This paper deals with the solution of the geometrical nonlinear problem of thin-walled steel structures using parallel finite element computations. The case of interaction of local buckling and overall lateral torsional buckling is considered. The incremental approach of Finite Element Analysis (FEA) and the Message Passing Interface (MPI) library for the achievement of the results was used. First the level of performance for the sequential tasks on machines with computer architectures such as PCs, workstations or mainframes is compared. In particular, the distribution of CPU time for solution and assembly of the incremental stiffness matrix, solution of the system of equations and the rest of the computational tasks is compared. The parallel tasks which include the formation of the incremental stiffness matrix were run on a multiprocessor shared memory SUN workstation and on a network of SUN workstations using MPI. The number of processes versus CPU time is compared. The main feature of this paper is to show how this highly nonlinear problem is computationally intensive and how parallel processing may improve the efficiency of the solution.

purchase the full-text of this paper (price £20)

go to the previous paper
go to the next paper
return to the table of contents
return to the book description
purchase this book (price £82 +P&P)