Seeing that FDS is a fire-spreading and smoke transport simulation software widely used in scientific research and engineering projects, this paper has done investigations on how to utilize the case of Steckler room fire test and made some quantitative studies over the mesh independency property by using the relative similarity analysis method. Further discussions have also been made with the sensitivity of the simulation results to the mesh grid in the fire source area, buoyancy plume area and smoke layer area, with the impact of the computational domain on the simulation result properly interpreted. The correlation value of the simulation results tends to increase while defining the mesh grid, and finally falls into the interval between 0.99 and 1. The impact of the mesh grid size on the correlation value of the simulation is likely to be most severe in the fire source area, less in the buoyancy plume area and the least in the smoke layer area. The correlation value of the results gained from the computational domain, the same in value with the Steckler room and that from the extended domain, tends to be relatively low, whereas that between the results in the domain extended to the distinctive size should be rather high. Thus, it can be concluded that the correlation analysis can be expected to become a reasonable method to assess the mesh grid independency. However, the sensitivity to the mesh grid size tends to drop from the combustion region to the smoke layer area, which suggests the grid size shrinking in the fire area is an effective way to gain more accuracy in an acceptable executive time. In addition, the choice of the computational domain may have an evident effect on the air flow-in and flow-out in the room, which is likely to further impact the temperature and velocity distribution in the building. Therefore, reasonable extension of the computational domain has been proved to be necessary to acquire accurate results.
safety technology and engineeringFDSnumerical simulationmesh gridcomputational domain
NETL
NSTL
万方数据
维普
