Applying parallel block predictor-corrector methods to a Space Shuttle main rocket engine simulation

The effectiveness of applying two types of parallel block predictor-corrector algorithms presented by L.F. Shampine and H.A. Watts (Math. Comput., vol.23, p.731-40, 1969; BIT, vol.12, p.252-66, 1972) and L.G. Birta and O. Abdou-Rabia (IEEE Trans. on Comput. vol.C-36, no.3, p.299-311 March 1987) to the simulation of a Space Shuttle main rocket engine is discussed. Comparisons between the sequential and parallel versions of the algorithms are made, based upon implementations of Shuttle main engine simulations on a four-node multiple-instruction multiple-data parallel processing environment consisting of Inmos T800 transputers. The results of these simulations are reported in terms of certain performance measurements, including execution time, processor utilization, simulation accuracy and simulation stability. The expected performance of expanded versions of the algorithms which employ more than four transputers is also calculated assuming a hypercube-type topology.<
>