Organisational Learning from Accidents and Incidents in European Railways

Accidents and incidents provide ample, although undesired opportunities to improve understanding of system processes with the aim to improve control of risks that are inherent in indented activities. Railway industry in Europe knows that accident review is important; moreover, many of not most railway workers have high safety awareness. However, learning from railway accidents by railway organisations is hardly organised despite all the effort over the last 150 years that has gone into collection of accident and incident data. Recent studies (also) within the context of SAMRAIL demonstrate that learning processes hardly exist, or only by chance due to devoted individuals (Koornneef and Kingston, 2003). The privatisation of formerly national railway companies also has contributed to fragmentation of rudimentary organisational learning processes (Hale et al., 2002, Lodge, 2003). Thus, learning from railways-related accidents is no longer a monopoly of one of the stakeholders in the railway industry. Lessons to be learned will be different for different stakeholders; opportunities to learn effectively will also be different. Some risks identified through learning from accidents or incidents might better controlled by another stakeholder. All these considerations indicate the necessity to conceive organisational learning from railway-related accidents and (near-miss) by and for the European railway industry in order to meet the interoperability objective. It is essential to identify what sorts of lessons can be learned from railway incidents at targeted levels within the Europe. These classes of learnable lessons differ for distinct levels. Starting from the collecting of raw accident/incident data value must be added for learning, e.g. about relevant system factors, to improve management of risks by ultimately at operational level. Higher-level learning requires data that is produced by processing lower level event data. Organisational learning (OL) is a systematic process by which operational surprises are identified and analysed, corrective actions are devised and implemented, and the effectiveness and impact of the corrective actions are monitored. An operational surprise occurs when somebody or some monitoring system detects operational conditions that were not anticipated and might lead to an unsafe state (Koornneef, 2000). Surprises are directly observed through hazardous events, such as SPAD or over-speeding, however there are some which have to be derived from benign events, such as omissions to acknowledge alarms. There could be many different causes for a surprise, however, the primary objective of OL is to identify manageable root causes, and devise appropriate corrective feasible measures that effectively prevent (surprise) recurrence and/or control the operational risk within acceptable levels. This devising is the function of `Learning Agency¿; implementation is a responsibility of appropriate line management. Data collected, processed and stored for this purpose also provided a sound basis for defining meaningful safety indicators. The SAMRAIL WP2.6 Workshop on Accident & Incident Data Requirements on September 17, 2003 (Koornneef and Kingston, 2003), revealed that clear insights in contents of and provisions for enabling effective learning from incidents by the participant's railway organisations are largely lacking; in addition, various barrier that obstruct such effective learning were identified. This document addresses key issues that need to be worked out in close cooperation with relevant stakeholders and proposes a roadmap to organisational learning from railway accidents and incidents within the EU.