UTILITY OF A FINITE-ELEMENT SOLUTION ALGORITHM FOR INITIAL-VALUE PROBLEMS

The Galerkin criterion within a finite element Weighted Residuals formulation is employed to establish an implicit solution algorithm for an initial-value partial differential equation. Numerical solutions of a transient parabolic and a hyperbolic equation, obtained using linear, quadratic and two cubic finite element basis functions, are employed to quantize accuracy and confirm and refine theoretical convergence rate estimates. The linear basis algorithm for the hyperbolic equation displays excellent accuracy on a coarse computational grid and a high-order convergence rate with discretization refinement. Good accuracy and a strong convergence rate in surface flux are determined for a non-homogeneous Neumann boundary constraint applied to a parabolic equation. The results amply demonstrate the impact of the non-diagonal finite element initial-value matrix structure on solution accuracy and/or convergence rate. © 1979.