You are in:Home/Publications/A fast multi-pole expansion applied to bending of plates integral equations

Ass. Lect. Mohamed Elsayed Mohamed Elsayed Nassar :: Publications:

A fast multi-pole expansion applied to bending of plates integral equations
Authors: Mohamed Elsayed Nassar
Year: 2013
Keywords: fast multipole
Journal: Not Available
Volume: Not Available
Issue: Not Available
Pages: Not Available
Publisher: Not Available
Local/International: International
Paper Link: Not Available
Full paper Mohamed Elsayed Mohamed Elsayed Nassar_FMMthesis9.pdf
Supplementary materials Not Available

A thesis submitted in partial fulfillment of the requirements for the degree of Master of Science in Engineering Mathematics Faculty of Engineering -Shobra Benha University This Thesis presents a new fast multi-pole boundary element formulation for the solution of Reissner’s plate bending problems. The solution of Reissner’s plate bending problems using the conventional direct boundary element method leads to a non-symmetric fully populated system of matrices. The complexity of the solution then becomes of the order O(N3) mathematical operations, where N is the total number of problem unknowns. Hence, the use of fast multi-pole technique becomes practically essential in case of solving large-scale problems by the direct boundary element method. The suggested formulation is based on representing the fundamental solutions as function of potentials. These potentials and their relevant fundamental solutions are expanded by means of Taylor series expansions. In the present formulation, equivalent collocations are based on the first shift expansion of kernels. This is achieved by representation of far field integrations by series expansions and carrying out summations of far clusters, whereas the near field integrations are kept to be computed directly. In the presented implementation, the fast multi-pole boundary element method is coupled with the iterative solver: Generalized Minimal Residual System (GMRES). The computational complexity is rapidly reduced to be O(N log N). Numerical examples are presented to demonstrate the efficiency, time saving, and accuracy of the formulation against the conventional direct BEM. The accuracy of the results is traced by cutting Taylor series to few terms. It was proven via numerical examples that three terms are enough to produce sufficient accuracy with substantial reduction of solution time.

Google ScholarAcdemia.eduResearch GateLinkedinFacebookTwitterGoogle PlusYoutubeWordpressInstagramMendeleyZoteroEvernoteORCIDScopus