Advanced Concepts for Condition Monitoring around the Wheel-Rail Interface

Investigators: Prof Roger Goodall, Dr Roger Dixon (Loughborough University); Dr Clive Roberts (University of Birmingham); Prof William Powrie (University of Southampton)
Researchers: Dr Guy Charles (Loughborough University); Dr Paul Weston, Mr Edward Stewart (University of Birmingham); Dr Jeffrey Priest (University of Southampton)

Background:
A feasibility assessment of innovative ideas in relation to condition monitoring of the wheel-rail interface was started at Loughborough within the first phase of RRUK, in particular the problem of real-time estimation of the wheel-rail profile. RRUK2 Project B8 has both continued this work and broadened the scope to include Birmingham and Southampton, firstly to investigate condition monitoring of switches and crossings, and secondly to link in some of the track monitoring concepts achieved in project A1.

The research on the condition monitoring of switches and crossings, which remains a significant proportion of the total cost of infrastructure, is aimed towards facilitating a move from calendar-based to condition-based maintenance will provide large benefits to operating costs.

The estimation of wheel-rail profile is a more speculative deliverable, but if real time profiles can be achieved then short-term variations will provide information of the condition of the rail head, long-term variations indicate wear of the wheel itself. The exact way in which the information would be utilised is to be determined as part of the research, but potential impact is large because this information is currently not available.

For railway track and the train, wheel irregularities, such as wheel flats, bad axles, are a major contributor to the performance of each. Current monitoring systems employed by NR are expensive and so only a small number exist at key sites. Therefore both freight and commuter trains on regional lines may never be assessed. This part of the project is therefore concerned with developing a monitoring technique for assessing irregularities in the wheel profiles of passing trains, which is both inexpensive and portable such that it can be utilised on any part of the rail network.

Original Project Scope and Objectives:
The overall aim is to employ advanced monitoring and processing concepts in order to provide the basis for a new generation of condition monitoring systems for railways principally focussed on vehicle-track interaction (VTI).

Specific objectives are as follows:

1. To establish methods and benefits of fusing the data obtained from vehicle and track instrumentation, with a view to providing more effective and robust indications of the condition of vehicles and track.
2. To develop concepts and techniques for providing meaningful monitoring of switches and crossings.
3. To research estimation techniques by which wheel-rail profile can be estimated while the vehicle is travelling along the track.
4. To develop inexpensive, portable track-based monitoring that can both broaden the applicability of monitoring around the railway network and provide the information related to objective 1.
5. To provide experimental validation of new techniques with respect to available track data.

Brief Summary of Research Methods:
A range of model-based and knowledge-based techniques are being used to process data gained from both vehicle and track measurements. A key feature is to understand in a general case the benefits gained by bringing together the data from the track and vehicle sensors, and this will be applied to both specific condition monitoring objectives (2 and 3).

The work on switches and crossings is extending some of the concepts developed as part of a previous DTp-funded project, as well as studying the incorporation of track-based sensors in a manner still to be determined.

The wheel-rail profile estimation is a scientifically-challenging problem that will start from a dynamic model as the main element to process the sensor data, but it is likely that prior knowledge of typical profiles will be built in to facilitate the non-linear estimation problem. Model-based condition monitoring approaches are central to the approach being followed, mainly concentrated on Extended Kalman Filters, but including work on Least Squares estimation methods.

The track monitoring aspects are focussed upon appraisal of wheel profile irregularities and their effect on track performance, in particular to assess the potential for ground based instrumentation, such as geophones, to quantify the degree of wheel irregularity It involves the development and implementation of track bed- based instrumentation to measure wheel irregularities, building on the work undertaken in project A1 and continuing in A4 with regard to use of ground based instrumentation. Analysis of data from site investigation is incorporated as part of the work, and correlation of such data with train mounted instrumentation will help differentiate the influence of train behaviour or track behaviour on sleeper velocities.

Work Done – including statement of key findings to date:
The following paragraphs describe achievements related to objectives 2, 3 and 4. A variety of opportunities are being followed with respect to experimental validation (objective 5).

Initial work used data collected from the Tyne and Wear metro to identify fey features of switches and crossing observable from in-service monitoring equipment. An improved six degree of freedom data acquisition system has been designed and constructed. This will be used during field tests with South Eastern trains scheduled for March 2008.

Initial feasibility studies have been completed aimed at estimating a nonlinear conicity function (in simulation results). These have applied Extended Kalman Filter and Least Squares Method approaches to the simulated vehicle response data and have highlighted key areas of difficulty in applying the Kalman Filter in higher complexity models.

Work done to date has focused on developing wireless data logging capabilities for the geophone system. The existing geophone measurement technique required hard wired connection between the various geophones. Due to NR policy this limited where the instrumentation could be used on the network. Lab trials have been completed using a wireless data logger provided by Microstrain Corporation, USA.

So far there has been limited progress in terms of understanding how integrating the vehicle and track measurement data (objective 1).

Work Remaining:
To date the work at Birmingham has mainly focussed on data collection. Further planned work includes analysis of data collected during tests, detailed measurement of test switches and crossings and development of an in-service test.

Continuation of the application of various condition monitoring methods, to include Particle Filters, multiple model Kalman Filter, Sliding Mode approaches. Identify the feasibility of a practical solution to estimation wheel-rail profile in real time, and establish a practical measurement set to perform this.

Work remaining includes field trial of the wireless data logger to ensure that electrical interference from the overhead lines, or third rail, does not occur. Once field trials are complete the instrumentation will be used to monitor sleeper velocities for a number train passages to obtain data. Subsequent analyses of the data will be conducted to assess the effectiveness of geophones in quantifying the magnitude of wheel irregularities. The monitoring will be conducted at various sites through out the UK.

A key outstanding item is to meet objective 1, i.e. to understand the value of bringing together track and vehicle measurement data. The effort in this respect is currently being applied to identifying testing opportunities when it will be feasible to gather coordinated data from the two measurement systems under development.<> Outputs:

• Poster from RRUK Workshop 2008
• Presentation from RRUK Workshop 2008
• A paper relating observable features of switches and crossings from in-service vehicles was presented at the International Conference on Railway Condition Monitoring which was held in Birmingham during November 2006.
• A paper relating to the work with South Eastern Trains will be presented at the International Conference on Railway Engineering (ICRE 2008) which will be held in Hong Kong at the end of March 2008.
• A paper relating to the work with South Eastern Trains will be presented at the World Congress on Railway Research (WCRR 2008) which will be held in Korea during May 2008.
• “Model-Based Condition Monitoring At The Wheel-Rail Interface”, International Association of Vehicle System Dynamics 20th Symposium (IAVSD 2007)
• “Wheel-Rail Profile Estimation”, International Conference on Railway Condition Monitoring 2006