Fatigue loads identification on orthotropic steel decks - Learning from strain measurements in practice

Initiation and propagation of fatigue cracks under heavy traffic loading is one of the most frequent causes of damage for road steel bridges. A particular case where fatigue phenomena have often been observed is the orthotropic steel deck (OSD). Due to its intricate layout with plated structural components in different planes and the inherent vulnerability of the employed welded connections to fatigue, bridges with this type of deck require accurate fatigue verifications. In the process of fatigue assessment an important part of uncertainty is the fatigue loading, commonly expressed in the form of stress variation at each detail. This stress-time history is typically related to certain bands of amplitude of stress cycles and their corresponding number. Thus, acquisition of relevant information for these aspects becomes important. Since, as representative as possible fatigue assessment is sought, strain measurements are utilised, so that bridge-specific load characteristics can be obtained. It is therefore examined to which extent relevant fatigue load information can be obtained through these monitoring in OSDs. It is further of interest to which extent, based on the identified load attributes, the stress histories at other details can be simulated. For practical investigation of the foregoing a case study is used, where strain measurements at specific locations of an OSD have been conducted. An approach is formulated based on bridge weigh-in-motion (B-WIM); a technique to weigh vehicles as they pass above a bridge. The main principle of B-WIM is the comparison of the measured response with a corresponding theoretical one, based, in this case, on a FE model. The peculiarities of the OSD result in various parameters that can affect the stresses and should be allowed for in the modelled response. The employed approach is structured in such a way so that the influence of these parameters can be examined. Application of the approach results into identification of transverse position and wheel loads. Then, simulated stress derivation at strain gauge locations based on identified characteristics and certain assumptions is realised. Comparison of these stresses with the corresponding measured ones gives an insight on the accuracy of the whole task. Unfortunately, acquired information on loads, transverse positions, wheel types etc. cannot be directly verified due to absence of reference values. Inspection of the comparisons between modelled and measured stresses shows a relatively good agreement in the overall shape, which implies close approximation of the wheel load-transverse position combination. In the vicinity of the sensors located under the deck plate, however, the effect of additional parameters like the wheel footprint increases. A number of factors thus act in a combined way affecting the stress ranges and leading, locally, to higher discrepancies. The discussed influence is further verified by a sensitivity analysis. Generally, it can be concluded that information such wheel loads, approximate transverse position of the wheel and wheel type (single or double) can be obtained through appropriate processing of the measured strain signal. Further, verifications need to be made, including comparisons with reference values so that an insight into the accuracy can be obtained. For further derivation of accurate stresses at deck plate details based on identified attributes, greater attention must be paid to the expected wheelprint (depending on the wheel load) as well as the correct inclusion of the possible stress reduction effect caused by the asphalt wearing surface.Structural and Building EngineeringStructural EngineeringCivil Engineering and Geoscience