Real-time fire sensing based on spectral shifts in structure-transmitted alarm sound field

Abstract

The alarm systems, commonly installed in modern infrastructures, can emit sound during accidents but fail to provide real-time hazard information. This work repurposes these systems into a dual-purpose network that not only warns occupants but also enables real-time hazard assessment. A standard chamber fire is selected as the benchmark scenario. Since solid is more suitable for sound propagation, the floor structure is selected as an acoustic waveguide, where fire heating can alter its elastic modulus and Poisson's ratio and induce measurable sound shifts including increased pressure and decreased frequency. A theoretical model is developed to correlate both fire power and burning area with these acoustic parameters. Model coefficients are defined by fine element simulations using wooden fires and concrete floors. Experiments involving propanol pool fires and aluminium metal floors are further carried out for model validation. Results show that the sound pressure variation and frequency attenuation are greater if the fire burns with larger power and spread area. Results also show that a single microphone is sufficient to detect these shifts and accurately predict fires across various conditions (R2 ≈ 0.81). This paper provides a cost-effective method to use existing alarm systems to enhance situational awareness and safety in complex infrastructures.