Process risk assessment uses big data


Case study 1: Refinery shutdown

Figure 2: Variation of likelihood of shutdown for a pressure differential variable over 4 months shows indicators at shutdown. Courtesy: Near-Miss ManagementAn industrial fluid-catalytic-cracking-unit in a large refinery experienced an unexpected shutdown in the middle of high-demand operations. Despite having a strong PSM program and alarm management philosophy in place, the immunity of process to disturbances deteriorated progressively until the disturbances overpowered the controllers and led to a shutdown.

Investigation afterward used risk analysis methodology to rapidly analyze more than 18 months of data, revealing that the deterioration occurred subtly over a period of several months. Plant systems and personnel that were in place could not see it. Increasing risk levels for several variables (before shutdown) were identified, indicating significant deviations from normal behavior, unseen with standard plotting, trending, and visualization of data. In particular, a pressure differential variable (for one of the stand pipes) was identified, exhibiting strong leading indicators before this shutdown. One leading indicator metric (likelihood of shutdown) increased gradually to 58% over several months. Gaps in standard risk assessment tools did not provide plant management with visibility of changing conditions.

By seeing changing risk levels dynamically (daily, weekly, monthly), a plant operations team can identify when problems begin and use leading indicators to take proactive measures to prevent and avoid operational problems. Predictive knowledge helps engineers and safety personnel monitor effectiveness of existing risk reduction measures and find issues early on, so management can allocate resources to the most needed areas.

Case study 2: Acid runaway

Figure 3: Variations of risk level for a key analyzer variable are shown over 5 months. Courtesy: Near-Miss ManagementAn industrial plant was unable to pinpoint the cause of an acid runaway incident, deemed to be a progressive event, not sudden human error. Figure 3 shows that one of the analyzer variables had been experiencing significant likelihood of exceeding its critical levels for several weeks before the incident. This one piece of information could have prevented the incident, with plant personnel taking corrective actions in advance. There were two periods where hidden data indicated events with high risk potential, not captured with existing risk assessment tools.

Many process conditions, like acid runaway, are difficult to detect with trending and visualization techniques. In such cases, big data analytics can highlight issues that are lost to other approaches and not observable to the human eye. Big-data analytics separates important from insignificant data.

Risks continue

Despite technological advances in automation and growing awareness for process safety, it is evident that industrial facilities still struggle with unexpected shutdowns and incidents. Current risk assessment methods fall short in their capacity to identify and predict risks dynamically. Harnessing process-specific big data with information on precursors indicates when risks are at their developing stages. This approach is independent of reporting by employees, often insufficient in capturing information.

With associated benefits such as transparency of risk information, automatic capture of problems, fact-based decision making, and effective monitoring of maintenance measures, predictive assessment of big data complements current PSM, hazard identification, and quantitative risk analyses techniques, and lays the foundation for next generation risk assessment. 


Kleindorfer, P.R., et al., “Accident Epidemiology and the U.S. Chemical Industry: Accident History and Worst-Case Data from RMP*Info,” Risk Analysis, 23 (5), pp. 865–881 (2003).

Kleindorfer, P., Oktem, U.G., Pariyani, A., and Seider, W.D., “Assessment of catastrophe risk and potential losses in industry,” Computers and Chemical Engineering, 47, 85-96 (2012).

Lauridsen, K., Kozine, I., Markert, F., Amendola, A., Christou, M., and Fiori, M., “Assessment of uncertainties in risk analysis of chemical establishments,” Summary Report on ASSURANCE project, Risk National Laboratory, Roskilde, Denmark (2002).

Pariyani, A., Seider, W.D., Oktem, U.G., and Soroush, M., “Incidents Investigation and Dynamic Analysis of Large Databases in Chemical Plants: An FCCU Case Study,” Ind. Eng. Chem. Res., 49, 8062-8079, 2010.

Phimister, J.R., Oktem, U., Kleindorfer, P.R., and Kunreuther, H., “Near-miss incident management in the chemical process industry,” Risk Analysis, 23(3), 445–459 (2003).

-By Ankur Pariyani , PhD, Ulku G. Oktem, PhD, and Deborah L. Grubbe, PE, Near-Miss Management LLC; Edited by Mark T. Hoske, content manager, CFE Media, Control Engineering and Plant Engineering,


Ankur Pariyani, PhD (corresponding author), is co-founder and chief technology officer of Near-Miss Management LLC, where he focuses on product development and innovation. He has authored several papers on risk analysis. Within the Near-Miss Management research team, he has developed breakthrough engineering techniques for identifying critical near-misses and predicting incidents in plants, setting a strong conceptual foundation for future growth. He holds a PhD and M.S. from the University of Pennsylvania and a B. Tech from the Indian Institute of Technology in Guwahati, all in chemical engineering.

Ulku Oktem, PhD, is co-founder and president of Near-Miss Management LLC, where she oversees product development and operations. She also serves as adjunct professor at the Operations and Information Management Department and as senior research fellow at the Risk Center of the Wharton School of the University of Pennsylvania. She holds a PhD in chemical engineering from the University of Delaware, M.S. from the Clarkson College, and B.S. from the Middle East Technical University. Dr. Oktem is globally recognized as a leading expert and researcher in the area of near-miss management systems. Her research has received international recognition and in-depth coverage in several academic journals and other publications. Prior experience includes managing product development and manufacturing of various specialty chemicals at Rohm & Haas Co. She also set up her own consulting company providing safety, health, and environmental training services to Fortune 500 companies.

Deborah L. Grubbe, PE, is chief marketing officer of Near-Miss Management LLC and is owner and president of Operations and Safety Solutions LLC, a consultancy that specializes in process safety leadership and safety culture. She is the former vice president of group safety for BP plc, which had its two safest years under her watch. Grubbe worked 27 years at DuPont, where she held corporate director positions in safety, operations, and engineering. Assignments included capital project implementation, process safety implementation, strategic safety assessments, manufacturing management, and human resources. From 2003-2012, Deborah served as a member of the NASA Aerospace Safety Advisory Panel, and was a consultant on safety culture to the Columbia Shuttle Accident Investigation Board. In May 2012, she received the NASA Exceptional Public Service Medal from NASA Administrator Charles Bolden. Deborah serves on the Purdue University College of Engineering Dean’s Advisory Council. She is an emeritus member of the Center for Chemical Process Safety and has worked with the National Academy of Sciences to support the Demilitarization of the U.S. chemical weapons stockpile. Deborah is a former member of the board of directors of American Institute of Chemical Engineers (AIChE), and is the current chair of the AIChE Institute for Sustainability. In 2002, she received the Purdue Distinguished Engineering Alumni Award, and was named “Engineer of the Year” in the State of Delaware. In May 2010, Deborah was awarded an Honorary Doctorate in Engineering from Purdue University.


See article about how Near-Miss Management LLC's commercial software, Dynamic Risk Analyzer, analyzes process and alarm data to dynamically determine risks of continuous processes.

<< First < Previous Page 1 Page 2 Next > Last >>

click me