A wavefront and bandwidth reduction algorithm
The finite element method of analysis involves the solution of large systems of linear algebraic equations with sparse coefficient matrices. In order to solve such a system of equations efficiently, computer storage as well as calculation time must be conserved.
A recent approach to utilizing this sparseness of the coefficient matrix is a frontal solution procedure involving an element assembly and factorization of the system equations. The efficiency of the frontal solution is dependent on the maximum element wavefront of the system which itself is determined by the order in which the elements are addressed in the assembly process.
This study outlines a procedure for resequencing the element data directly in order to reduce the maximum element wavefront encountered. Of specific concern is the order in which decisions are made in choosing a succeeding element for renumbering. Several examples to which the algorithm has been applied are presented. These examples show that the suggested minimum frontal growth algorithm is also efficient in reducing the bandwidth associated with a banded solution of the sparse matrix.
Thesis77.F843.pdf
2.9 MB
Adobe PDF
a4b5fc9981ba9145e11c6b002d3e39d0