An-Najah Staff

This study evaluates the adequacy of different methodologies to predict the plastic capacity and response caused by non-uniform thermal gradients through the depth of beam–columns that are loaded only axially at the centroid. Three models with different levels of complexity were used to evaluate the fire response of beam–columns under non-uniform temperature gradients: (1) code-based equations; (2) a fiber-beam element model; and (3) a shell element model that discretizes the full cross section and length and is capable of capturing local (i.e. plate) instability. The code-based equations do not predict the response satisfactorily since these equations do not properly consider temperature gradients. The fiber-beam element and shell model results correlate well to the thermal and structural response of the beam–columns tested experimentally with varying parameters. If local buckling is not expected at ambient temperature, complex shell elements are not necessary when the failure mode is fully plastic and fiber-beam elements, which are simpler and less “computationally expensive” than shells, suffice. The experiments and models also validated equations that consider thermal gradients and predict the plastic capacity and structural response of these members, which includes a moment reversal due to a shift in the section center of stiffness with increasing temperatures.

Fire tests were performed to investigate the mechanics and capacity of steel beam–columns that develop a thermal gradient through their depth when exposed to fire. Wide-flanged specimens were loaded axially and tested vertically in a furnace recently commissioned at Michigan State University. The placement of insulation simulated a realistic three-sided heating scenario such as that experienced by a column on the perimeter of a building frame. Specimens were tested with several combinations of load level, fire scenario, and direction of the thermal gradient (which dictates the direction of bending). The different combinations of tested parameters had a significant influence on the fire response of these columns, which all failed by full section yielding due to a combination of axial load (P)(P) and moment (M)(M). These columns developed bending moments in response to through-depth thermal gradients as well as a moment reversal due to a shift in the section’s center of stiffness. The plastic resistance to combinations of axial load and moment was also affected by the thermal gradients such that the critical section, located in the hottest region along the column length, was where moment was the smallest (not the largest, as would be intuitively expected). The experiments and computer models showed good agreement with the predicted demands (i.e. bending moment reversal) and capacity (i.e. changes in the plastic P–MP–M capacity).