Emergency shutdowns of propylene production plants: Root cause analysis and availability modeling

Unplanned shutdown events in chemical process plants cause revenue losses due to production curtailments, repair costs, and harmful environmental and safety effects on plant personnel and the surrounding community. Several recent studies have used news articles and databases to perform statistical analyses and modeling of the causes and consequences of chemical process accidents, but these studies do not analyze the operating deviations that led to shutdowns. This work structures narratives about unplanned shutdown events into data, analyzes the causes of those events, and uses the collected data to model the failure probability of process plants. A case study comparing two propylene production technologies, steam cracking and propane dehydrogenation is developed. We use a set of 118 shutdown events reported in four sources across a five-year span for 7 propylene production plants. Using this data, the three leading causes for unplanned shutdowns in both propylene production technologies were determined to be: equipment trips/shutdowns, utility supply interruptions, and weather-related events. A power law process (PLP) for data from multiple repairable systems is used to model the failure behavior for both technologies. For the data collected, this procedure indicated an increasing failure intensity function for both technologies, with propane dehydrogenation increasing more significantly. In addition, both maximum likelihood estimates (MLEs) and confidence intervals for the PLP parameters are presented. Since these are repairable systems, an estimated inherent availability is determined for both technologies. Finally, recommendations that address how to reduce the occurrence of these events and mitigate their consequences are discussed.

Automated summary

Unplanned shutdown events in chemical process plants lead to revenue losses, repair costs, and environmental and safety impacts. This study analyzes the causes of these events and models the failure probability of process plants, providing recommendations for reducing the occurrence and mitigating the consequences of shutdown events.