by
Professor Lewis Lesley, BSc, PhD, CEng, MICE, FCIT, FIFF, MUITP, FRSA
Professor of Transport Science,
Liverpool. J. M. University
| Summary | ||||
| 1.0. | Introduction | |||
| 2.0 | Analytical. | |||
| 2.1 | Finite Element Analysis | |||
| 2.2 | 2 & 3 dimensional beam/spring analysis. | |||
| 2.3 | Noise and vibration transmission. | |||
| 3.0 | Laboratory Tests. | |||
| 3.1 | Description of Rail Samples Tested | |||
| 3.2 | Description of the Test Rig | |||
| 3.3 | Non Destructive Testing | |||
| 3.3.1 | The Endurance Testing Procedure | |||
| 3.3.2 | Static Tests. | |||
| 3.3.3 | Dynamic Tests | |||
| 3.3.3.1 | Ambient Temperature | |||
| 3.3.3.2 | Temperature Controlled | |||
| 3.3.4 | Electrical Resistivity. | |||
| 3.3.5 | Results | |||
| 3.4 | Destructive Testing | |||
| 3.4.1 | Failure over Foundation Void | |||
| 3.4.2 | Rail pull out Test | |||
| 3.4.3 | Conclusions from Destructive Test | |||
| 3.5 | Previous Relevant Tests | |||
| 3.6. | Conclusion of Laboratory Test | |||
| 4.0 | Field Trials. | |||
| 4.1 | Rotherham Bus Station. | |||
| 4.2 | South Yorkshire Supertramway. | |||
| 5.0 | Relevant Technical Publications | |||
| 6.0 | Conclusion. | |||
| References | ||||
| Appendix I | Notes on Polyethane grouting employed |
The LR55 track system has been tested numerically, in laboratory and in field trials. A mathematical model for the dynamic and static behaviour of the track has been validated in laboratory tests. Nearly 100 years of track loading has been simulated in laboratory and nearly 50years of loading in a highway environment from the field trials. The track system has completed its testing programme without failure.
In an attempt to resolve some of these issues, as well as addressing some technical consideration like electrical stray currents and noise and vibration transmission a completely new rail track system has been developed. The new rail has no vertical web and is suspended at its top level. (Fig.1) rather than supported from its foot, like traditional rails. (Fig. 2).
Such a different rail form has generated considerable scepticism, and mirrors the difficulty of getting flat bottom (Vignoles) rails accepted in place of traditional bullhead rail in the UK. While Vignoles rails were invented in the third quarter of the 19th Century, they were not accepted as a standard on Britain's railways until 1951, and have only recently been accepted for London Underground.
In order to address these concerns a comprehensive battery of tests; computer simulation, laboratory and in field have been undertaken. This report sets out those tests and their results.
Finite Element Analysis (FEA) enables complex structures to be examined for the loading and stress paths. FEA is now a practical option even on PC's, for very complex strucures. The LR55 track system is a rather simple structure compared to say an aircraft fuselage, but nevertheless has been modelled successfully. This is achieved by dividing the rail, supporting grout, concrete foundation trough and ground into interlinked cells. (Fig. 3).
BSC plc undertook a FEA to determine where areas of high stress occured in the LR55 track installation. Two areas of high stress where identified:
In neither case was the level of stress near the elastic limit of either material. Investigations were made on changing the radius of the rail flange corner or flange thickness but made little difference to the level of stress in the steel.
Including a radius in the concrete trough corner nearly eliminated the high stress in the PU grout.
Further FEA have been undertaken at the Liverpool J.M. University. This has included further modelling of the sub base under the concrete trough to determine how the ground transmits static and dynamic loads from the rail head. The rail and concrete trough were also analysed to determine the transmission of dynamic forces.
Rail tracks behave non linearlly and non isotropically. A simple analytical model is therefore not possible. An explicit model has however been constructed representing the LR55 track system as; rail (beam), PU grout (horizontal and vertical springs), concrete trough (beam) and ground (vertical springs). (Fig. 5).This model was developed by one of Professor Lesley's PhD. students at the Liverpool J.M. University. It has been used to model a variety of ground conditions and to examine the performance of different concrete trough sections. In all cases the maximum BR 25 tonne axle load has been assumed. (Figs. 6-8).
As well as different strength ground and sub base (from loose sand to rock) (Figs. 9-10) the behaviour of the track has been modelled over voids (up to 2m long) appearing under the track concrete trough. The model shows that the track is self supporting and this has been demonstrated in a field trial and validated in laboratory testing. (Fig.11).
A mass-sping-mass-spring model for noise and vibration attenuation into the ground has been constructed. (Fig.12). This predicts a 30dB reduction in the range 0 - 20 Hz and 50dB reduction in the >100Hz range. Before and after noise measurements on a field trial will be used to validate the model.
A battery of tests, static and dynamic, have been undertaken in Professor Lesley's Structural Testing Laboratory in the Liverpool J.M. University. These have been undertaken since 1992. The tests use a standard hydraulic press able to deliver up to a 500kN (about 50tonne) load, either statically or dynamically. In all cases the tests pieces were instrumented with strain and displacement gauges. The applied load was measured with a load cell. All data was recorded contemporaneously with a computer, which was subsequently used to analyse the results obtained.
Sample rails, each 1 metre long were cast by Edgar Allen Engineering Ltd. from standard manganese steel used for normal rails and crossings. The rails had three different depths 60mm, 70mm and 80mm. The first and third samples have plain rounded bases, while the second has notches in the railside. (Figs. 13-14).
These rails were bonded into a weak mix cast concrete bases using two part polyurethane grouting. Attached to the underside of the rails were strain gauges which were placed 0.2, 0.5 and 0.7 metres along the length of the rail, in the centre line.
Some of the tests were conducted in dry conditions. In others the rail and concrete base were covered by 25mm of water retained by a tank created around the concrete base and sealed with silicon mastic. (Fig. 15).
A final battery of tests were conducted using a 6 metre long rail provided by Edgar Allen Engineering Ltd. bonded with a PU grout into a pre-stressed reinforced concrete trough provided by Tarmac Building Products Ltd. and mounted on a compacted sand bed some 750mm deep. This has been used to validate earlier analytical model results.
The rails embedded in their concrete bases were mounted on a hydraulic impulse press, normally used to test highway pavement specimens. A load cell is placed over the centre of the upper surface of the rail and loads imposed via a large steel ball, raised in 0.1 kN steps. The imposed loads are static, or dynamically varied. The strain gauges plus a displacement gauge at the centre of the rail were connected to a computer based data logging system. (Fig.16).
Each rail sample embedded in its concrete base was mounted under the hydraulic impulse press and load cell. At the start of each test increasing static loads were imposed in incremental steps from 0.0 to 100 kN and reference readings from the displacement and strain gauges taken. The rail was then subjected to 1.25 million cycles of sine wave dynamic impulses with a frequency of 15Hz, and a maximum load of 100 kN. (Fig. 17).
A further static graduated load test up to 100 kN was then undertaken. The rail was then subjected to another 1.25 million cycles. Lastly the static load was repeated. The before, middle and after results were then plotted for comparision. 15Hz was selected to represent an operating train speed of about 100 km/h.
This procedure was then repeated for a further 2.5 million cycles with the sample rail submerged in water. Giving 5 million cycles in total for each sample, over 12 years under typical urban operating conditions. The other rail samples were subjected to the same procedures as the initial rail, each having 2.5 million cycles, dry and 2.5 million cycles wet. Giving an accumulated total of 15 million cycles for the initial three rail samples examined in detail.
These have been undertaken on 8 different track samples at up to the equivalent of 80 tonne axle loads. Two different PU grouts have been tested for performance. Three different rail heights (60mm, 70mm and 80mm) were tested, as well as rails having notches in the lower rail side to determine any difference of performance which resulted. After these tests the 80mm rail section was selected for further testing, without side notches. The analytical structural models have been validated on a 6m long track sample. In both cases the rail returns to its neutral position very close to the loading point. (Fig. 18).
Tests were undertaken of a simulated LRV operating speed of 100km/hr with an equivalent 20tonne axle load. This was achieved with a load of 100kN applied sinusoidially at a frequency of 15Hz. In total some 200 million passing axles have been simulated (about 400 years of service on a typical LR system. About 80 million axle loads were tested with the track samples flooded, to determine the degree of water penetration and possible failure. No water penetration or failure was recorded.
One sample was tested dynamically in a temperature chamber in the range -10°C to +60°C. This recorded the hardening at lower temperatures and the softening at higher temperatures of the bonding PU grout, which did not impair the performance of the track system. (Fig.19).
The control of stray currents in normal conditions is determined by the electrical resistivity of tracks. One metre long track samples in the laboratory measured better than 20,000 mega Ω resistance. The resistivity of both PU grouts tested were similar. This translates in the field to over 10,000 Ωkm resistivity. In practice a short stretch of LR55 attached to conventional tracks was found to have a resistivity of over 1000 Ωkm. For street tramways, the Railway Inspector requires a resistivity of over 100 Ωkm.
The preliminary results were plotted for each rail section, differentiated between wet and dry samples. The results show that the rail and PU grout are resilient and obey Hooke's Law. The results also confirm earlier tests undertaken by the University of Calgary. (Ref.1). Samples of rail bonded with both PU materials have been lab tested, giving similar and satisfactory results.
It is worthy of comment at how short was the length of the vertical diplacement of the rail, restoring to the neutral position close to the loading point. (Fig.20). The simulated dynamic loads tested, up to 100 kN, represents a railway axle load of about 20 tonnes. The maximum static load was 400 kN (about 80 tonne axle load).
Both are well in excess of the maximum axle loads experienced on light railways or tramways. The simulated speed of 100 km/h is also higher than normally found in light railways or tramways.
These results have shown that from a static laboratory test using a hydraulic impulse cell to simulate 100 kN loads at 15Hz, equivalent to 100 km/h, sections of LR55 low profile rail with different rail depths were able to withstand and recover from the loads imposed, which were equivalent to maximum main line railway standards. A total of 15 million simulated axle loads were imposed on one grout, equivalent to about 35 years of operation, the PU bonding grout replicated earlier tests at the University of Calgary and was able to withstand the ingress of water from 2.5 million cycles submerged for each sample tested. Over 30 million simulated axles have been tested on the other PU grouting System bonded rail samples.
The rail and bonding material behaved according to Hooke's Law, and no evidence emerged that the elastic limit was approached in any of the samples. Nor was there any evidence of a deterioriation in the performance of the PU grout, which continued to behave elastically. However, these laboratory tests did not simulate a rolling wheel passing along the rail or the impact of sine wave deformation and harmonics created by adjacent wheels. Nor were these tests able to simulate the impact of heavy road vehicles passing transversely across the rail. Those conditions have been fully tested by lengths of rail installed in a highway pavement and on a working tramway or light railway which is mounted in a highway. These field trials are recorded below.
A rail sample was simply supported at a one metre spacing and the load applied in the centre. The load was increased in 10kN increments. The track failed at 296 kN (about 30 tonnes) representing an axle load of nearly 60 tonnes. (Fig.21).
To determine the resistance of the PU grout to thermal expansion forces, which could cause the rail to be dispaced from the foundation trough, a rail was pulled out of the trough with a force increasing in 10kN increments. The one metre track sample failed at a force of 36kN. This is an order of magnitude higher that the thermal expansion forces experienced in normal street track locations. (Fig. 22).
In both cases the PU Grout bond did not fail. The PU remained bonded to the rail and to the concrete trough. The concrete troughs failed in tension. Both tests indicate considerable overcapacity compared to likely loads to be met in normal highway pavements.
The University of Calgary, Department of Civil Engineering undertook similar battery of tests in the 1980s, which came to very similar conclusions to the tests of the LR55 system in Liverpool. A letter from the Light Rail System in Calgary has confirmed the durability of the PU Grouting bond, installed in 1988.
The other PU Grouting material was used in 1994 to bond a 10m length of double conventional railway track at a level crossing on the South Yorkshire Supertramway. This has proved very satisfactory. Since then it has been used to bond a 1km length of double street conventional tramway track in Sheffield on the Hillsborough line. Again this has been entirely satisfactory.
A wide variety of laboratory tests have resulted in consistant data on the behaviour of the LR55 track under static and dynamic axle loads greater than the present UK mainline railway maximum (25 tonnes). The LR55 track behaves elastically, can be installed on weak foundations (like compacted sand) and will successfully bridge voids without failure. Testing full sized but short rail samples has provided considerable confidence on the reliability, robustness and durability of the LR55 track system.
Two field trials have been sucessfully mounted.
A 10 metre length of LR55 rail was installed in March 1993 in the entrance to Rotherham Bus Station, were buses crossed, turned around and passed along the rail. The trial was concluded in Sept. 1995 after about 2 million buses had impacted on the LR55 sample. (Figs.23-24)
The rail and trough were instrumented with strain gauges and the results captured on a data logger. Regular analysis of this data showed that the track remained in good condition. After a year all gauges had exceeded their design life but continuing visual monitoring confirmed that the LR55 track had not deteriorated or failed. This trial represented about 30 years traffic in a typical urban radial road.
This was also confirmed by a trench dug under the site in Sept. 1994 to expose the track construction. The concrete trough showed no signs of failure or deterioration. This was inspected by interested parties including the Railway Inspectorate, public utility representatives etc. The trench showed the shallow construction of the track and that concrete troughs are self supporting over trenches.
Over a weekend in March 1996 a 16m length of LR55 track was installed on a single track section of the Supertramway, at a point where some 300 LRVs and 100 HGVs impact per day (a total of about 120000 vehicles pa, or about 6million tonnes pa). (Figs. 25-26). A failure of this sample would halt the main line to Meadowhall. The track was installed one rail per night time possession, without interuption of day time rail service or road traffic stand by bus and temporary level crossing. This installation has been monitored regularly, and the alignment measured for stability.
The installation was accepted by South Yorkshire Supertram Ltd. into its maintenance responsibilty in Sept. 1996. Before installation noise measurements were taken, and an after noise measurement will be taken to confirm the noise attenuation performance of the LR55 track system.
As a result of the testing programme and research work at the Liverpool John Moores University, a large corpus of refereed scientific publications has been achieved. These are listed fully at the end of this Report. (Refs. 2-35).
The LR55 track system has been subjected to a battery of tests simulating over 200 years on a light rail system and validated by more than 6 years on a real light rail system. Several other new track systems have been introduced in the last ten years, eg. Manchester, Sheffield, and currently Wolverhampton. None has been tested as comprehensively as the LR55 track system.
Two different two part polyurethane (PU) grouts have been used in the testing programme.