Conference Agenda

In railway bridge dynamics, the accurate prediction of deck accelerations is essential to ensure ballast stability, track quality, and the overall structural integrity of the bridge. Adequate modelling of the track, bridge, and subsoil is often a difficult challenge due to the conflicting requirements of accuracy and efficiency of the chosen modeling approach. Therefore, the modeling approaches found in the literature often focus either on efficiency by making simplifying assumptions about the foundation, the subsoil, the track and the bridge, or on accuracy by modeling of the geometry and material of the structural components and the subsoil in detail. The objective of this contribution is to assess the reliability of an efficient modelling strategy that includes the track, bridge, and foundation on the subsoil in the form of a simple two-dimensional lumped parameter model in comparison to a sophisticated, fully three-dimensional transient coupled finite element – boundary element approach. The fundamental modal properties of both models and their respective subsystems are evaluated and used to optimize the simplified approach. Special emphasis is placed on the accurate consideration of the dissipative effect of propagating waves in the infinite domain of the soil. Final conclusions regarding the applicability as well as the limitations of the simplified approach are drawn from the comparison of the acceleration response predictions due to the passage of a train for both models.

Fatigue damage can be evaluated in both the time and frequency domains. While time-domain methods are more accurate than frequency-domain methods, they are also more computationally expensive, particularly for long input signals. Frequency-domain methods, on the other hand, are more computationally efficient, but require stationary and Gaussian input signals. To apply the frequency-domain method to estimate fatigue in steel bridges, this assumption must be satisfied. However, the input vibrations derived from vehicle loads do not meet this requirement. As a result, the fatigue life of the main girders obtained from the frequency method was only about a thousandth of that from the time-domain method. This research aimed to reduce the discrepancy in estimated fatigue life between the time and frequency methods.

Although the previous study used 166 cases for the stress definition, modal transformation and reduction technique contributed to use 20 cases, which made the stress definition more accurate. Furthermore, the grouping and compression of input vibrations significantly contributed to minimizing the discrepancies. The grouping was based on the vehicle weights, while the compression strategy involved shortening the intervals between two vehicles to the longer passing time of the two, but only if the original intervals were longer than the passing times. As a result of these methodological improvements, the error margin in the fatigue life estimation was reduced to 10%.

In addition, this research demonstrates that the frequency-domain method significantly shortens calculation time. Remarkably, the method allows for the fatigue estimation of a 10-day input signal to be completed for about 8.6 hours, which is nearly the same as for a 3.5-hour input. It is important to highlight that this calculation efficiency is notably greater than expected in the time domain, where the estimated calculation time is around 4,100 hours.

A high-speed train passing a bridge structure at a critical speed can induce large bridge vibrations. In particular, the acceleration of the bridge deck is a critical design parameter because exceeding the normative limit can cause ballast instability resulting in track misalignments. Therefore, it is of utmost importance to realistically predict the acceleration response of the bridge in the design process based on a sufficiently accurate yet computationally efficient mechanical model. In engineering practice and research, a variety of modeling strategies exist with varying degrees of sophistication. For the structure, these can range from simply supported Euler-Bernoulli beam models that capture the bridge to two beams representing the structure and the track separately that are coupled by vertical and horizontal spring-damper elements, to sophisticated 3D finite element models that capture the components of the bridge and the track (and possibly the subsoil) in detail. The train modeling strategy ranges from single loads to detailed train models with multiple DOFs. Many bridges are single-span structures, but especially high-speed railway bridges span large valleys and thus consist of several single-span structures of the same length, weakly coupled by the superstructure of ballast and rail. The objective of this study is to evaluate the effect of different modelling strategies on the numerical prediction of the dynamic response of such weakly coupled ballasted bridge structures. In particular, different beam models with and without soil-structure-track interaction as well as with and without consideration of the weak coupling between the adjacent bridge structures are considered. The results of deterministic (semi-probabilistic) analyses and probabilistic failure assessments are performed and compared for the different models.

With an increasing number of bridges being decades old since their construction, it is crucial to assess the residual seismic performance of existing bridges, taking into account changes in structural parameters due to ageing, to ensure the resilience and robustness of road networks. To this end, we investigate the approximate Bayesian computation (ABC) model updating using seismic response data, in which a well-known sequential Monte Carlo sampler, called transitional Markov chain Monte Carlo (TMCMC) is employed to gradually minimize the Euclidian distance between the simulated acceleration response time histories and the corresponding observed acceleration time histories. The target system is a seismically isolated bridge pier with deterioration of both the rubber bearing and the reinforced concrete pier. Firstly, the effects of different input ground motions and different noise-to-signal ratios on the ABC model updating results are investigated. Furthermore, the nonlinear dynamic analysis is performed using the updated bridge pier model against different input ground motions to demonstrate the effectiveness of the ABC model updating for estimating the residual seismic performance considering the ageing deterioration.

Bridges play a pivotal role in railway transportation infrastructure. However, they are susceptible to the effects of time and wear. Concurrently, the ever-increasing demand for mobility requires higher travel speeds and imposes heavier loads, further challenging their integrity and safety, thus necessitating diligent monitoring to ensure railway network safety. Traditional Structural Health Monitoring (SHM) techniques, employing fixed sensors on bridges, offer reliable data collection but are limited in their ability to comprehensively inspect numerous bridges within a network. An approach utilizing mobile vibration-based monitoring, employing sensors on passing trains, presents a promising alternative. By operating such trains at regular intervals, continuous data collection becomes feasible, providing crucial insights into bridge deterioration over prolonged observation periods. To this end, this work proposes a model-based methodology to extract modal parameters of bridges based on acceleration data collected by traversing trains, focusing on a truss bridge located on the Loetschberg axis of the Swiss Federal Railways (SBB) network. The Loetschberg axis serves as a key north-south route through the Alps, holding significance for Switzerland and Europe as a whole. Leveraging acceleration data collected from the diagnostic vehicle (gDfZ) of SBB and the available multibody vehicle model of the diagnostic vehicle, the proposed approach utilizes a model-based Kalman filtering approach for state and input estimation coupled with a subspace identification method to determine bridge frequencies. The aim is to monitor changes in the bridge’s modal frequencies in time, enabling assessment of its remaining lifespan and prompt repairs in case of damage. By ensuring the safety and reliability of this critical railway infrastructure, we aspire to contribute to the seamless operation of rail transportation within Switzerland and across Europe.

A realistic and economical dynamic assessment of railway bridges requires, first and foremost, input parameters that correspond to reality. In this context, the applied damping properties of the structure have a decisive influence on the results in the prediction of resonance effects and further in the assessment of the compatibility between rolling stock and railway bridges. Concerning the damping factors used in dynamic calculations, the standard prescribes damping factors depending on the type of structure and the span. However, these factors can be regarded as very conservative values which do not represent reality. As a result, in-situ measurements on the structure are often necessary to classify a bridge categorized as critical in prior dynamic calculations as non-critical. Regarding in-situ tests, a measurement-based determination of the damping factor is inevitably accompanied by a scattering of the generated results due to the measurement method used and also as a result of the individual scope of action of the person evaluating the test and this person's interpretation of the measurement data.

With this background, this contribution presents new methods and analysis tools for determining the damping factor, intending to reduce the scatter of the results and limiting the scope of action of the person evaluating the test. Methods and analysis tools are discussed for methods in the time and frequency domain. Based on in-situ tests on 15 existing railway bridges, the data-based procedure for determining the damping factor is explained, and the methods are compared in the time and frequency domain. It is shown that a clearly defined evaluation algorithm can significantly reduce the scattering of results. Furthermore, it is shown that the excitation method substantially influences the determined damping factors.