Ultimate strength prediction of elliptical hollow section columns in compression and bending

In composite structures, tubular sections are widely utilized as structural members due to their effective load carrying capacities and visual aspects. The well-known tubular members include rectangular, square and circular hollow sections. Elliptical hollow section (EHS) is a relatively new shape of high strength, hot milled steel sections. In the recent years, it has been used in constructional building applications thanks to its intrinsic aesthetics as well as its constructional opportunities of sections having minor and major axis features. In this study, the prediction models for ultimate load carrying capacity (Nu) of EHS steel tube columns under the effect of compression and bending through popular soft computing methods named gene expression programming (GEP) and artificial neural networks (NN) were aimed to be derived. For this, experimental data from the literature were collected for training and testing the models. The column specimens were tested for flexural buckling in their minor and major axes considering with and without eccentricity. The parameters were selected as axis of buckling, eccentricity in y and z axes, larger and smaller outer diameters, wall thickness of the section, yield and tensile strength of steel, and member length. Explicit formulation of the Nu through GEP and NN were compared to the actual values to analyze the robustness and the repeatability of the proposed models.