Ignition ability prediction model of biomass fuel by arc beads using logistic regression

Wildland–urban interface fires are a severe fire hazard due to biomass ignition caused by arc beads. This study investigated how the energy, diameter, and number of arc beads affect biomass ignition probabilities. An improved experimental method was used to generate arc beads of various arc energies. Here, α-cellulose materials, which are well-characterised as biomass, were used as fuels. A high-speed camera recorded ignition phenomenology, revealing two ignition behaviours of rolling and embedding. The results revealed that an electrical fault arc energy of approximately 175 J was the most dangerous ignition condition. Ignition phenomenology was categorised into ignition and non-ignition, and it was observed that ignition could only occur during the bead rolling process. Contrarily, non-ignition occurred when its bounce even spun plenty of times. Ignition limits, namely the ignition region, potential ignition region, and non-ignition region, were determined. Furthermore, a novel predictive logistic regression-based ignition probability model was established, which indicated that the ignition occurrence of an arc bead was highly dependent on the diameter of the arc bead. The developed mathematical model can reasonably predict the ignition ability of biomass fuel ignition by arc beads.