Numerical and Experimental Evaluation on the Behaviour of Cold-Formed Steel Box Struts and Prediction of Experimental Results Using Artificial Neural Networks

Cold-formed Steel Square and rectangular hollow sections are superior to conventional structures in torsion, compression, bending, fatigue, tension and shear. Failure of such columns in compression is on account of buckling of the member which is dependent on the slenderness ratio. Rectangular and square hollow box struts of slenderness ratios 10, 15 and 20 with fixed end conditions are tested to failure. The parameters that are varied include flat width to thickness ratio and shape. Totally, fifteen specimens are tested to study the behaviour of the strut both experimentally and analytically using ANN approach. The change in the cross-sectional shape from a long rectangle to square does not significantly increase the ultimate load carrying capacity. The non-linearity in the load versus axial shortening behaviour was found to start at 90% of the ultimate load for almost all the specimens tested. The ultimate load carrying capacity were calculated using codal provision and also predicated by ANN which is very close to the experimental values of the specimen.