The nonsymmetric sections commonly found in open-web steel joists present difficulties when evaluating eccentric loads applied to a single chord angle, which result in additional bending moments and torsion that are not typically considered in standard design practice. This study presents an experimental and computational approach towards qualifying the response of a steel joist with nonsymmetric sections and eccentric loading. An experiment was conducted in which a joist was subjected to uniform top-chord loading with an eccentric hanging load applied to one chord angle, while displacements and strains along the bottom chord were recorded. A second-order structural analysis with a novel beam-element formulation that considers nonsymmetric sections was able to accurately estimate the deflections and rotations of the joist bottom chord. Several alternative computational models were created to determine the effects of common design assumptions on the analysis output. It was found that model variations including using a first-order analysis, simplifying the hanging load condition, and using doubly-symmetric section behavior only resulted in changes to the applied torsion and corresponding rotations, which led to worse predictions of stresses and deflections on the chord angle. Detailed measurements of each cross-section geometry in the joist were obtained using a handheld three-dimensional laser scanner. Despite the significant changes in torsional cross-section properties, the computational model with measured cross-section properties resulted in minimal variations compared to the model with nominal section properties. The work presented herein provides insight into the torsional response of steel joists with eccentric loads.

This paper demonstrates how a novel line element, which is available for use in MASTAN2, can account for nonsymmetric cross-section behavior resulting in a reasonable prediction of steel joists subjected to nonstandard loading, specifically eccentric loading to one bottom chord angle. The modeling approach was based on joist modeling recommendations from the Steel Joist Institute to align with existing design practices. The analysis results are shown to capture the movement of the overall joist as well as the twisting of the individual chord angle when eccentrically loaded. Therefore, the novel line element could be used to determine a conservative prediction for the maximum normal stress supported by a joist chord under torsion. Details are provided on how the stress values were determined as well as implications of modeling assumptions and analysis types through a series of alternative modeling and analysis approaches. These alternative analyses indicated that accurately defining the torsion on the cross-section is a key factor in properly distributing the supported moment between strong- and weak-axis bending. Future design provisions and recommendations will need to address the disconnect between the standard design assumption that angles have no warping stiffness and the fact that minimal warping stiffness causes significant theoretical local stress concentrations at connections and locations of applied loads.