A resilience-based causal framework for conducting safety analysis

Lauchlan J Clarke
2020
A causal relationship is a relative and innate concept that is necessary for reasoning and intervening within an uncertain domain. Almost every field of study has established causal frameworks designed to utilise salient information to satisfy a specific purpose. Due to its comparatively recent maturation and a perceived alignment of purposes, the existing causal framework adopted in formal safety analysis borrows heavily from other fields, such as law and engineering. Although safety
more » ... ies are becoming increasingly sophisticated, their underlying causal assumptions have mostly continued unchallenged despite their direct influence on safety outcomes. Recently, safety methodologies based on the current causal framework have received criticism related to their ability to meet increased public safety expectations when applied to complex socio-technical systems. While these criticisms are justified, they relate to a specific framework of causality rather than the use of causality itself. Based on the qualitative concepts of Resilience Engineering and Safety-II, this research explicates a new causal framework capable of generating proactive safety recommendations in complex socio-technical systems. Based on this framework, a methodology is developed that utilises a functional decomposition scheme to learn the structure of causal networks. The approach incorporates a broad range of information and focusses on the decisions and tasks required for a system to achieve its purpose. The new methodology is applied to benefit safety in a novel maritime transhipping operation. Implicitly, the approach develops a causally structured reservoir of information relevant to the functioning of the system. This information develops the systems potential to anticipate and respond to variability. Explicitly, the networks act as an oracle for exploring hypothetical contexts and assessing the effect of proposed recommendations.
doi:10.25959/100.00034820 fatcat:vvtpvnun65bi5bpuq2rcu4eepy