Parallel transient dynamic non-linear analysis of reinforced concrete plates

This article describes how parallel processing may improve the computationa l efficiency of the transient dynamic non-linear analysis of reinforced con crete plates subjected to blast or seismic loading. A parallel scheme for t he time marching procedure is presented using the explicit Newmark's algori thm. The finite element formulation with the material modelling (of a strai n-rate sensitive hardening-softening, elasto-viscoplastic model accounting for cracking and crushing) is described. The complexity of the material law s and their modelling, induced by the extreme loading conditions, requires vast computational effort per time step. It will be shown that very high co mputational efficiency may be obtained by decomposing the finite element me sh into a number of sub-domains for distributed analysis on multiple proces sors. This high efficiency is achieved using distributed memory MIMD archit ectures (including transputer networks), where the routing of the messages represents significant overhead in terms of inter-processor communication b ecause of limitations resulting from the processor network topology. Throug h examples it will be demonstrated how this efficiency depends on the probl em size (i.e. the level of refinement of the problem idealisation), the num ber of sub-domains and the status of the analysis with regards to the state s of the material. (C) 1999 Elsevier Science Ltd and Civil-Comp Ltd. All ri ghts reserved.