ارزیابی قابلیت اطمینان مدل Eurocode برای سنجش خستگی پل های فولادی Reliability evaluation of the Eurocode model for fatigue assessment of steel bridges

1. Introduction Designing a steel bridge against fatigue damage requires estimations of the load effect and the resistance at critical details. The load effect needs to be estimated using a structural model and depends on the traffic intensity and the type of vehicles travelling over the bridge. In reality, a bridge is typically loaded by a range of different vehicles with varying axle configurations and gross weights which is not feasible to consider during the design phase. In practice, the design is performed using predefined simplified load models described in regulations. The Eurocode is the governing regulation for design of bridges in Europe. In part EN 1993-2 [1], a verification format considering fatigue is defined as cFfDsE2 ≤ DsC cMf (1) where DsE2 is a damage equivalent stress range and DsC is the fatigue strength, both valid for 2 million load cycles. The partial safety factors cFf and cMf represent the model uncertainties associated with the load model and the fatigue strength, respectively. The fatigue strength is a characteristic value based on fatigue tests under constant amplitude stress conditions. The damage equivalent stress range is determined as DsE2 = kV2Dsp (2) where k is a damage equivalent factor, V2 a damage equivalent impact factor, and Dsp the stress range for the load p. The stress range is defined as the difference between the maximum stress and the minimum stress caused by the load model evaluated using influence areas [1]. For road bridges, the load to be considered is the fatigue load model 3 (FLM3). The model based on damage equivalent factors was suggested already in [2] for railway bridges. The calibration considering traffic loads on road bridges is described in [3,4], and [5]. The characteristics of the FLM3 and the associated damage equivalent factors were calibrated using traffic measurements from Auxerre on the motorway A6 between Paris and Lyon in France [5]. In the background documents, statistical considerations regarding vehicle interaction were considered using stochastic queueing theory. The validation of the model was, however, performed in a deterministic manner against the traffic measurements from Auxerre. The accuracy of the verification model (1) and the associated FLM3 has been subject to discussions in, e.g., [6,7], and [8]. The model is suspected to render misleading results for regions with traffic intensities different from the reference location in Auxerre. And the model itself is claimed to be too simplistic to capture all possible cases of bridge geometry and traffic scenarios. In Sweden doubts about its validity have been raised mainly because now, fatigue more often becomes decisive in the design of new bridges than before when the superseded Swedish regulations were governing [9].