How to use HIAD data
HIAD data collection is a service provided by JRC to safety experts, educators, students and technologists active in the field of hydrogen. Please consider that the events described in HIAD are not intended to serve as an instrument for passing judgement on individual companies or countries associated with an accident. A blame culture surrounding the database would greatly reduce the sharing of information. Hence, the HIAD Excel file provides only literature references, it does not provide the first or secondary data sources documents, which remain managed solely by JRC. JRC will provide these on an ad-hoc basis, following a request for specific research purposes and only on a sub-set of the total dataset.
A proper use of the data in HIAD should also consider the following caveats:
• Do not conclude from the data in HIAD that ‘hydrogen is not safe’. Every technology, once deployed, will always be affected by unplanned, unwanted events. HIAD is a tool developed to assist safe improvements of hydrogen technologies.
• HIAD events descriptors have been designed to draw a lesson learned and improve the safety of hydrogen technologies, not to compare the safety of hydrogen technologies with the safety of other technologies. Do not use HIAD data to answer the ill-posed question: ‘which technology is safer?’
• A collection of incidents may picture a rather negative image of a technology or process. In the cases collected in HIAD it is also possible to learn what went right (in risk mitigation, in emergency handling, in avoiding escalation). For inspiration, see the very enlightening paper of P.R. Amyotte, What went right, Process Safety and Environmental Protection, 135 (2020) 179-186, https://doi.org/10.1016/j.psep.2019.11.043
• The present structure of HIAD data is not done for deriving quantitative failure probabilities of specific components, because most of the event descriptions do not provide the details required and the statistical reliability enabling this type of analysis.
• Be very cautious before drawing general conclusions, because they could be biased by the types and availability of the primary sources used. For example, the historical and geographical distributions of the accidents are predominantly reporting European and North American events. Moreover, certain industrial sectors are more represented than others, because they are committed to investigate and publicly report their accidents, while others are not.
Quality of data
HIAD events are assessed, interpreted, reviewed and validated by JRC experts before entering the database. The overall quality of the data depends mainly on the quality and the level of details provided by their sources. To guide the users and allow a down-selection based on event descriptors quality, JRC provide a quality label for each event. The quality label scale is ranging from 1 to 5, with 5 being the highest quality label. Events with quality label from 2 to 5 are shared with users. Unvalidated events (quality label 1) are not made public. They wait for possible improvements based on new information becoming available, or eventually for cancellation.
For questions, updates, feedback and collaboration proposal, please write to JRC-PTT-H2SAFETY@ec.europa.eu
