Optimizing the spectral based fire detection strategy with high-fidelity radiation spectra

Accurate detection of an unwanted fire in its early stage is crucial in mitigating fire and preventing damages. The detection stratety must also avoid false alarms and the associated disruptions in workplaces. Thermal radiation -based detection is the fastest detection method, and commonly used in e.g. petroleum industry. The main challenge is to distinguish the radiation spectra of fires from those of other sources, e.g. hot objects or solar spectrum. The principles of radiation -based detection have been known for long, but open data and worked-out feasibility studies are rare. In this work, comparing the experimentally measured high resolution spectra of several fires with the numerically obtained line-by-line spectra of hot black body emitters, we present an optimized fire detection strategy and propose its implementation using gradient heat flux sensors with three optical filters with optimal cutoff limits.