Cause and Effect Analysis for Decision Support in Alarm Floods

Abstract

The presence of faults in large-scale facilities, quite frequently propagates plant-wide due to energy integration and consequently triggers a sequence of alarms, which are known as alarm floods. For the purpose of decision support in the root cause identification of alarm floods, this paper presents a framework to combine causality inference using both process and alarm data. The method includes the following main steps: first, alarm floods are identified from alarm data; second, root cause alarms are detected through cause and effect analysis; third, associated process variables are extracted and causal relations are confirmed. Using this method, the number of alarm tags analyzed for root cause identification is short-listed. Causal relations between the corresponding short-listed process variables are then used to support root cause analysis. The practical utility of the method is demonstrated by application to an industrial dataset for which both process and alarm data are available.