Transient Analysis of Inelastic Structural Large Deformation Applied with Parallel Computation

• Main Authors: Wei-Yu Hsu, 徐維佑
• Other Authors: Yaw-Jeng Chiou
• Format: Others
• Language: en_US
• Published: 2001
• Online Access: http://ndltd.ncl.edu.tw/handle/44761845724524291301

碩士 === 國立成功大學 === 土木工程學系 === 89 === The objective of this research is to investigate the implementation of parallel computation for simulations of nonlinear behavior of steel structures subjected to dynamic loads. A parallel explicit finite element approach is employed to facilitate inelastic structural analysis more efficient. A convected material frame (C.M.F.) is adopted to deal with the large geometric behavior of structures. This approach is a modification of the co-rotational approximation by incorporating an adaptive convected material frame in the basic definition of the displacement vector and strain tensor. In the formulation, each discrete element is associated with a local coordinate system that rotates and translates with the element. For each load increment, the corresponding strain-displacement and nodal force-stress relationships are defined in the updated local coordinates, and based on the updated element geometry. The rigid body motion and deformation displacements are decoupled for each increment. This modified approach incorporates the geometrical non-linearities through the continuous updating of the material frame geometry. A two surface model is used to describe the material non-linearity, which has been proven to be capable of predicting the inelastic cyclic behavior of steel structures with good agreement, as is compared with the experimental results by many former researchers. The model accounts for the importance cyclic characteristics of structural steel, even with the yield plateau, such as, the decrease and disappearance of the yield plateau, reduction of the elastic range and cyclic strain hardening. A generalized nonlinear function is used to derive the inelastic constitutive relation, and the Bauschinger effect and the cyclic strain hardening is treated in this model. The equation of motion is integrated by an explicit procedure and it involves only vector assemblage and vector storage in the analysis by assuming a lumped mass matrix of diagonal form. The computations are primarily vector operations, which are well suited for parallel processing in the proposed approach. All the computation is carried out on the PC clusters in Linux and conventional parallel machine IBM SP2 at the National Center for High-Performance Computing (NCHC). The message passing software package, MPI, is utilized as a parallel construct for data communication and message passing among processors. Numerical studies have been made to investigate the speedup and efficiency of the parallel code. The performance studies indicate that this explicit algorithm is highly adaptive for parallel processing.