Systems Theory Based Accident Analysis for the Process Industry

Process plants (refineries, chemical plants, petrochemical, pharmaceutical, etc.) deal with a large amount of potentially dangerous materials (toxic, inflammable, explosive) and many times in extreme conditions (such as high temperatures and pressures). This can lead to equipment failures, plant shutdowns or even worse accidents with catastrophic consequences.

In spite of the safety layers of protection (basic control, alarms, SIS, protective systems, etc.) there are accidents every day with losses up to 1.000 million $ each year only in the US refineries. The existence of accidents is due mainly to the increasing complexity of the process plants. This complexity appears because of two factors. The first one is a more complex process structure (energy integration, minimum waste, higher demands on yield and production, environmental constraints), the second one is a more complex control system (systems that performs many more tasks than before with a non predictable software). This complexity problem is even worse because both factors are not independent but they highly interrelated.

The goal when investigating an accident should be to know why the accident happened and how similar losses can be avoided in the future. Unfortunately, nowadays most of the accidents reports use an event-based model approach. They analyse the facts, following the chain of events searching for a root cause for the accident. When a cause root is found (many times still a human operator) the investigation is closed. This approach usually lacks in depth of why the events occurred and the opportunity to extract important information for (the prevention) of future accidents is lost.

In this paper we analyse the application of a new methodology for the investigation of accidents in the process (chemical,  petrochemical, oil&gas) industry. The methodology is called CAST (Causal Analysis based on STAMP). STAMP (System-Theoretic Accident Model and Processes). It is a systems theory based methodology to ensure systems safety developed by prof. Leveson at MIT.  This model considers safety as an emergent property of the system that is enforced by safety constraints. The key point is to control the behaviour of the components and the system to guarantee that the safety constraints are met. CAST can be used to identify the questions that need to be answered to fully understand why the accident occurred. CAST extends the analysis to the entire sociotechnical system searching for weaknesses in the existing safety control structure and identifying the changes needed not only to remove the causal factors but the systemic ones as well. At the moment CAST has been applied to other domains (aviation, transport, etc.) but it has not been applied in depth to the process industry.

In this work we apply CAST to a real industrial accident and compare the analysis of the official report with the one obtained using the systemic approach. Conclusions on the application, strengths and weaknesses, of the methodology to the process industry are presented.

References

Leveson, N. (2011). Engineering a safer world. MIT Press.