Final Report Summary - MONITORAIL (Long range inspection and condition monitoring of rails using guided waves)
Executive summary
The aim of MONITORAIL is to develop a cost effective wireless condition monitoring system utilising long-range ultrasonic inspection and acoustic emission monitoring technologies. The system is to be used to monitor critical areas of rails where the probability of defects is high or there is limited access to carry out conventional inspection techniques. This will improve the European railway system for better efficiency and safety.
The project will reduce the substantial costs related to rail inspection, and contribute to achieving the target set by the European rail industry to reduce the overall maintenance expenditure by 30% by 2020. A significant increase in the availability of the network is anticipated owing to increased reliability of detection of potential failures.
The main technical objectives of the project are: Development of long-range ultrasonic and acoustic emission methods for accurate flaw detection in rails; Development of novel permanently mounted transducer arrays and electronic hardware for this monitoring task; Development of data analysis tools for automated defect recognition; Development of enabling technology for continuous structural health monitoring, including wireless communications and energy harvesting and storage methods to power the system.
Project context and objectives
To accomplish the project objectives, the work activities have been organised into a number of discrete Work Packages (WP). Those were divided into research (WPs 1 to 6), demonstration (WP7), exploitation and dissemination (WP8) and management (WP9).
The description for those work packages are shown below:
- WP1 Project specifications;
- WP2 Theoretical study and modelling;
- WP3 Development and evaluation of transducers for rail monitoring;
- WP4 Development of advanced signal processing and analysis tools and flaw sizing;
- WP5 Software and communication system development;
- WP6 System integration and testing;
- WP7 Demonstration and large-scale laboratory trials;
- WP8 Exploitation and dissemination;
- WP9 Project Management
Project results
The details of technical and scientific results for each work package are presented as follows:
WP1 Project specifications are 100% complete in month 5. The end-user, Network Rail, led this work package with the support of all the consortium members. Contributions from all partners involved in this WP were in accordance with the latest DoW without deviation with regards to the WP objectives and use of resources. A specification document was produced and submitted to the European Commission as part of D1.1. The document includes the requirements of the condition monitoring system and also describes the critical areas of the rail track that have been selected for investigation. Network Rail has provided information concerning the operating conditions of the rail network in order to assess their effect on the rail inspection system. Network Rail has also supplied information concerning current non-destructive testing techniques used for the in-service monitoring of rails. The limitations of current inspection techniques have been discussed within the consortium and the expectations of the end-users for the newly developed techniques are defined in this document. For Task T1.1 project requirements, the Small and Medium-sized Enterprises (SMEs), Research and Technological Development partners (RTDs) and the end-user discussed the main aims of the project at the kick-off meeting (9 February 2011 at TWI, UK). Furthermore, several meetings took place (most of them at Network Rail facilities) where critical areas of inspection for monitoring were identified and the real rail industry needs were clarified. The project requirements for the MONITORAIL system have been drafted and agreed by all the participants in the deliverable D1.1. The project consortium have also indirectly tried to involve other end-users from different European Union (EU) countries to make sure that the MONITORAIL technology covers the EU rail industry requirements and not just one specific problem for one specific end-user. For Task T1.2 project specification of required monitoring performance to detect defects, discussions during meetings and phone conferences with Network Rail have taken place in which a number of critical applications for the MONITORAIL system were identified. The project consortium with the help of Network Rail have identified that the incidence of rail breaks from corrosion assisted fatigue from the rail foot is a major cause of concern. The foot region is difficult to access for inspection by the widely-used moving ultrasonic inspection methods so that such defects area hard to detect. This is therefore the primary focus of the technique development, although the approach is applicable to the whole section of the rail. Defects in the foot are more likely where corrosion is more severe, such as at level crossings or in tunnels. Network Rail estimate that for rail foot defects, a semi-elliptical crack with threshold depth of 5mm (35 mm2) or more could be critical. However, loss of depth of the foot due to corrosion can in some circumstance be less critical and up to 20% of the cross section (2 mm) could be lost before the fault becomes critical. A range of samples that will be used for the laboratory trials in WP6 has been selected in this work package based on the parameters specified. The samples will vary from feature-free rail samples to real rail tracks. These have been provided by the all the partners, and Network Rail. These samples will be used to develop the MONITORAIL technology and assess the defect detection sensitivity of the system and its reliability over time. TWI has provided rail samples for the pilot experiments. Also further experiments were carried out on a Network rail sample located at the University of Birmingham, UK. An additional two samples mounted on sleepers and clamped with clips have been provided by Network Rail to carry out experiments at TWI to be more representative of real track conditions. It is planned that the final demonstration trials will be performed on a Network rail test track at High Marnham, UK. A detailed table with the list of rail samples can be seen in the Deliverable 1.1 report.
WP2 Theoretical study and modelling was 100% complete at month 9 and was reported as Deliverable D2.1 There were no deviations with regards to the WP objectives and use of resources. Aerosoft was the lead partner work with strong involvement of TWI and Brunel University, UK. The characteristics of ultrasonic guided waves in the complex rail geometrical profile have been identified in this work package. The theoretical generation of the dispersion curves showed that large number of wave modes were present within the rail at the frequency range between 20-90 kHz. This can be a real challenge in terms of using guided waves as an inspection method on rails due to the difficulties of propagating a single wave mode at known velocity within the structure. Hence, the rail cross section was divided into three sections; the head, web and foot. Wave modes with the vibrations solely existing in each section were filtered down and the so-called F3, T2 and F2 modes were selected for the head, web and foot respectively. The modelling helped the consortium to take this decision in order to the reduce the complexity of the problem and facilitate the interpretation of the signals As determined in WP 1, the rail foot has been the focus of the investigation. The F2 guided wave mode in the foot gives good sensitivity across the whole foot section. Hence, significant amount of development focused on optimising the transducer array for generating the F2 wave mode in the foot. The developments considered the accessibility to the foot, whether the access to the top side, bottom side or from both sides (i.e. top and bottom of the foot). Several scenarios have been found to offer suitable means of excitation with respect to the rail foot access. This is based on their reflections from the rail far end with minimum presence of other higher order wave modes as well as 1-D Fast Fourier Transform spectrum analysis, from which it has been possible to quantify the nature of the wave modes that are propagating within the rail foot. Further work was also conducted on propagating ultrasonic guided waves in the rail web and head. Network Rail has highlighted that there is no access issues on the rail web. However, the top and one side of the rail head cannot be accessed for the condition monitoring purposes as these are the running areas. This issue will be considered for future trials. An experimental investigation has been carried out in the foot section of a rail structure, demonstrating 2 mm as the minimum detectable defect depth. The time reversal focusing technique has been applied to the same rail structure. As has been shown in deliverable 2.1 the far-end reflection is clearly visible on transducers in the experiments carried out in TWI lab. Tests on a real track section with clips and sleepers showed that there are many signals arising from these features. This will require the feature extraction. This is to be developed within the project by Cereteth in Tasks 4.2 and 4.3.
WP3 Development and evaluation of transducers for rail monitoring was 100% complete in month 16. Vermon has led this work package with the support of all the RTDs. Vermon has also provided some customised transducers for the execution of the experiments. The aims of this WP are to develop a novel permanently mounted sensor array for the combined acoustic emission and long-range ultrasonic testing techniques, and to develop a modular pulser-receiver unit, capable of operating the MONITORAIL system. Task 3.1 involved the selection of suitable transducers and the development of a permanently attached version of those transducers. The technology chosen in the second period had been proven to be highly robust and applicable to commercial inspections but had not been used in a permanently mounted context. Deliverable 3.1 describes the transducers, the design decisions and explored the adhesive attachment of the transducers to the rail. Task 3.2 took the transducer elements and combined them into an array to support spatial discrimination and localisation of the defects within the rail. The external rail laboratory, created during the second period, is described in deliverable D3.2. The work performed for Task 3.3 was also described in deliverable D3.2 - Energy sourcing system and storage device. An analysis of vibrational energy available from the rail is followed by a comprehensive review and investigation into the current state-of-the-art in harvesting technology. The deliverable concludes with a comparison of the power available from the energy harvesting systems and the requirements of the planned system. Under Task 3.4 a modular system was developed which is controlled by a PC and supports the acquisition of Acoustic Emission data from the array and the performance of long-range ultrasonic inspections using transducers across the array.
WP4 Development of advanced signal processing and analysis tools and flaw sizing was 100% complete in month 18 and was reported as Deliverable D4.1. Task 4.1 was complete in the first reporting period, having investigated the use of wavelets for de-noising. The work performed in Task 4.2 extended the work already performed on the section of rail track at the University of Birmingham (UoB). The rail laboratory at TWI was completed and was the basis for this extended piece of work. Many transducers were strategically placed around the clip system holding the rail onto the sleepers. A variety of signals at different frequencies were injected into the system and the resulting signals at different positions with respect to the injected signal and the clip system were investigated. The resulting acquired signals were analysed with the Fast Fourier Transform and the Short Time Fourier Transform. The conclusions of this work were that this clip system, as distinct from the older clip system on the UoB rail track, caused very little distortion of the injected signals and were found to have very little affect on the propagating guided waves.
A series of parameters were derived from the guided wave signals as the basis for the multi parametric analysis performed as an essential precursor to the flaw sizing, classification and location of the defects. The prototype array was used to generate guided waves in the rails of the rail laboratory at TWI. The obvious effect of temperature was considered first in order to provide subsequent signal processing stages with more consistent, compensated data. A min-max normalisation was found to eliminate most of the temperature variable features. An investigation was then performed into optimal conditions for the extraction of the signal envelope. The intention was to extract the shape of the waveform and reduce to an even greater extent the deleterious effect of temperature on the signal. A Savitzky-Golay smoothing filter was found to most effectively prepare the signal for a Hilbert transform. The extracted shape was found to reject temperature effects well. Several features were considered as the basis for the multi parametric analysis including first order statistical features extracted from the time domain data, the spectral edge frequency derived from a power spectral density analysis, low level indicators derived from a Short Time Fourier Transform and finally, the wavelet transform was used to generate a series of signals which were partitioned at different resolution levels prior to their use in the flaw recognition task.
The flaw sizing, classification and location of Task 4.3 were performed by using a novel fuzzy criterion able to handle the discrimination power of the features and the complementary characteristics between the features. Having automatically selected the features and their respective contributions, the data could be used to train a Support Vector Machine to perform accurate classifications of the inputs. A hierarchical classification technique was developed to further improve the accuracy of the flaw detection. The technique was applied to holes and cracks. For cracks under laboratory conditions, the algorithm managed an average recognition rate of 91.96% while the average recognition rate for holes was found to be 93.3%.
WP5 Software and communication system development was 100% complete in month 18 and was reported as Deliverable D5.1. In Task 5.1 during the second period, an array of longer range wireless technologies were examined having established the capabilities of the developed CloudSensor technology and its range across Paris of 9 km. The preferred technologies were GSM and IEEE 802.11 b/g implemented with directional antennae. Long-range wireless links were demonstrated in the external rail laboratory at TWI. Task 5.1 concluded with a detailed examination of the considerations for sending data over wireless links. D5.1 describes in detail the developed communication software system architecture in two major sections. Initially the MONITORAIL integrated software is described with the workflows leading to the graphical user interface. The functional model of the MONITORAIL integrated system is then described with the corresponding level decompositions.
WP6 System integration and testing was 100% complete in month 23 and was reported as Deliverable D6.1 (also in Periodic Report 2). Sub-components of the modular system were integrated with a compact, low power PC controlling it into a weather proof enclosure ready for trackside installation. The PC has user interfaces for each of the constituent components of the system and also a supervisory software package that automates the operation of each of the constituent parts of the system.
The guided wave sub system has independent channels able to take full advantage of the array of transducers in order to localise the defect.
The power consumption of the system has been optimised with a view to trackside installations without infrastructure power. The supervisory software has the ability to control the duty cycle of guided wave inspections as part of the condition monitoring program and is therefore able to control the systems power consumption. The system also supports a variety of communications methods according to the power available at any particular installation point and the available infrastructure and resources. The system is able to exploit the CloudSensor technology developed during the project and is able to use GSM and long-range IEEE 802.11 b/g implemented with directional antennae.
The chosen PC provides sufficient storage of data to allow the system to be automatically operated without any intervention for at least 1 year. Depending on the communications infrastructure and bandwidth available to the system, the system can be completely remotely managed allowing for even greater periods without direct human intervention.
The system was installed in the external rail laboratory at TWI for the laboratory testing phase of the project. The performance of the system was assessed in a variety of challenging weather conditions and found to exceed expectations. In particular, the system found defects specified by the end-user (35 mm2 transverse defects off centre in the foot) in both free and unclipped rails.
Acoustic emission monitoring was performed on a section of rail over a period of two months while a crack was introduced and grown through cyclic loading of the rail. The resulting fatigue of the material caused the crack to propagate across the rail. The system was shown to be able to distinguish acoustic emissions from the crack propagation events versus emission as a result of noise.
The system was accepted and deployed externally in the TWI rail laboratory to support the final tests of the system and the large scale demonstrations in WP 7.
The wireless system was particularly advantageous in the challenging weather conditions allowing remote management of the unit and full transfer of data collected from the external rail laboratory at TWI. The system was able to make the data available to partners via the internet indirectly through an FTP server. Access to the system was also possible through a virtual private network.
Automation, communication and remote control difficulties were resolved during the course of the work package.
Potential impact
High levels of maintenance cost adversely affect the financial performance of rail operations. The system will improve the efficiency through optimising the split between the cost of the initial investment and maintenance on rail tracks. The automated monitoring of the infrastructure and the associated data processing will aid the development of predictive methods of maintenance for the rail tracks and the better scheduling of the maintenance interventions. This will result in fewer maintenance interventions.
Moreover, the technology is able to detect hidden defects (or irregularities that can propagate to critical defects) before they lead to catastrophic rail failure. The goal of the MONITORAIL system was to detect small cracks in rails, particularly in the foot where they are difficult to detect by other means. The final results presented the systems reliability for successful detection of such defects within rail samples under simulated service conditions (attached to the sleepers with securing clips). The data and analysis provided by the MONITORAIL system is available immediately over a variety of communication media and could be used to improve the design of the rail tracks by providing better understanding on the origins of the defects and the causes of their propagation.
Indirect savings will be also made in terms of the operational cost since accidents and incidents involving broken rails will be minimised. Forced closure of a line with freight and passenger disruption for unscheduled repair interventions to be done is another great annual expenditure that will also be minimised by the introduction of the new MONITORAIL system.
The exploitation of this project will also create a number of new jobs primarily within the partner countries but also in other countries within the EU. It is estimated that 183 new jobs will be created in the EU. Further growth will create more job opportunities.
Project website
The public website address for the project is http://www.monitorail.eu/. This website will be kept as the main platform for any communications related to the MONITORAIL project beyond the project term.
The aim of MONITORAIL is to develop a cost effective wireless condition monitoring system utilising long-range ultrasonic inspection and acoustic emission monitoring technologies. The system is to be used to monitor critical areas of rails where the probability of defects is high or there is limited access to carry out conventional inspection techniques. This will improve the European railway system for better efficiency and safety.
The project will reduce the substantial costs related to rail inspection, and contribute to achieving the target set by the European rail industry to reduce the overall maintenance expenditure by 30% by 2020. A significant increase in the availability of the network is anticipated owing to increased reliability of detection of potential failures.
The main technical objectives of the project are: Development of long-range ultrasonic and acoustic emission methods for accurate flaw detection in rails; Development of novel permanently mounted transducer arrays and electronic hardware for this monitoring task; Development of data analysis tools for automated defect recognition; Development of enabling technology for continuous structural health monitoring, including wireless communications and energy harvesting and storage methods to power the system.
Project context and objectives
To accomplish the project objectives, the work activities have been organised into a number of discrete Work Packages (WP). Those were divided into research (WPs 1 to 6), demonstration (WP7), exploitation and dissemination (WP8) and management (WP9).
The description for those work packages are shown below:
- WP1 Project specifications;
- WP2 Theoretical study and modelling;
- WP3 Development and evaluation of transducers for rail monitoring;
- WP4 Development of advanced signal processing and analysis tools and flaw sizing;
- WP5 Software and communication system development;
- WP6 System integration and testing;
- WP7 Demonstration and large-scale laboratory trials;
- WP8 Exploitation and dissemination;
- WP9 Project Management
Project results
The details of technical and scientific results for each work package are presented as follows:
WP1 Project specifications are 100% complete in month 5. The end-user, Network Rail, led this work package with the support of all the consortium members. Contributions from all partners involved in this WP were in accordance with the latest DoW without deviation with regards to the WP objectives and use of resources. A specification document was produced and submitted to the European Commission as part of D1.1. The document includes the requirements of the condition monitoring system and also describes the critical areas of the rail track that have been selected for investigation. Network Rail has provided information concerning the operating conditions of the rail network in order to assess their effect on the rail inspection system. Network Rail has also supplied information concerning current non-destructive testing techniques used for the in-service monitoring of rails. The limitations of current inspection techniques have been discussed within the consortium and the expectations of the end-users for the newly developed techniques are defined in this document. For Task T1.1 project requirements, the Small and Medium-sized Enterprises (SMEs), Research and Technological Development partners (RTDs) and the end-user discussed the main aims of the project at the kick-off meeting (9 February 2011 at TWI, UK). Furthermore, several meetings took place (most of them at Network Rail facilities) where critical areas of inspection for monitoring were identified and the real rail industry needs were clarified. The project requirements for the MONITORAIL system have been drafted and agreed by all the participants in the deliverable D1.1. The project consortium have also indirectly tried to involve other end-users from different European Union (EU) countries to make sure that the MONITORAIL technology covers the EU rail industry requirements and not just one specific problem for one specific end-user. For Task T1.2 project specification of required monitoring performance to detect defects, discussions during meetings and phone conferences with Network Rail have taken place in which a number of critical applications for the MONITORAIL system were identified. The project consortium with the help of Network Rail have identified that the incidence of rail breaks from corrosion assisted fatigue from the rail foot is a major cause of concern. The foot region is difficult to access for inspection by the widely-used moving ultrasonic inspection methods so that such defects area hard to detect. This is therefore the primary focus of the technique development, although the approach is applicable to the whole section of the rail. Defects in the foot are more likely where corrosion is more severe, such as at level crossings or in tunnels. Network Rail estimate that for rail foot defects, a semi-elliptical crack with threshold depth of 5mm (35 mm2) or more could be critical. However, loss of depth of the foot due to corrosion can in some circumstance be less critical and up to 20% of the cross section (2 mm) could be lost before the fault becomes critical. A range of samples that will be used for the laboratory trials in WP6 has been selected in this work package based on the parameters specified. The samples will vary from feature-free rail samples to real rail tracks. These have been provided by the all the partners, and Network Rail. These samples will be used to develop the MONITORAIL technology and assess the defect detection sensitivity of the system and its reliability over time. TWI has provided rail samples for the pilot experiments. Also further experiments were carried out on a Network rail sample located at the University of Birmingham, UK. An additional two samples mounted on sleepers and clamped with clips have been provided by Network Rail to carry out experiments at TWI to be more representative of real track conditions. It is planned that the final demonstration trials will be performed on a Network rail test track at High Marnham, UK. A detailed table with the list of rail samples can be seen in the Deliverable 1.1 report.
WP2 Theoretical study and modelling was 100% complete at month 9 and was reported as Deliverable D2.1 There were no deviations with regards to the WP objectives and use of resources. Aerosoft was the lead partner work with strong involvement of TWI and Brunel University, UK. The characteristics of ultrasonic guided waves in the complex rail geometrical profile have been identified in this work package. The theoretical generation of the dispersion curves showed that large number of wave modes were present within the rail at the frequency range between 20-90 kHz. This can be a real challenge in terms of using guided waves as an inspection method on rails due to the difficulties of propagating a single wave mode at known velocity within the structure. Hence, the rail cross section was divided into three sections; the head, web and foot. Wave modes with the vibrations solely existing in each section were filtered down and the so-called F3, T2 and F2 modes were selected for the head, web and foot respectively. The modelling helped the consortium to take this decision in order to the reduce the complexity of the problem and facilitate the interpretation of the signals As determined in WP 1, the rail foot has been the focus of the investigation. The F2 guided wave mode in the foot gives good sensitivity across the whole foot section. Hence, significant amount of development focused on optimising the transducer array for generating the F2 wave mode in the foot. The developments considered the accessibility to the foot, whether the access to the top side, bottom side or from both sides (i.e. top and bottom of the foot). Several scenarios have been found to offer suitable means of excitation with respect to the rail foot access. This is based on their reflections from the rail far end with minimum presence of other higher order wave modes as well as 1-D Fast Fourier Transform spectrum analysis, from which it has been possible to quantify the nature of the wave modes that are propagating within the rail foot. Further work was also conducted on propagating ultrasonic guided waves in the rail web and head. Network Rail has highlighted that there is no access issues on the rail web. However, the top and one side of the rail head cannot be accessed for the condition monitoring purposes as these are the running areas. This issue will be considered for future trials. An experimental investigation has been carried out in the foot section of a rail structure, demonstrating 2 mm as the minimum detectable defect depth. The time reversal focusing technique has been applied to the same rail structure. As has been shown in deliverable 2.1 the far-end reflection is clearly visible on transducers in the experiments carried out in TWI lab. Tests on a real track section with clips and sleepers showed that there are many signals arising from these features. This will require the feature extraction. This is to be developed within the project by Cereteth in Tasks 4.2 and 4.3.
WP3 Development and evaluation of transducers for rail monitoring was 100% complete in month 16. Vermon has led this work package with the support of all the RTDs. Vermon has also provided some customised transducers for the execution of the experiments. The aims of this WP are to develop a novel permanently mounted sensor array for the combined acoustic emission and long-range ultrasonic testing techniques, and to develop a modular pulser-receiver unit, capable of operating the MONITORAIL system. Task 3.1 involved the selection of suitable transducers and the development of a permanently attached version of those transducers. The technology chosen in the second period had been proven to be highly robust and applicable to commercial inspections but had not been used in a permanently mounted context. Deliverable 3.1 describes the transducers, the design decisions and explored the adhesive attachment of the transducers to the rail. Task 3.2 took the transducer elements and combined them into an array to support spatial discrimination and localisation of the defects within the rail. The external rail laboratory, created during the second period, is described in deliverable D3.2. The work performed for Task 3.3 was also described in deliverable D3.2 - Energy sourcing system and storage device. An analysis of vibrational energy available from the rail is followed by a comprehensive review and investigation into the current state-of-the-art in harvesting technology. The deliverable concludes with a comparison of the power available from the energy harvesting systems and the requirements of the planned system. Under Task 3.4 a modular system was developed which is controlled by a PC and supports the acquisition of Acoustic Emission data from the array and the performance of long-range ultrasonic inspections using transducers across the array.
WP4 Development of advanced signal processing and analysis tools and flaw sizing was 100% complete in month 18 and was reported as Deliverable D4.1. Task 4.1 was complete in the first reporting period, having investigated the use of wavelets for de-noising. The work performed in Task 4.2 extended the work already performed on the section of rail track at the University of Birmingham (UoB). The rail laboratory at TWI was completed and was the basis for this extended piece of work. Many transducers were strategically placed around the clip system holding the rail onto the sleepers. A variety of signals at different frequencies were injected into the system and the resulting signals at different positions with respect to the injected signal and the clip system were investigated. The resulting acquired signals were analysed with the Fast Fourier Transform and the Short Time Fourier Transform. The conclusions of this work were that this clip system, as distinct from the older clip system on the UoB rail track, caused very little distortion of the injected signals and were found to have very little affect on the propagating guided waves.
A series of parameters were derived from the guided wave signals as the basis for the multi parametric analysis performed as an essential precursor to the flaw sizing, classification and location of the defects. The prototype array was used to generate guided waves in the rails of the rail laboratory at TWI. The obvious effect of temperature was considered first in order to provide subsequent signal processing stages with more consistent, compensated data. A min-max normalisation was found to eliminate most of the temperature variable features. An investigation was then performed into optimal conditions for the extraction of the signal envelope. The intention was to extract the shape of the waveform and reduce to an even greater extent the deleterious effect of temperature on the signal. A Savitzky-Golay smoothing filter was found to most effectively prepare the signal for a Hilbert transform. The extracted shape was found to reject temperature effects well. Several features were considered as the basis for the multi parametric analysis including first order statistical features extracted from the time domain data, the spectral edge frequency derived from a power spectral density analysis, low level indicators derived from a Short Time Fourier Transform and finally, the wavelet transform was used to generate a series of signals which were partitioned at different resolution levels prior to their use in the flaw recognition task.
The flaw sizing, classification and location of Task 4.3 were performed by using a novel fuzzy criterion able to handle the discrimination power of the features and the complementary characteristics between the features. Having automatically selected the features and their respective contributions, the data could be used to train a Support Vector Machine to perform accurate classifications of the inputs. A hierarchical classification technique was developed to further improve the accuracy of the flaw detection. The technique was applied to holes and cracks. For cracks under laboratory conditions, the algorithm managed an average recognition rate of 91.96% while the average recognition rate for holes was found to be 93.3%.
WP5 Software and communication system development was 100% complete in month 18 and was reported as Deliverable D5.1. In Task 5.1 during the second period, an array of longer range wireless technologies were examined having established the capabilities of the developed CloudSensor technology and its range across Paris of 9 km. The preferred technologies were GSM and IEEE 802.11 b/g implemented with directional antennae. Long-range wireless links were demonstrated in the external rail laboratory at TWI. Task 5.1 concluded with a detailed examination of the considerations for sending data over wireless links. D5.1 describes in detail the developed communication software system architecture in two major sections. Initially the MONITORAIL integrated software is described with the workflows leading to the graphical user interface. The functional model of the MONITORAIL integrated system is then described with the corresponding level decompositions.
WP6 System integration and testing was 100% complete in month 23 and was reported as Deliverable D6.1 (also in Periodic Report 2). Sub-components of the modular system were integrated with a compact, low power PC controlling it into a weather proof enclosure ready for trackside installation. The PC has user interfaces for each of the constituent components of the system and also a supervisory software package that automates the operation of each of the constituent parts of the system.
The guided wave sub system has independent channels able to take full advantage of the array of transducers in order to localise the defect.
The power consumption of the system has been optimised with a view to trackside installations without infrastructure power. The supervisory software has the ability to control the duty cycle of guided wave inspections as part of the condition monitoring program and is therefore able to control the systems power consumption. The system also supports a variety of communications methods according to the power available at any particular installation point and the available infrastructure and resources. The system is able to exploit the CloudSensor technology developed during the project and is able to use GSM and long-range IEEE 802.11 b/g implemented with directional antennae.
The chosen PC provides sufficient storage of data to allow the system to be automatically operated without any intervention for at least 1 year. Depending on the communications infrastructure and bandwidth available to the system, the system can be completely remotely managed allowing for even greater periods without direct human intervention.
The system was installed in the external rail laboratory at TWI for the laboratory testing phase of the project. The performance of the system was assessed in a variety of challenging weather conditions and found to exceed expectations. In particular, the system found defects specified by the end-user (35 mm2 transverse defects off centre in the foot) in both free and unclipped rails.
Acoustic emission monitoring was performed on a section of rail over a period of two months while a crack was introduced and grown through cyclic loading of the rail. The resulting fatigue of the material caused the crack to propagate across the rail. The system was shown to be able to distinguish acoustic emissions from the crack propagation events versus emission as a result of noise.
The system was accepted and deployed externally in the TWI rail laboratory to support the final tests of the system and the large scale demonstrations in WP 7.
The wireless system was particularly advantageous in the challenging weather conditions allowing remote management of the unit and full transfer of data collected from the external rail laboratory at TWI. The system was able to make the data available to partners via the internet indirectly through an FTP server. Access to the system was also possible through a virtual private network.
Automation, communication and remote control difficulties were resolved during the course of the work package.
Potential impact
High levels of maintenance cost adversely affect the financial performance of rail operations. The system will improve the efficiency through optimising the split between the cost of the initial investment and maintenance on rail tracks. The automated monitoring of the infrastructure and the associated data processing will aid the development of predictive methods of maintenance for the rail tracks and the better scheduling of the maintenance interventions. This will result in fewer maintenance interventions.
Moreover, the technology is able to detect hidden defects (or irregularities that can propagate to critical defects) before they lead to catastrophic rail failure. The goal of the MONITORAIL system was to detect small cracks in rails, particularly in the foot where they are difficult to detect by other means. The final results presented the systems reliability for successful detection of such defects within rail samples under simulated service conditions (attached to the sleepers with securing clips). The data and analysis provided by the MONITORAIL system is available immediately over a variety of communication media and could be used to improve the design of the rail tracks by providing better understanding on the origins of the defects and the causes of their propagation.
Indirect savings will be also made in terms of the operational cost since accidents and incidents involving broken rails will be minimised. Forced closure of a line with freight and passenger disruption for unscheduled repair interventions to be done is another great annual expenditure that will also be minimised by the introduction of the new MONITORAIL system.
The exploitation of this project will also create a number of new jobs primarily within the partner countries but also in other countries within the EU. It is estimated that 183 new jobs will be created in the EU. Further growth will create more job opportunities.
Project website
The public website address for the project is http://www.monitorail.eu/. This website will be kept as the main platform for any communications related to the MONITORAIL project beyond the project term.