Two algorithms for superconvergent stress recovery based on MLS and finite points method

Two methods of stress recovery have been suggested and investigated in this paper. The first one is introduced by using moving least square method (MLS) OF and superconvergent points. The second method is based on the satisfaction of equilibrium equations at some nodes for which the recovery is applied. Simultaneous solution of these equations increases computational time. So, the second method is more expensive than the first one. A numerical example is used to compare the stresses recovered by these two methods with corresponding FEM, well-known SPR method and analytical solutions. The effect of various orders of basis functions and the values of weight functions on stress recovery by the proposed methods is also investigated. The present research indicates that the two methods, and especially the first one, represent acceptable accuracy over the domain and even on the boundaries, in comparison with SPR method, and also good convergency is achieved.