The paper presents the development of a three-dimensional (3D) beam element for the analysis of steel structures in fire that properly accounts for the degradation of the torsional strength and stiffness owing to thermal exposure. The element is well-suited for the analysis of members that are subjected to significant torsional effects, as steel members with open cross-sections subjected to torsion, lateral-torsional buckling, and torsional buckling. The element is based on a corotational formulation and small strains assumption, while the local formulation was developed according to both the Timoshenko and Bernoulli beam theories. The stress-strain relationship and thermal expansion of steel were implemented according to Eurocode EN 1993-1-2 (CEN 2005). A comprehensive numerical analysis was performed and results for several case studies are presented in the paper to assess the performance of the 3D beam element. The outcomes indicated very good agreement when compared with shell-based models, highlighting promising capabilities of modeling steel structures in fire. Moreover, the proposed element showed more accurate results with respect to beam elements included into two commercial software, i.e., ABAQUS and SAFIR.

In this paper, fire safety engineering principles were applied to a steel-framed office building. The case study consisted of a 15-storey steel moment-resisting frame designed in Japan. Once the performance criteria had been defined, two severe fire scenarios for the unprotected structure, implying a different degree of building collapse, were identified and modelled in zone models, OZone and consolidated fire and smoke transport (CFast), as well as in the computational fluid dynamics software Fire Dynamics Simulator (FDS). Based on the results of the fire development simulation, several finite-element thermo-mechanical analyses were performed with SAFIR. OZone and CFast models, which are much less computationally demanding, provided comparable failure mode and time with respect to FDS. Since the steel frame was seismically designed as a moment-resisting frame in the two main horizontal building directions, the columns were particularly stocky. Moreover, they were only partially heated because they were located on the compartment edges. For these reasons, columns did not exhibit failure, in contrast to the assumptions taken in the risk-ranking process in relation to the evaluation of the consequences, suggesting a revision of such estimations and a possible iterative procedure for the definition of critical fire scenarios for the structure.