Optimization of industrial problems using parallel processing under distributed environments

A parallel optimization algorithm implemented in a distributed computing environment was applied to nonlinear engineering problems. We deal with the Parallel Variable Distribution (PVD) algorithm, discussing how to handle nonlinear constraints and proposing a new domain-partitioning heuristics. The quality of the proposal was first assessed by analyzing its performance for several small nonlinear models associated with classical engineering problems. Then, the parallel distributed code was employed to solve the rigorous model of an existing expander plant, whose constraint-evaluation stage was more complex. Satisfactory speed-up and efficiency values were achieved.