How to improve the model partitioning in a DSS for instrumentation design

Abstract

In this article, we present an extension to the observability analysis algorithm known as Direct Method (DM), which is an essential piece of a Decision Support System (DSS), whose purpose is to help engineers in the instrumentation design of industrial process plants. The main goal of this new version of DM algorithm is to improve the global performance of the DSS. The new variant, known as Extended Direct Method (EDM), incorporates a novel module that quantifies the degree of nonlinearity associated with the variables and equations of the mathematical model for the process plant. This new quantification procedure conforms a heuristic search, which guides the traditional DM towards a decomposition of the mathematical model in subsystems with the smallest possible degree of nonlinearity. The obtained results show that the EDM is able to reduce the nonlinearity of the subsystems obtained through the observability analysis procedure. At the same time, it can successfully increase the generated quantity of minimum size subsystems, thus also contributing to improve the DSS?s global performance.