The effect of variabilities in the manufacturing tolerances on the ultimate flexural capacity of steel beams

Abstract

Steel fabrication involves the production of large, complex welded assemblies of rolled steel products, inherently leading to dimensional variations due to high-temperature processes used in manufacturing and joining. The purpose of this research is to examine the validity and safety of the tolerance tables established by the Eurocode (EN10034: 1993) in terms of the ultimate flexural capacity of the beam. Both geometric properties and material properties have been treated as random variables. With the Monte Carlo Simulation, 2000 realistic and randomised combinations were produced in order to compare the randomised moment capacity with the nominal plastic moment capacity. The analysis began with a comprehensive test in which all variables were randomised. This initial test provided the probability of failure for all beams under realistic conditions. Subsequent tests were conducted with certain variables held constant at their nominal values to assess the influence of each random variable on the plastic moment capacity. Further detailed analysis involved three purposely selected beams—specifically IPE80, IPE360, and IPE750x173—chosen for their varied section depths. The results from these tests were depicted in scatterplots, which facilitated the visual identification and understanding of any trends previously noted in the general analysis. Through the in-depth analysis, it was concluded that the flange thickness is the most critical dimensional property across all section sizes. Additionally, a highly important controlling factor is the variability of the yield stress. In fact, the in-depth analysis demonstrated that the effect of the remaining three-dimensional variables on the plastic moment capacity varies by section size. Specifically, the geometric properties, such as section height and width of section, have a diminishing influence as the section size increases, while the impact of web thickness on the probability of failure becomes more pronounced with larger section sizes.