Although mandatory in most high-risk industries, the safety management system (SMS) is often criticized as burdensome and complex. Through its requirement to formalize all main activities, the SMS is perceived as bureaucratic and a vehicle for pure compliance and Safety I (one). Furthermore, the SMS is often detached from an organization’s core activities, goes against local practice and does not deliver the safe performance that was hoped for. By comparing the model behind SMS with specific requirements for process capability, this paper identifies a safety fractal that reflects the basic requirements that are needed to control safety related activities at all levels within an organization. It is further argued that the constituent elements of this safety fractal are particularly suitable to organize resilient performance, provided that resilience is explicitly identified as the safety strategy to follow and, as such, consequently implemented. This approach is then positioned against common safety management concepts as management system maturity, leadership and safety culture, leading to a systematic and a more comprehensive view on how to measure safety performance and resilience.

The concept of a safety management system (SMS) to control the risks of operational activities has been introduced in high-risk industries already some decades ago. SMS requires accidents/incidents to be reported and analysed and measures to be taken to prevent future events. Additionally, national investigating bodies have been given the role of independently investigating serious events, with the same goal. The current practice in accident and incident investigation however, does not provide a systematic approach to analyse elements of SMS. As a direct consequence, the opportunity to use these investigations for introducing sustainable system changes is often missed. The paper describes the SAfety FRactal ANalysis (SAFRAN) method that is developed to guide investigators to explore the composing elements of an SMS in a natural and logic way, starting from the findings close to operations that explain the occurrence – being the elements accident investigators are first confronted with. The paper further informs on the application of the SAFRAN method to review a selected set of published railway accident investigations, all reporting on occurrences related to over-speeding, possibly resulting in a (lethal) derailment. The depth and focus of the performed investigations is assessed and compared with a reference mode of expected findings that would result from an analysis that is applying the SAFRAN logic. This demonstrates the need, in order to introduce sustainable changes, to focus accident analysis on an organisation's capability of managing the variability that might put successful process performance at risk.