No abstract

Analysis of helicopter accidents is fundamental to the development of effective mitigation measures. However, there are significant weaknesses in the current methods. This paper proposes a new structured framework for the analysis of accidents in helicopter operations, which includes the determination of the appropriate timeframe for analysis, establishment of terminological definitions, identification of the relevant accident variables and data sources, and the execution of a three-fold statistical analysis strategy. It is implemented in the context of worldwide offshore helicopter operations, enabling the identification of a number of areas for priority intervention including the need for a global harmonisation of accident data capture and dissemination, causal factor-based metrics for the calculation of accident rates, better pilot support for night-time and arrival operations, and a new safety paradigm to mitigate rare accidents. The framework is recommended for the analysis of helicopter accidents and incidents to support safety improvement.K E Y WO R D S 1. Accident analysis.2. Helicopter safety. 3. Predictive safety.

Several hazardous industries have embraced Heinrich's premise that incidents of negligible safety consequences are precursors to accidents in a statistical sense. However, in few such industries has research verified the truth of that assumption. This paper explores the relationships between accidents and reported incidents in the context of oil and gas–related offshore helicopter operations by using the accident investigation reports published by the UK Air Accident Investigation Branch and the incidents filed under the British Mandatory Occurrence Reporting Scheme between 1997 and 2010. Classification frameworks were developed to enable the independent analysis of accidents and incidents in relation to specific variables of interest. Frequencies and statistical associations that could have indicated the precursor relationship were explored. From the results of the analysis, the paper highlights potentially severe shortcomings in the assumptions underpinning incident data collection as well as the process with which incident data are generated. For example, the paper unveils the existence of sudden failures that cannot be reliably anticipated or reported and draws attention to a potentially flawed incident-reporting culture. Given the results, the paper informs stakeholders in the industry of specific initiatives to ensure that the right lessons are learned from past occurrences (e.g., through ways of collecting incident data that will not rely solely on reporters) and how these could be used to inform future interventions, for example, through the analysis of potential consequences of incidents, as a complement to the analysis of frequencies.

No abstract