Search

This thesis concerns the dynamic behaviour of railway catenary systems, particularly Norwegian systems. The aim of this PhD study is to increase the knowledge of the dynamic behaviour of these systems. This includes analyses of field measurements sampled during daily train operation and parameter studies by developing and using a detailed and validated numerical model. A range of field measurements was obtained and analysed during the study. These include displacements, accelerations, rotational velocities, contact forces and geometries. A new and purposespecific wireless monitoring system designed for field measurements was developed during the thesis and was used for sampling accelerations and rotational velocities. The monitoring system consists of up to ten wireless sensors and one master unit. The sensors can be mounted arbitrarily over a range of 1400 m. Triggering is mainly accomplished by passing trains, but manual triggering is made possible. Time synchronization of the sensors is ensured by a developed scheme. Furthermore, closerange photogrammetry and integrated accelerations were used to estimate and verify the displacement uplift time series. The analysed recorded contact forces were obtained from a database constructed by data from an overhead line recording locomotive. In addition, catenary geometry was measured by laser. The study develops, presents and uses different methods for assessing the dynamic behaviour of railway catenary systems by evaluating both numerical simulations and field measurements. The methods include the use of power spectral density (PSD), short-time Fourier transforms (STFT), covariance-driven stochastic subspace method (Cov-SSI), histograms, cross-correlation and extreme value probability distributions. The thesis demonstrates that the dynamic behaviour of railway catenary systems has a high grade of variability. This includes high variability in the frequency content that is shown to be substantially dependent on the position in a span, the properties of the span, the properties of each catenary section and the loading. This demonstrates the importance of establishing system frequencies and positionspecific frequencies. The variability of the loading depends on the train speed, static uplift, type of pantograph and wear on pan heads. This makes it favourable to perform both overall analyses, including many train passages, and analyses of single train passages to obtain the full picture. In addition, the importance of the variability in the response at one point in the catenary during a train passage is demonstrated and requires that the analysis should be performed both for segments of the time series and for the whole time series. The study clearly suggests distinguishing pre- and post passage parts based on their difference in frequency content. In addition, clear differences found in the lateral and vertical response indicate that both should be included in a frequency investigation to improve interpretations. The dynamic response was demonstrated to be a good tool for parameter study. It was used to evaluate changes in tension forces, the effect of rapid changes in contact wire height and differences depending on curvature. A damping study including analysis of data from three different catenary sections and several train passages resulted in proposed Rayleigh damping coefficients for numerical models. Above all, this study highlights the complexity of the pantograph–catenary interaction.

The dynamic behaviour of the railway catenary system is of great importance as the speed of the train increases. The aim for this study was to make a numerical model of the catenary system in three dimensions to be able to investigate what should be done with existing catenary systems in Norway to increase the train speed. In Norway there is a high ratio of curves with small radius, thus it is very important to include three dimensions. Abaqus has been used to obtain the numerical model. Both static and dynamic analyses have been done on this model. The results from the analyses have been compared with measurements, and this has shown that the model is able to describe the dynamic behaviour of the catenary system with sufficient accuracy. In future work it is recommended to look especially into how the contact interaction between the contact wire and the pantograph is modelled.

Populariteten til tverrspente limtrebuer har økt den siste tiden grunnet rask byggetid og god tilpasningsevne. Treverk er betraktet som et karbonnøytralt byggemateriale og dermed ansett som et sentralt byggemateriale for å oppnå Norges klimamål om å bli karbonnøytralt innen 2030. Designkriteriet for disse lette, slanke konstruksjonene er hovedsakelig begrenset av bruksgrensetilstanden. Stivheten til ikke-strukturelle komponenter er som regel neglisjert i dynamiske analyser. Det har blitt vist at asfaltdekket påvirker de dynamiske egenskapene til tverrspente limtrebruer. Det er dermed essensielt å nøyaktig forutse asfaltens bidrag for de dynamiske egenskapene for ulike temperaturer. Denne masteren analyserte akselerasjonsdata fra to ulike tverrspente limtrebruer som befinner seg i Raufoss og Brumunddal. Bruene ble eksitert ved bruk av hopptester og modalhammertester. Tidsseriene ble analysert ved bruk av BFD, SSI-Cov og SSI-DD. SSI-metodene er egentlig kun egnet for å analysere hvitt støy, men metodene greide å estimere de dynamiske egenskapene fra modalhammertestene. SSI-Cov ga bedre estimater for egenfrekvenser og dempingsforhold, men metoden strevde med energidissipasjonen mellom modalformene til nærliggende moder. En Python-kode som modellerte, kjørte og hentet ut modalformer og egenfrekvenser fra de modellerte tverrspente limtrebruene i ABAQUS ble implementert. ABAQUS-modellen estimerte det dynamiske bidraget til asfalten som en funksjon av temperatur og lastfrekvens. Stivheten forårsaket av det kontinuerlige asfaltlaget ble modellert som en rotasjonsstivhet. Rotasjonsstivheten ble optimert for å minimere feilen til bruenes første egenfrekvenser. Modellen ga gode estimater for egenfrekvensene samt gode estimater for modalformer. Modalformene ble nesten ikke påvirket av rotasjonsstivheten ettersom opplagerkantene ikke deformerte seg lineært. Lineær deformasjon av opplagerkantene var forventet. Stress-laminated Timber bridges have recently grown in popularity due to their speed of manufacture and great adaptability. Timber is considered carbon-neutral and is therefore considered an essential building material in Norway’s mission to achieve carbon-neutrality by 2030. Design criteria of these lightweight, slender structures are mainly governed by serviceability criteria. Stiffnesses from non-structural components are typically neglected during dynamic analysis. The asphalt pavement has been proven to greatly influence the dynamic properties of the Stress-laminated Timber bridges. Accurately predicting the asphalt’s contribution for different temperatures is therefore essential to accurately predict the dynamic properties of the bridges. This thesis analyzed acceleration data from two stress-laminated timber bridges located in Raufoss and Brumunddal, respectively. The bridges were excited by Jumping Tests and Modal Hammer Tests. These data sets were analyzed using BFD, SSI-Cov, and SSI-DD. The SSI methods are only meant to analyze white-noise data. However, the methods accurately managed to predict the dynamic properties from the Modal Hammer Test. SSI-Cov yielded superior estimates for natural frequencies and damping ratios but struggled with the energy dissipation between mode shapes for closely spaced modes. A Python script that modeled, ran, and extracted estimated mode shapes and natural frequencies of Stress-laminated Timber bridges in ABAQUS was implemented. The material properties of the asphalt pavement were estimated as a function of temperature and loading frequency. The stiffness caused by the continuous layer of asphalt, was modelled as a rotational stiffness in ABAQUS. The rotational stiffness was optimized to minimize the error for the bridges’ first natural frequency. The results yielded accurate predictions of natural frequencies and mode shapes. The mode shapes were, however, hardly affected by the rotational stiffness. This was caused by the boundary edges of the ABAQUS model being distorted.

The thesis involves development of code for simulating the contact interaction between pantographs and catenary systems through the use of FEA software. The resulting code is then used to investigate how certain model features; the cant height and overlapping spans, affect the dynamic response.

The main objective of this thesis is to measure and determine the dynamic properties of a measurement pantograph, and then create a numerical model with the same properties. The purpose of the model is to give a better representation of the actual behavior of the pantograph compared to the previously used lumped mass models. The thesis presents the process from very simple operation of the pantograph and contact force measurements, to more advanced studies of the attachment of the accelerometers and the development of test setups and procedures to measure the dynamic properties. The measurements done on the dynamic properties are then analyzed using scripts in MATLAB and the modal analysis software ARTeMIS. Using the results from the dynamic analysis, a numerical 3D model is created in Abaqus CAE with the same dynamic behavior as the pantograph in the lab.