Thesis for the Degree of Master of Science


Year 2011
Octavio Pozo Garcia

Classification and analysis of faults and their diagnosis methods in the process industry

The current competitive environment found in process industry demands the optimization of the production processes. Untreated faults have serious consequences on a process or, in the worst cases, to the whole plant and the industry sector. In this context, the development of fault detection and fault diagnosis methods has become a high priority task. Traditionally, fault detection and diagnosis methods are classified according to their working methodology. However currently, there is no classification or methodology that relates the types of faults with the information needed for a fault detection and diagnosis system to perform their task efficiently.

The aim of this thesis is to establish the existence of a relationship between the features generated by a fault and the fault detection methods used to discover the presence of such fault. In literature part of this work, topics related to fault detection, fault diagnosis and their current classifications are presented. In the experimental part, the introduction of a classification of fault features, seeks to allow a better understanding of the detection methods in the context of the signals or symptoms they use. While the presentation of process devices and their most common faults illustrate the features that can be generated by a single component in a system. Finally, to fully illustrate the relationship between fault features and detection methods, a methodology for the selection of fault detection methods is presented and exemplified.

This info last modified 20 Jun 2018 by Jerri Kämpe-Hellenius