Solution algorithms for nonlinear transient heat conduction analysis employing element-by-element iterative strategies (Q1067505)

An important result from the element-by-element (EBE) algorithm proposed by \textit{T. J. R. Hughes}, \textit{I. Levit} and \textit{J. Winget} in J. Eng. Mech. Div. ASCE 109(2), 576-585 (1983) is that the storage and solution costs tend to grow linearly with problem size, whereas the resources required by traditional finite element algorithms tend to grow geometrically as the problem size is increased. This paper is a development of the EBE solution techniques on the following aspects: the first is the development of techniques for solving linear equation systems arising from the application of the finite element method to problems in nonlinear transient heat conduction. The conclusion is that the one-pass, reordered, Crout EBE factorization applied to the optimal definition of the matrix of preconditioned systems and coupled with a precondition conjugate gradient iteration algorithm is particularly effective for solving symmetric positive-definite matrix systems. Convergence studies have verified the analysis in the approximate factorization. The use of step-size selection strategies to control the error in the transient solution is considered. Also, the modifications required to implement EBE algorithms on parallel-processor machines are proposed.