What the paper contributes
This first-author fire-safety milestone develops a transparent workflow for converting qualitative and semi-quantitative hydrogen accident records into screening-level physical descriptors. It deliberately distinguishes measured information from assumptions and reconstructed outputs.
Method
Four HIAD 2.1 tables were merged into a master dataset. First-order engineering relationships were used to estimate fuel inventory, heat release rate, flame-length scale, explosion-energy potential and notional blast indicators.
Main finding
The incidents span several orders of magnitude in reconstructed energy release. Direct correlations with fatalities and damage were weak, emphasising the role of exposure, mitigating factors, confinement, ignition timing and missing reporting fields.
Why it matters
The paper bridges empirical incident learning and physics-based interpretation without presenting reconstructed values as measured validation data. It also identifies what future hydrogen incident databases need to record.