Advancement in monitoring and abnormal situation management in chemical processes

Resumen

In the last decade the interest on monitoring systems has increased following the demand of better management of the plants according to more restrictive economical and environmental conditions. As a consequence, the number of accidents has been reduced and the economical performance of process plants has been improved. Despite a promising start towards hybrid fault diagnosis systems (i.e., systems that integrate more than one monitoring methodology) the continuous increase in the complexity of the chemical processes generates the necessity of faster and more robust monitoring systems. In addition, an important direction for research is the integration of diagnostic tasks with other process operations such as sensor location, regulatory control, data reconciliation and supervisory control. Besides, in batch and semi-continuous processes, a fault diagnosis system should be integrated with higher levels of plant information such as the planning and scheduling activities. The objective of this thesis is to address the process monitoring problem from an integrated perspective, considering all the designed methodologies into an advanced plant information system framework, and not only the fault diagnosis system in itself. The thesis is structured as follows. The first chapter explains the main motivations and objectives of the work, while Chapter 2 consists of a literature review that addresses the most significant issues around process monitoring. Chapter 4 presents with some detail some of the tools and techniques more used in this thesis. After that, Chapter 5 and Chapter 6 introduce two methodologies for sensor location. The first one focuses in sensor networks accuracy while the later focuses on sensor networks reliability. Chapter 7 presents a methodology to select sensor signals to improve process monitoring performance. Chapter 8 presents a straightforward methodology to identify the process delays to improve process monitoring.