Proactive Alarms Monitoring using Predictive Technologies

Abstract

Chemical plants are now built to have large number of integrated and interlinked process units so as to optimize production and reduce waste. When an abnormal situation occurs, the automation systems alert the operators through alarms and help orient them to the new state. However, as most processes are highly-coupled, many simultaneous alarms can occur resulting in a flood of alarms. This can create confusion among operators, who must diagnose and rectify the fault before the condition escalates. Sometimes when the abnormal situation cannot be properly diagnosed, the operator will then activate the emergency shut-down, resulting in loss of productivity. In this work, we propose a novel alarm management framework, called proactive alarms monitoring scheme that provides operators with anticipatory information on incipient alarms that could happen within a certain time-window. This anticipatory information, which is built around a predictive algorithm, seeks to improve the sense-making facilities offered by the alarm system. As such, it allows plant operators to adopt a more proactive approach in managing alarm floods during various abnormal situations. © 2012 Elsevier B.V.