Reliability-Based Fracture Toughness Requirements for Steel Highway and Pedestrian Bridges

Abstract

Although the brittle fracture toughness requirements currently prescribed for Canadian steel highway bridges in CSA S6 appear to be meeting the needs of designers and users, these provisions exhibit shortcomings in terms of their lack of a probabilistic basis, in contrast with other failure modes, and their lack of flexibility, relative to international design provisions in Europe and the United States. A probabilistic basis of these provisions has only recently been established. The results of this effort highlighted a potential lack of conservatism for high yield strength plates and offered the potential to develop new design tools for explicitly considering important factors known to affect brittle fracture risk, such as plate thickness or demand-to-capacity ratio (DCR). At the same time, CSA S7—a new guideline recently developed for the design of pedestrian bridges—only considers a reduced strain rate in the relaxation of toughness requirements for Canadian steel pedestrian bridges when compared to those for highway bridges, and neglects the possible effects of differences in usage patterns, material properties, common weld details, susceptibility to fatigue, and typical governing limit state. In order to define improved brittle fracture provisions for highway and pedestrian bridges, address industry concerns about the scarcity of sufficiently tough hollow structural sections (HSS), and propose new design tool formats for selecting Charpy V-notch (CVN) test requirements, this project set out to describe a methodology for establishing reliability-informed fracture toughness requirements using Monte Carlo simulation, a fracture toughness master curve, finite element analysis (FEA), and new and existing highway traffic and pedestrian probabilistic live load models. The presented work includes improvements to a previously-developed probabilistic assessment tool and the verification of existing magnification factor solutions for common weld details, including transverse attachments and butt welds. Additionally, new magnification factors are determined for the K-joints in typical pedestrian bridges using the FEA software ABAQUS. A pedestrian live load model is defined based on similar concepts used to prescribe live loads in buildings. Even when making conservative assumptions about parameters, this model is found to reflect the low probability of the pedestrian live load exceeding the nominal load, let alone the factored design load. For a range of plate thicknesses, climates, target reliability indices, weld details, and live load models, simulations of different combinations of yield strength, DCR, and CVN test temperature are conducted using the assessment tool. Results are fitted to two-degree surfaces of test temperature as a function of yield strength and DCR. Three design tool formats are created using the results: 1) surface coefficient lookup tables; 2) contour plots; and 3) augmented design tables of CVN test requirements. The toughness requirements for highway bridges are found to be stricter as plate thickness, yield strength, and DCR increase. For most weld details, except for the most severe longitudinal attachment, it is shown that toughness requirements for pedestrian bridges can almost be neglected, particularly when considering the yield strengths and DCRs typical of pedestrian bridges. To support these conclusions, future research could focus on improving existing highway live load models with recent gross vehicle weight data and continue to establish realistic parameters for the developed pedestrian live load model. Additional work on finite element models would improve understanding of fracture toughness requirements for K-joints and other weld details fabricated with HSS members. Lastly, the integration of a fatigue crack growth model could be used to justify the use of lower target reliability indices by explicitly considering the effects of inspection and fatigue crack growth, allowing engineers to take advantage of the relaxation of requirements and higher calibrated crack depth demonstrated through the sensitivity studies in this project.