I have an interesting example of a harmonic Lamb wave propagating in an infinite isotropic plate. The analytical solution for the velocity dependence on frequency is well known, it is called dispersion relation. However, my attemts to use FEM for launching Lamb wave at sufficiently large fequencies, failed due to necessity to decrease element size with increasing frequency.