Stochastic analysis of semi-rigid steel frames using a refined plastic-hinge model and Latin hypercube sampling

Abstract

This paper proposes a stochastic analysis model that combines the refined plastic-hinge analysis method with the Monte Carlo simulation to predict the realistic resistance of steel structures. The model treats the uncertainty quantifications of the material properties, geometry, and semi-rigid connections as independent random variables in the discretized stochastic fields using Latin hypercube sampling. The critical resistance of the structures in each simulation is determined using the general displacement control method and advanced analysis algorithms. The statistical approach is employed to compute the mean values and the standard deviation of the load-carrying capacity of the steel frames. The study finds that the uncertainties in material properties are the most sensitive, and the correlation of these parameters with the strength of structures is also significant. To improve the load and resistance factor design criterion, a resistance factor of 0.93 is suggested for the beam-columns in the frames under axial force, torsion, and biaxial bending moments. A stochastic response spectrum is proposed to define the boundary of strength, where a clearly defined plastic mechanism is attained. Furthermore, this study provides valuable insights into the stochastic resistance of steel frames, which are essential for practical engineering applications.