Vibration and Impact Monitoring of Tram Operations
Lead Research Organisation:
University of Cambridge
Department Name: Engineering
Abstract
This project aims to develop a system for monitoring the condition of railway track using sensors mounted on in-service trains and trams, as part of the infrastructure and urban systems theme.
The surface roughness of railway track increases during service. Along with other forms of deterioration, such as settlement at rail joints, this leads to an increase in the level of vibration generated as a vehicle runs along the track. From its origin at the wheel-rail interface, vibration propagates through the track structure, through the ground and into nearby buildings, where it may then cause disturbance as perceptible vibration and/or re-radiated noise.
A significant portion of railways in the UK run through densely populated urban areas. The proximity to residential properties, and sensitive buildings such as concert halls and hospitals, means that groundborne vibration is a major concern of urban railway operators, who must work continuously at track maintenance to keep noise and vibration within acceptable levels. Currently, this work relies on manual track inspections, typically undertaken fortnightly, to identify areas of deterioration before they become a significant source of vibration. Such inspections are time consuming and costly. There is a clear need for a more efficient monitoring method.
This research aims to develop a track condition monitoring system that can be permanently installed on service rail vehicles running on an operational network. It builds on recent prior work that saw five axlebox accelerometers and associated data acquisition and positioning equipment fitted to one of Midland Metro's CAF Urbos 3 trams. This equipment has been collecting useful data over Midland Metro's rail network since 10th August 2016. Similar equipment may be fitted to one of London Underground Limited's trains in the near future to provide a second source of data for this research.
The main objective is to derive the wheel-rail interface force spectrum and the railhead's roughness spectrum from the measurements of vertical axlebox acceleration and vehicle location and speed, and relate these spectra to the amount of groundborne noise injected into the ground as the vehicle runs over the rail. This requires the entire mechanical system between the rail and the axlebox to be considered. Inherent challenges in the signal processing include separating out the vibration caused by wheel roughness, and the necessary correction for vehicle speed, both of which enable measurements from different passes over the same rail section to be compared directly.
The next objective is to create a virtual condition map of the network. The track will be divided into short segments over each of which the roughness spectrum is computed and presented. Peak and/or average levels at each track segment will be plotted over location for the entire network to give the operator an overview of the condition of the network. Further research is required into the most appropriate segment lengths that offer a useful compromise between wavelength resolution and spatial resolution along the track.
A networked system will be developed that automatically collects vibration data from the vehicles by means of wireless communication, and updates a virtual condition map in real time. Algorithms will be developed to identify vibration transients (caused by misaligned turnouts, crossings, poor rail joints etc.) in addition to exceedances in railhead roughness levels. It is anticipated that the levels of vibration will be monitored in order to schedule rail maintenance before these levels rise above a certain limit.
The surface roughness of railway track increases during service. Along with other forms of deterioration, such as settlement at rail joints, this leads to an increase in the level of vibration generated as a vehicle runs along the track. From its origin at the wheel-rail interface, vibration propagates through the track structure, through the ground and into nearby buildings, where it may then cause disturbance as perceptible vibration and/or re-radiated noise.
A significant portion of railways in the UK run through densely populated urban areas. The proximity to residential properties, and sensitive buildings such as concert halls and hospitals, means that groundborne vibration is a major concern of urban railway operators, who must work continuously at track maintenance to keep noise and vibration within acceptable levels. Currently, this work relies on manual track inspections, typically undertaken fortnightly, to identify areas of deterioration before they become a significant source of vibration. Such inspections are time consuming and costly. There is a clear need for a more efficient monitoring method.
This research aims to develop a track condition monitoring system that can be permanently installed on service rail vehicles running on an operational network. It builds on recent prior work that saw five axlebox accelerometers and associated data acquisition and positioning equipment fitted to one of Midland Metro's CAF Urbos 3 trams. This equipment has been collecting useful data over Midland Metro's rail network since 10th August 2016. Similar equipment may be fitted to one of London Underground Limited's trains in the near future to provide a second source of data for this research.
The main objective is to derive the wheel-rail interface force spectrum and the railhead's roughness spectrum from the measurements of vertical axlebox acceleration and vehicle location and speed, and relate these spectra to the amount of groundborne noise injected into the ground as the vehicle runs over the rail. This requires the entire mechanical system between the rail and the axlebox to be considered. Inherent challenges in the signal processing include separating out the vibration caused by wheel roughness, and the necessary correction for vehicle speed, both of which enable measurements from different passes over the same rail section to be compared directly.
The next objective is to create a virtual condition map of the network. The track will be divided into short segments over each of which the roughness spectrum is computed and presented. Peak and/or average levels at each track segment will be plotted over location for the entire network to give the operator an overview of the condition of the network. Further research is required into the most appropriate segment lengths that offer a useful compromise between wavelength resolution and spatial resolution along the track.
A networked system will be developed that automatically collects vibration data from the vehicles by means of wireless communication, and updates a virtual condition map in real time. Algorithms will be developed to identify vibration transients (caused by misaligned turnouts, crossings, poor rail joints etc.) in addition to exceedances in railhead roughness levels. It is anticipated that the levels of vibration will be monitored in order to schedule rail maintenance before these levels rise above a certain limit.
Organisations
Studentship Projects
| Project Reference | Relationship | Related To | Start | End | Student Name |
|---|---|---|---|---|---|
| EP/N509620/1 | 01/10/2016 | 30/09/2022 | |||
| 1946489 | Studentship | EP/N509620/1 | 01/10/2017 | 30/09/2022 | Tobias Carrigan Donfrancesco |
| Description | The wavelength spectrum of railhead roughness on railway track as per EN 15610 can be derived from axle-box acceleration measured on passenger trains in revenue service. The effect of any wheel roughness is removed using a comb filter. It is important for the dynamic properties of the train and track to be correctly determined for the derivation of roughness to be accurate: the train's properties may be calibrated separately from those of the track using conventional rail roughness measurements, which enables the track's properties to be updated for different sections of track without the need for recalibration. Furthermore, the track stiffness may be estimated from the same axle-box acceleration measurement. The present techniques derive the rail roughness accurate to 5 dB at wavelengths above 20-30 mm even with variations in track stiffness. The conditions and assumptions for this are: * The track is underground and consists of rails directly fastened to the concrete tunnel invert (data from ballasted track is currently being examined); * The rail type (e.g. UIC-54) and the spacing between wheelsets is known; * Either the speed of the wheelset fitted with axle-box accelerometers is measured precisely, to enable a comb filter to be used to remove wheel roughness from the measurement, or the wheel roughness is less than the rail roughness; * Axle bearing roughness is lower than rail roughness (this is met in the majority if not all cases). "Axle-box acceleration" is measured using accelerometers fitted to the railway vehicle's axle boxes (or a non-moving part of the bogie closest to the axle bearing). |
| Exploitation Route | Some railway operators, including the London Underground, are investigating the use of axle-box accelerometers to monitor rail roughness on their rail networks, but without the means to derive roughness from axle-box acceleration to a consistent accuracy. The software-based signal processing techniques developed to accurately derive spectra of rail roughness from axle-box acceleration can be deployed in-situ to this effect, enabling continuous monitoring of rail roughness and potentially reducing the need to take direct measurements of rail roughness. |
| Sectors | Transport |
| Description | Pecha Kucha presentation (Cambridge Science Festival) |
| Form Of Engagement Activity | A talk or presentation |
| Part Of Official Scheme? | No |
| Geographic Reach | Local |
| Primary Audience | Public/other audiences |
| Results and Impact | I presented a Pecha-Kucha-style presentation about rail roughness ('roaring rails') and the project in two workshops during the Cambridge University Science Festival event, drawing interest from the general public. |
| Year(s) Of Engagement Activity | 2018 |