New Methods of Finite Element Stiffness Matrix's Formation,Assemblage and Storage Suitable for Preconditioned Conjugate Gradient Method in Hydraulic Structures

包劲青,杨强,刘耀儒
DOI: https://doi.org/10.3969/j.issn.1002-5634.2009.04.001
2009-01-01
Abstract:Based on the analysis of algorithmic characteristics of preconditioned conjugate gradient methods(PCG),new methods of stiffness matrix formation,assemblage and storage were proposed for finite element analysis in hydraulic structures,which the stiffness matrix was of great scale and change frequently.In new methods,stiffness matrix was formed by nodes instead of by elements in traditional methods.Strategy of position remembering of stiffness matrix was proposed,where only non-zero coefficients and their positions in the stiffness matrix are stored,because zero coefficients were not involved in calculation during iterations in PCG.Examples show that the new methods are more suitable for PCG than traditional methods,both on the aspect of time consumption for stiffness matrix formation,assemblage,and the storage scale.
What problem does this paper attempt to address?