A PARALLEL BLOCK FRONTAL SOLVER FOR LARGE-SCALE PROCESS SIMULATION - REORDERING EFFECTS

For the simulation and optimization of large-scale chemical processes, the overall computing time is often dominated by the time needed to s olve a large sparse system of linear equations. We describe here a par allel frontal solver which can significantly reduce the wallclock time required to solve these linear equation systems using parallel/vector supercomputers. The algorithm exploits both multiprocessing and vecto r processing by using a multifrontal-type approach in which frontal el imination is used for the partial factorization of each front. The alg orithm is based on a bordered block-diagonal matrix form and thus its performance depends on the extent to which this form can be obtained. Results on several large scale process simulation and optimization pro blems are presented, with emphasis on the effect of different matrix r eorderings to achieve bordered block-diagonal form.