Automating HAZOP studies using D-higraphs

Abstract

In this paper, we present the use of D-higraphs to perform HAZOP studies. D-higraphs is a formalism that includes in a single model the functional as well as the structural (ontological) components of any given system. A tool to perform a semi-automatic guided HAZOP study on a process plant is presented. The diagnostic system uses an expert system to predict the behavior modeled using D-higraphs. This work is applied to the study of an industrial case and its results are compared with other similar approaches proposed in previous studies. The analysis shows that the proposed methodology fits its purpose enabling causal reasoning that explains causes and consequences derived from deviations, it also fills some of the gaps and drawbacks existing in previous reported HAZOP assistant tools.