Experimental study and machine learning on the maximum temperature beneath tunnel ceiling induced by adjacent tandem fires in longitudinally ventilated tunnel

Abstract

This study experimentally investigated the effects of longitudinal ventilation on the maximum temperature rise under tunnel ceilings induced by adjacent tandem fires. Because the tunnel is a narrow building structure, the fire source may ignite the surrounding flammable materials (such as adjacent vehicles, etc.), causing the simultaneous burning of multiple fires. The experiments were conducted in a 1:8 reduced-size tunnel. The ceiling temperature were measured for 60 repeatable test conditions, including five separation distances between adjacent fire sources at various heat release rates with different longitudinal wind speeds. As regards the effect of fire separation distance, the maximum ceiling temperature rise was recorded and measured for five different fire separation distances; as the fire separation distance increased to 0.9 m for a given test condition, the maximum ceiling temperature rise gradually decreased. Meanwhile, the maximum ceiling temperature rise decreased as the crosswinds increased. A new correlation was proposed between the maximum ceiling temperature rise as a function of the change in fire plume due to the fire separation distance and the Froude number representing flame tilting caused by crosswinds. Meanwhile, machine learning algorithm is used to predict the fire heat release rate. The model was shown to be able to effectively explain the experimental results and can provide a reference for examining the evolution of tunnel fire situations.