Capabilities and Limitations of the Fire Dynamics Simulator in the Simulation of Tunnel Fires with a Multiscale Approach

Abstract

In tunnel fire safety engineering, long tunnels are generally simulated with one-dimensional fluid dynamics models (1D). Short tunnels can be simulated with a three-dimensional fluid dynamics model (3D). The 1D simulations are faster than 3D simulations, but they lack accuracy in the region where the flow is three dimensional, which is typically around the fire source. In this paper, we present a multiscale approach where large parts of the tunnel are simulated with the 1D approach, while a limited portion, close to the fire, is simulated with a 3D model. This allows to simulate with a good accuracy smoke dynamics and the induced flow field near the fire whilst large parts of the ventilation system and the tunnel's structure are simulated with 1D, reducing thus the computational costs. The present paper explores the capabilities and limitations of the Fire Dynamics Simulator, FDS 6.6.0, in the simulation of tunnels with a multiscale approach. FDS has an embedded 1D model connected with the hydrodynamics solver (3D) which allows to implement the multiscale approach without making a link to external models. In the first part of the paper, the fan-induced flow field is simulated with the 3D model in order to generate an operative map where the pressure rise induced by the fan is a function of its volume flow rate. The operative map is later implemented in the 1D model without the need to model explicitly jet fans in FDS. The second part of the paper addresses the modelling of a fire in a tunnel. The fire has two effects on the response of the ventilation system: (1) local pressure losses near the fire section and, (2) higher pressure losses along the tunnel due to the high temperature of the combustion products. Very promising results were obtained with the multiscale approach. Nevertheless, the results show also that, in the 3D portion of the tunnel (where the fire is located), large velocity fluctuations may lead to unphysical pressure distributions, especially for large fires (above 3 MW) and when a fine mesh is used. Therefore, further analysis and investigation is required in order to be able to deal with higher fire loads such as a 100 MW HGV fire or a 200 MW tanker fire.