A numerical study of gas explosion with progressive venting in a utility tunnel

Abstract

A numerical model of a progressive vented gas explosion is presented. A CFD tool in combination with correlation analysis and an artificial neural network (ANN) were utilized to establish and refine the numerical model. The experimental results of 44 fixed vented gas explosions and one progressive vented gas explosion with moving obstacles were used to validate the numerical accuracy. The results indicated that the method to estimate the activation pressure of the pressure relief panels for a fixed vented gas explosion achieved a lower overpressure prediction compared to that for a progressive vented gas explosion. The progressive venting procedure was modelled by two-layer pressure relief panels with the upper layer having activation pressures with a linear ascent trend. The vents on the tunnel had an insignificant impact on the explosion load after being lifted over the tunnel top, and their falling process was unnecessary to be modelled. A non-negligible impact of the obstacles inside the tunnel on the flow field upon being pushed away from their initial positions was demonstrated. By employing an ANN, the critical parameters in the numerical model were determined, which were used to accurately replicate the experimental results. The findings clarified a revenue for the modeling of a progressive vented gas explosion as well as some shortcomings of the CFD tool.