Analysis of underlying causes of investigated loss of containment incidents in Dutch Seveso plants using the Storybuilder method

In the Netherlands there are around 400 “Seveso” sites that fall under the Dutch Major Hazards Decree (BRZO) 1999. Between 2006 and 2010 the Dutch Labour Inspectorate's Directorate for Major Hazard Control completed investigations of 118 loss of containment incidents involving hazardous substances from this group. On the basis of investigation reports the incidents were entered in a tailor-made tool called Storybuilder developed for the Dutch Ministry of Social Affairs and Employment for identifying the dominant patterns of technical safety barrier failures, barrier task failures and underlying management causes associated with the resulting loss of control events. The model is a bow-tie structure with six lines of defence, three on either side of the central loss of containment event. In the first line of defence, failures in the safety barriers leading to loss of control events were primarily equipment condition failures, pre start-up and safeguarding failures and process deviations such as pressure and flow failures. These deviations, which should have been recovered while still within the safe envelope of operation, were missed primarily because of inadequate indication signals that the deviations have occurred. Through failures of subsequent lines of defence they are developing into serious incidents. Overall, task failures are principally failures to provide adequate technical safety barriers and failures to operate provided barriers appropriately. Underlying management delivery failures were mainly found in equipment specifications and provisions, procedures and competence. The competence delivery system is especially important for identifying equipment condition, equipment isolation for maintenance, pre-start-up status and process deviations. Human errors associated with operating barriers were identified in fifty per cent of cases, were mostly mistakes and feature primarily in failure to prevent deviations and subsequently recover them. Loss of control associated with loss of containment was primarily due to the containment being bypassed (72% of incidents) and less to material strength failures (28%). Transfer pipework, connections in process plant and relief valves are the most frequent release points and the dominant release material is extremely flammable. It is concluded that the analysis of a large number of incidents in Storybuilder can support the quantification of underlying causes and provide evidence of where the weak points exist in major hazard control in the prevention of major accidents.     

Harnessing database resources for understanding the profile of chemical process industry incidents

Analyzing historical databases can provide valuable information on the incident occurrences and their consequences for assessing the safety of the chemical process industry. In this study, the RMP and HSEES databases were utilized to understand the patterns and the factors influencing chemical process industry incidents. Frequency exceedance curves were generated by utilizing the different incident consequences from the databases to understand the profile of societal loss from reported incidents. Understanding the statistics and trends of the historical incidents could serve as important lagging indicators in order to assess the probable proximity to major consequences from the low-probability/high-consequence incidents. To this regard, the safety pyramids were also generated to better understand the relationship between the different consequences of the reported incidents. Furthermore, the safety pyramids were analyzed in comparison with the traditional safety pyramid proposed by Heinrich to understand the US process industry incident occurrence trends.     

What can the trove of CSB incident investigations teach us? A detailed analysis of information characteristics among chemical process incidents investigated by the CSB

Understanding the commonalities among previous chemical process incidents can help mitigate recurring incidents in the chemical process industry and will be useful background knowledge for designers intending to foster inherent safety. The U.S. Chemical Safety and Hazard Investigation Board (CSB) reports provide detailed and vital incident information that can be used to identify possible commonalities. This study aims to develop a systematic approach for extracting data from the CSB reports with the objective of establishing these commonalities. Data were extracted based on three categories: attributed incident causes, scenarios, and consequences. Seventeen causal factors were classified as chemical indicators or process indicators. Twelve chemical indicators are associated with the hazards of the chemicals involved in the incidents, whereas five process indicators account for the hazards presented by process conditions at the time of the incident. Seven scenario factors represent incident sequences, equipment types, operating modes, process units, domino effects, detonation likelihood for explosion incidents, and population densities. Finally, three consequence factors were selected based on types of chemical incidents, casualties, population densities, and economic losses. Data from 87 CSB reports covering 94 incidents were extracted and analyzed according to the proposed approach. Based on these findings, the study proposes guidelines for future collection of information to provide valuable resources for prediction and risk reduction of future incidents.     

Analysis of equipment failures as contributors to chemical process accidents

A database study of chemical process accident cases was carried out. The objective of the study is was to identify the reasons for equipment based accidents. The most frequent accident causing equipment were piping (25%), reactors and storage tanks (both 14%) and process vessels (10% of equipment accidents). The six most accident-prone equipment is process related involve nearly 80% of accidents.78% of equipment accident contributors are technically oriented including design and human/technical interface faults. Purely human and organizational reasons are the most common accident contributors for storage tanks (33%), piping (18%) and heat transfer equipment (16% of causes). For other equipment the technical accident causes are most common.The accident contributors were divided to main and sub-contributors. On average process equipment failures have 2.2 contributors. The contributors, which frequent and act often as main contributors, should be focused. These risky contributors were identified for several equipment types. Also a deeper analysis of the accident causes and their interconnections was made. Based on the analysis a checklist of main risk factors was created for hazard identification on different types of equipment.     

Priority analysis for risk factors of equipment in a hydraulic turbine generator unit

Ensuring the safe operation of hydropower stations is one of the key challenges for electric generation. Clearly the safe operation of such systems can only be archived with proper and effective maintenance scheduling. The objective of this study is to analyze, rank and prioritize the risk factors responsible for equipment failures of a hydraulic turbine generator unit (HTGU) based on operating data and expert elicitation. Here a simple qualitative risk evaluation model is proposed able to consider seven typical failures in HTGU. The proposed tool is applied for the risk prioritization of equipment failures, e.g. shaft torsion, misalignment, rotating fault, axis bend, runner fault, water guide, and wicket gate of a hydropower station in China. The obtained results have been compared against the actual statistics of equipment failures of a hydropower station in China, considered showing good agreement. All of these results provide theoretical guidance for digitalization realization of equipment failures.     

Subsea pipelines incidents prevention: A case study in Brazil

There are more than 4000 subsea pipelines in Brazil. These pipes include umbilicals, drilling risers, flexible risers, rigid risers, hybrid risers, flowlines, and export pipelines. Despite all standards, regulations, guides, and risk management tools designed to avoid events, subsea pipeline incidents still occur, revealing possible failures in companies' risk control. Identifying similarities between different subsea pipeline failure events is crucial to improving the design, risk management practices, and regulation requirements, besides promoting accident prevention. This paper proposes applying the life cycle and management practices combined to analyze subsea pipeline incidents from the RDI (Detailed Incident Report) and investigations reported to ANP (Brazilian National Agency of Petroleum, Gas, and Biofuels), the Brazilian safety regulatory agency. Furthermore, subsea pipeline incidents data were analyzed: correlated circumstances, consequences, and causes. The results show that most riser and flowlines causal factors are related to equipment failures, and recurrent root causes are design errors and integrity control. Based on the proposed approach, it was possible to identify gaps in most riser and flowlines accident investigations since there are few causal factors, root causes, and the absence of riser and flowlines failure mode and mechanisms. Therefore, the development of accident recommendations can be compromised. Thus, this paper proposes improvements to current Brazilian regulations to clarify the minimal subsea pipeline accident investigation requirements.     

Migrating an incident reporting system to a CCPS process safety metrics model

A major chemical company established a formal incident investigation and reporting system several years ago. The original system focused heavily on worker-related injuries, illnesses, and near-misses and was used primarily to track statistics reportable to the Occupational Safety and Health Administration (OSHA). This Occupational Injury and Illness (OII) approach has been recognized to be an ineffective tool for measuring, predicting, and preventing process safety incidents. The Center for Chemical Process Safety (CCPS) recently published guidelines on how to establish safety metrics for the measurement and reduction of process safety incidents. The process safety metrics approach relies upon leading and lagging metrics to improve organization process safety. This paper is a case study of the analysis of one organization’s incident database, which represents approximately five years of data from over a dozen facilities. The aim of this investigation was to extract useful process safety metrics from the incident investigation and reporting system, which would be pertinent to the types of process units and process functions at these facilities. This paper will discuss the approach taken to extract process incident information from an OII-based database and present the difficulties of performing an analysis on such a database. This paper provides guidance on how to migrate an existing OII-based reporting system to a program that includes process safety metrics in accordance with industry best practices.     

Determining a realistic ranking of the most dangerous process equipment of the ammonia production process: A practical approach

OCI Nitrogen seeks to gain knowledge of (leading) indicators regarding the process safety performance of their ammonia production process. The current research determines the most dangerous process equipment by calculating their effects resulting from a loss of containment using DNV GL's Phast™ dispersion model. In this paper, flammable and toxic effects from a release from the main equipment of an ammonia plant have been calculated. Such an encompassing approach, which can be carried out for an entire plant, is innovative and has never been conducted before. By using this model, it has been demonstrated that the effects arising from an event of failure are the largest in process equipment containing pressurized synthesis gas and ‘warm’ liquid ammonia, meaning the ammonia buffer tanks, ammonia product pumps, and the ammonia separator. Most importantly, this document substantiates that it is possible to rank the most hazardous process equipment of the ammonia production process based on an adverse impact on humans using the calculated effect distance as a starting point for a chance of death of at least 95%. The results from the effect calculations can be used for risk mapping of an entire chemical plant or be employed and applied in a layer of protection analysis (LOPA) to establish risk mitigation measures.     

Global process safety incidents in the pharmaceutical industry

The objective of this research is to analyse global process safety incidents within the pharmaceutical industry in terms of their consequences and factors contributing to the incidents. There were 73 process safety incidents leading to 108 fatalities found between 1985 and 2019. Trends between the number of incidents, number of fatalities, location, and contributing factors were identified and summarized. The most reported fatalities occurred in 2018 & 2019. 83% of fatalities occurred in China and India. Explosions were associated with 71% of incidents, which resulted in 89% of fatalities. For most of the international incidents, incident investigations were not available and thus insufficient details were available to determine the causes. Contributing factors were available or estimated from available data for about half of the incidents, with the following most common: hazard awareness & identification; operating procedures; design; safeguards, controls & layers of protection; safety culture; and preventive maintenance. These findings can be used as a basis to improve process safety performance in the pharmaceutical industry.     

Fuzzy dynamic risk-based maintenance investment optimization for offshore process facilities

A methodology for maintenance planning is developed which helps in improving the reliability of the components and safety performance in process facilities. This methodology helps design an optimum safety maintenance investment plan by integrating the optimization techniques and a fuzzy dynamic risk-based method. Intuitionistic Fuzzy Analytic Hierarchy Process (IFAHP) is applied to deal with uncertain data. The proposed approach employs multi-experts’ knowledge which helps to optimize the maintenance investments. A separator system in an offshore process facility platform is selected as a case study to demonstrate the application of the proposed methodology. A practical example in the separator system is surveyed and potential failures and Basic Events (BEs) are identified. Finally, a risk-based maintenance plan is provided for future safety investment analysis. The results indicate that the developed methodology estimates the risk more accurately, which enhances the reliability of future process operations.     

Operational risk assessment of chemical industries by exploiting accident databases

Accident databases (NRC, RMP, and others) contain records of incidents (e.g., releases and spills) that have occurred in the USA chemical plants during recent years. For various chemical industries, [Kleindorfer, P. R., Belke, J. C., Elliott, M. R., Lee, K., Lowe, R. A., & Feldman, H. I. (2003). Accident epidemiology and the US chemical industry: Accident history and worst-case data from RMP*Info. Risk Analysis, 23(5), 865–881.] summarize the accident frequencies and severities in the RMP*Info database. Also, [Anand, S., Keren, N., Tretter, M. J., Wang, Y., O’Connor, T. M., & Mannan, M. S. (2006). Harnessing data mining to explore incident databases. Journal of Hazardous Material, 130, 33–41.] use data mining to analyze the NRC database for Harris County, Texas.Classical statistical approaches are ineffective for low frequency, high consequence events because of their rarity. Given this information limitation, this paper uses Bayesian theory to forecast incident frequencies, their relevant causes, equipment involved, and their consequences, in specific chemical plants. Systematic analyses of the databases also help to avoid future accidents, thereby reducing the risk.More specifically, this paper presents dynamic analyses of incidents in the NRC database. The NRC database is exploited to model the rate of occurrence of incidents in various chemical and petrochemical companies using Bayesian theory. Probability density distributions are formulated for their causes (e.g., equipment failures, operator errors, etc.), and associated equipment items utilized within a particular industry. Bayesian techniques provide posterior estimates of the cause and equipment-failure probabilities. Cross-validation techniques are used for checking the modeling, validation, and prediction accuracies. Differences in the plant- and chemical-specific predictions with the overall predictions are demonstrated. Furthermore, extreme value theory is used for consequence modeling of rare events by formulating distributions for events over a threshold value. Finally, the fast-Fourier transform is used to estimate the capital at risk within an industry utilizing the frequency and loss-severity distributions.     

Threshold values for domino effects caused by blast wave interaction with process equipment

The present study focuses on the definition and assessment of overpressure threshold values for the damage to equipment caused by blast waves originated by primary accidental scenarios. A revision of literature data and of the available damage probability models was carried out. Threshold values were proposed for different categories of process equipment, taking into account either damage levels or release intensities following the loss of containment. Specific threshold values for domino effect were also proposed.     

Accident data for the Semantic Web

This paper describes concerns about the documentation, dissemination and use of lessons learned from mishap investigations, impediments posed by current practices, and opportunities for improvement. Lessons are presently developed, documented and stored primarily in narrative form and relational databases, and disseminated in many forms and media, including the Internet. Current practices pose many impediments to maximized development, dissemination and use. Investigation process research and new data concepts behind the Semantic Web, exploited elsewhere, offer potential opportunities to overcome these impediments. To exploit these opportunities, formation of a working group to develop an improved Semantic Web-friendly mishap investigation lessons learning system is proposed. An example illustrating an alternative approach is described to support a reasonable expectation that an alternative lessons learning system could be developed.     

Method for identifying contributors to chemical process accidents

The paper presents a new method for identifying contributors to chemical process accidents by exploiting knowledge on causes of past accident cases. Accident reports from the Failure Knowledge Database were analyzed and utilized for hazard identification. The accident information gathered was used as a basis to develop an accidents ranking and points-to-look-for approach for the safe design and operation of chemical process equipment. In the method, accident contributors including technical, design and operation errors of major process equipment types and piping are identified. The method is applicable throughout the process lifecycle, even for process changes in the early design stages. The Bhopal tragedy is used as a case study to demonstrate and test the method. The proposed method can predict on average up to 85% of accident causes and design and operation errors.     

Managing process chemicals,technology and equipment information for pilot plant based on Process Safety Management standard

Injuries, accidents or even fatalities while working in pilot plant are reported worldwide. The OSHA Laboratory Standard and Hazard Communication Standard have been used as a guideline to manage safety of laboratories and pilot plant. In spite of the implementation of these standards, incidents which result in injuries and property loss are continuously occurring. The implementation of OSHA Process Safety Management (PSM) Standard in pilot plant is expected to further reduce the risks of accidents. This paper presents a new system for managing process chemicals, technology and equipment information in pilot plant and the concept is developed based on Process Safety Information (PSI) element of PSM 29 CFR 1910.119(d). It provides organized strategies to manage documentations, communicate information, and written program for maintaining, revising and updating related information. Process and Instrumentation Diagram (P&ID) is used as a foundation for data management. Implementation of this system at the CO₂ Hydrocarbon Absorption System pilot plant as a case study is examined and discussed.     

Ageing management and monitoring of critical equipment at Seveso sites: An ontological approach

A safe “ageing” of Seveso establishments is a challenge for both operators and regulators. To this scope, Seveso III Directive required to integrate the equipment integrity issue into the safety management system for the major accident prevention; at the same time, the Italian Authority adopted a short-cut method for a quick ageing evaluation, which awards the application of the best techniques to control integrity and prevent deterioration-related failures. In this paper, the use of the ontology has been proposed to support decision-making about the implementation of technical solutions to control equipment ageing and comply the requirements of the Seveso legislation. To contrast deterioration mechanisms, the rapid development of data intensive smart sensors should be exploited and, in this frame, the automated on-line direct monitoring of equipment conditions, based on innovative low-cost sensors, is a novelty and promising solution. The developed ontology-based system points towards the adoption, when possible, of on-line monitoring. This solution provides much more data than traditional measurements and it is essential for the operators to understand how to merge concurrent information and data and to adequately control equipment deterioration. The ontology-based approach appears a viable solution even for this purpose. To demonstrate its potentiality, a real use-case has been used, where the model has been tested in finding the best technical solutions to improve the ageing management of an atmospheric distillation unit of a refinery in order to comply with safety requirements. A further use-case is given to show how the model can be used to react, after real-time damage signals, to restore safety conditions by means of an adequate decision-making.     

Calibrate failure-based risk assessments to take into account the type of chemical processed in equipment

Currently, failure-based risk assessments in the process industry do not empirically take into account the type of chemicals processed in equipment, mainly because chemical-specific failure rate data barely exist. This paper suggests a methodology to calibrate failure-based risk assessment predicated on the chemical being processed in equipment. The methodology uses a data mining tool known as the association rule. Specifically, the lift association rule is utilized (the Lift Methodology). By extracting equipment failure information from incident databases based on the chemical involved in the process, the Lift Methodology leads to more accurate equipment-related risk assessment.     

Self-Organizing Map and clustering algorithms for the analysis of occupational accident databases

Data mining techniques are a powerful method for extracting information from large databases. Among these techniques, clustering and projection of data from high-dimensional spaces hold a main role, since they allow to discover hidden structures in the data set. Following this approach, this paper presents a data analysis method designed to help the management and investigation of occupational accident databases. The purpose is to discover the most common sequences of events leading to accidents for devising preventive actions. To this aim, we developed a two-level approach based on the joint use of the Kohonen’s Self-Organizing Map and the k-means clustering algorithm. This approach allows not only to group the accidents in different classes but also to visualize them in a way understandable for the analyst. The method has been applied with satisfactory results to a large database of occupational accidents occurred in the Italian wood processing industry. A comparison with the Hierarchical Clustering method confirmed the effectiveness of the proposed approach.