Expanding the FDS Simulation Capabilities to Fire Tunnel Scenarios Through a Novel Multi-scale Model

Abstract

AbstractComputational Fluid Dynamics (CFD) is widely used to simulate tunnels and partially substitute on-site tests. As technology advances, new application opportunities appear; some examples are the optimal operation of ventilation and emergency systems, risk assessment of tunnels and training of the operators. Even when the computational capacity of computers has grown, CFD is still constrained by the large amount of computational resources needed in long tunnels. This introduces a need for methods able to reduce the amount of time required for simulations. To face this need, a novel 1D–3D multiscale model is presented in this paper. The model incorporates the code Whitesmoke into FDS (Fire Dynamics Simulator) through a direct coupling. Whitesmoke manages the fluid dynamics, temperature and concentration of species in the 1D portion, while FDS calculates these fields in the portion where fire occurs. Using this multiscale model, the computation time for long tunnels is reduced, proportionally to the 1D length in the domain. Also, additional simulation capabilities particularly useful for tunnel analysis are obtained. Some new characteristics are pressure boundary conditions can be easily imposed at the tunnel portals or at the ventilation shafts; the characteristic curves of the fans/jet-fans can be included, also considering the degradation effects due to smoke propagation; the piston effect can be properly considered. Our research verifies most of its capabilities, also clarifying its limitations and the criteria used to set the domain for the analysis. As a final step, the model is tested in a tunnel with a cross section of 4.8 m and 600 m of length with a 2 MW fire, comparing its performance with a full 3D FDS simulation. The difference in temperature and velocity is minimal for most of the domain, making It a good way to optimize resource usage in large simulations. Furthermore, the multiscale manages to reduce the computational time of more than a 50%.