Machine learning approach for stress analyses of steel members affected by elastic shear lag

In the design of steel members based on beam theory, normal stresses in cross sections are generally determined neglecting shear strain influences (shear lag). However, in cross sections with wide flanges compared to member lengths, shear lag can significantly influence normal stress states. To capture corresponding effects in practical applications, this paper presents a novel stress calculation approach based on machine learning (ML). For implementing the approach, neural networks are employed as supervised ML algorithm using training data generated by finite element calculations (shell models). The performance of the approach is validated using cross sections with wide flanges revealing high accuracy of the neural network. By subsequently interpreting the ML model, influences of different parameters, such as cross section parameters, on the normal stress distributions are quantified, providing deeper understanding of the mechanical problem solved by ML.