Improved finite element domain decomposition method using local and mixed Lagrange multipliers

This paper presents the two-dimensional structural computational algorithm based on the finite element domain decomposition technique. The proposed algorithm is compared with the well-known FETI approach with regard to convergence and efficiency characteristics. The proposed approach uses both local and global Lagrange multipliers together with an augmented Lagrangian formulation to achieve computational robustness and efficiency. The present algorithm is relatively simple and leads to improved convergence and efficiency characteristics. Numerical results are compared with those obtained by the original and dual-primal FETI methods to demonstrate the improved efficiency of the proposed approach. And implementation on a parallel computer is conducted for the proposed FETI methods. Finally, sparse matrix algorithm is implemented on the proposed FETI methods.