Ductile fracture of steel structural members under combined shear and tension was observed in past strong earthquakes, but studies on ductile fracture under the combined stress state have been very limited in the field of steel structures. This study aims to experimentally and numerically investigate the coupling effect of shear stress and tensile stress on ductile fracture of both mild steel and high-strength steel. Two series of specimens are manufactured in which they are under pure shear and combined shear and tension. Meanwhile, a generalized simulation process, including generalization of the whole stress-strain data up to the fracture and fracture parameter calibration using only tensile coupons, is proposed to simulate ductile fracture of different structural steels. The generalized simulation process is proven to be capable of simulating ductile fracture under the combined stress state with good accuracy and simplicity.

To study extremely low-cycle fatigue performance of steel bridge structures, a total of nine steel bridge piers with unstiffened thick-walled box sections were tested in this study. The effects of various parameters, including cross-sectional shape (with or without an extension part), loading pattern, and width-thickness and slenderness ratios were investigated. Experimental results demonstrate that ductile cracks occurred at the column base in all specimens regardless of cyclic-loading histories and that ductile crack initiation life is sensitive to loading pattern, cross-sectional shape, and the width-thickness ratio. In all tests, the ductile crack began in the pier-to-base weld, and subsequent cyclic loads resulted in the ductile cracks growing at the weld toes along web and flange directions and then propagating into the base metal, which triggered an obvious decrease in strength capacity. A significant delay was observed between ductile crack initiation and maximum strength capacity. Moreove...

This article presents a damage evaluation method for predicting ductile crack initiation in steel bridge piers. In the proposed method, a damage index is introduced to evaluate ductile crack initiation. Seventeen different analytical models, including steel columns and beam-column connections in steel bridge pier frames, are created based on the shell element model, which takes local refined meshing into account. The plastic strain and damage index at strain concentration location are calculated. Moreover, effects of welded joint models, extension part, loading patterns and cross-section parameters are considered and discussed. Comparisons between experimental and analytical results show good prediction of ductile crack initiation. Meanwhile, the present method realizes prediction of ductile crack initiation at structural component level.

The strain-based prediction model combining the Miner's rule and Manson-Coffin's relationship provides a local parameter for evaluating the ductile crack initiation of steel structures, and some modified models based on it were proposed to evaluate extremely low-cycle fatigue (ELCF) behaviors of steel structures. Previous research has confirmed these local models to be an accurate index for ductile crack initiation in steel bridge piers, however it is found to quite depend on the mesh size of the numerical model used. In this study, a non local damage parameter is presented and successfully applied to ductile crack initiation life assessment of steel bridge piers subjected to earthquake-type cyclic loading. The non local damage parameter is based on averaging the strains over the effective plane using a weight function in the exponential form, and introduces the non local damage parameter to replace the local state variable. Finite element analysis with three different mesh sizes is employed. Comparisons of the local and non local solutions with those of experiments indicate that the non local prediction model can predict the ductile crack initiation of steel bridge piers with good accuracy regardless of the specimen geometries and loading histories, meanwhile the mesh independent nature of the non local model is also demonstrated.