Best practices for modeling structural boundary conditions due to a localized fire

Recent studies suggest the assumption of uniform heating that is used in current structural fire design cannot be assumed conservative, especially if the fire is expected to burn locally. Aside from design equations, which have limited applicability, a common approach to simulating structural members subjected to a localized fire is modeling the fire-structure interaction using a coupled computational fluid dynamics (CFD)-finite element (FE) model. In the existing literature, a wide range of methods and parameters are used when determining the boundary conditions at the fire-structure interface, specifically regarding the representation of net heat flux, heat transfer coefficient, and surface emissivity of steel. The purpose of this study is to investigate various methods for representing the boundary conditions in terms of accuracy and computational efficiency and then identify best practices. In conclusion, our study found that net heat flux predicted by adiabatic surface temperature, a nonconstant heat transfer coefficient, and a surface emissivity of 0.9 for steel was the most reliable thermal boundary condition in a coupled CFD-FE model of a localized fire. These recommendations are based on the two cases studied here, and caution should be used when applying these results to future studies.