A least-square radial point collocation method for adaptive analysis in linear elasticity

This paper presents a least-square radial point collocation method (LS-RPCM) that is formulated based on the strong formulation and the local approximation using radial basis functions (RBFs). Aiming to solve the instability problem observed in the conventional RPCM using local nodes, a simple and yet effective procedure that uses the well-known least-square technique in a carefully designed manner has been proposed to restore the stability. Since stable solution can now be obtained, the LS-RPCM is then extended for adaptive analysis. Attractive features of the meshfree strong-form method that facilitate the implementation of adaptive analysis are demonstrated via a number of examples in this work. A robust residual based error estimator and a simple refinement procedure using Delaunay diagram are adopted in our adaptive scheme. Stable and accurate results are obtained in all the numerical examples.