Industrial Application of Human Factors Safety Standards

The main objective of this thesis is to investigate how selected standards are applied within an industrial context and to suggest strategies to improve their application. This thesis focuses on the human part of large and complex systems. Humans are essential in any large and complex system, but their complexity makes it virtually impossible to fully predict their behaviour and impact on such systems. Safety standards should be an aid to handle such complexity in a safe, efficient and effective manner. Standards adapt several strategies, such as prescriptive or goal-based and process or product focus, to utilize advantages and manage challenges related to humans as parts of safety-critical and safety-related systems. A survey of selected HSE standards used in the Norwegian petroleum industry revealed that the standards differ in how goal-based or prescriptive and how processor product-oriented they are and recent standards were not more goal-based than older ones. Studies of the application of ISO 11064 revealed that the standard is well appreciated measured by the opinion of its users. However, many users of the standard also found it to be too comprehensive to be used efficiently and effectively. Also, the application of the standard must be adapted to each project. Studies of the partly prescriptive CRIOP methodology revealed that it is well appreciated measured by the opinion of its users, but CRIOP also requires adaptation in its application to be effective and efficient in use. Even strictly prescriptive standards are adapted in their application to real world problems. An analysis of the application of the prescriptive standard ON-S1 in Australian rail occurrence reporting revealed that the introduction of this standard significantly increased the number of reported serious injuries. This was due to a mismatch between a prescriptive requirement and the corresponding application. Human operators as users of safety-related and safety-critical systems must make correct decisions and safe actions to avoid accidents. One reason for applying HF safety standards is to help operators achieve that goal. However, according to existing literature, human error is a causal or contributing factor in 60-90 percent of accidents and incidents. The validity of these figures will be discussed in the thesis. An adaptation of the Human Factors Assessment and Classification System (HFACS) for use in the Oil & Gas industry revealed that latent failures on the organizational level were the most common error type. The thesis suggests applying a safety case-based approach to model requirements and recommendations in standards to bridge the gap between goal-based and prescriptive perspectives in standards. There appears to be a gap between requirements and recommendations in standards, the way standards are actually applied and the way compliance to the standards is demonstrated. Since many systems contain legacy systems, reused components or COTS (Commercial off the Shelf), a retrospective safety case can be produced to integrate such system parts in an overall case. Modelling templates or patterns for standards can make standards easier to read, understand and use. Such an approach can also provide means for reuse, by reusing known safety argument structures and combinations of evidence. The thesis focuses primarily on industrial application of standards, but the knowledge provided can be valuable input to the drafting of standards as well. Also, regulators can use such an approach by comparing system safety models to libraries of patterns or models to identify deviations from known solutions that might require further analysis for approval. Efficient and effective comparison of system solutions to best practice patterns or models will however require a more formal notation of the argument structure than what is available today. One solution to this could be to further develop the Goal Structuring Notation (GSN) into a more formal notation, or even integrate GSN with UML.