A systematic fault root causes tracing method for process systems

Abstract

Fault diagnosis and abnormal event management are key issues in process industry. For decades of research, many approaches have been proposed and put into practice. The Signed Digraph (SDG) model based diagnosis approach seems to be more effective in tracing the root causes of abnormality in large scale process system like petrochemical industry, especially when system becomes more complex nowadays [1]. Bu t SDG model based diagnosis approach has two inherent deficiencies. Firstly, the SDG models developed by current me thod often hold false causalities which will cause spurious diagnosis results. Secondly, the diagnosis result does not indicate the mechanical or electrical root fault causes directly. The root mechanical or electrical fault causes are more helpful for maintenance than root abnormal state variables given by SDG results. According to these deficiencies of SDG, a systematic root fault causes tracing method is proposed. First of all, causal ordering analysis based on the many physicochemical equations and empirical formulas in process systems is introduced to construct accurate SDG models. Then, the hierarchical dependent relation between state variable of SDG and its root failure causes is described by polychromatic sets. The dependent relation is further transformed into Fault Trees (FTs) which are used to deduce the root faults causes. In this method, the SDG model and the FTs are integrated into one systematic diagnosis model. The top level of this model is used to represent state variables of different equipments in a process system, and the bottom level are used to deduce the root fault causes of the equipment abnormality. This method integrates the advantages of SDG and FTA. Two examples are illustrated to show the modeling and diagnosis process, and thereby verify the practicality and validity of the proposed approach in process system fault tracing.