This project addresses a central problem underlying the effort to automate the process of design software development: the lack of reliable techniques for solving large sets of nonlinear algebraic equations that form the basis of chemical process design computations. Such techniques, if available, would be an essential component in a system capable of generating the required numerical software rapidly and reliably by machine. Software development costs and delays could be reduced, yielding an increase in engineering productivity. The objectives of the research are to produce a numerical solution technique that will provide a rigorous guarantee of successful convergence; the capability to identify and compute multiple solutions; the ability to deal with feasibility constraints defined by inequalities; and the precise diagnosis of underspecification, overspecification, and infeasibility of complex system models. The new approach is based on an implicit enumeration strategy employing analytically-derived bounding functions. The research will encompass development of the detailed algorithm with analysis of its theoretical properties, followed by software implementation and computational testing on full-sized chemical processing system models.
