Behaviour of ballasted track during high speed train passage William Powrie and Jeffrey Priest University of Southampton
Outline of talk Background and Aims Sub-base issues: effect of train speed on vertical track movements during train passage - numerical analysis - field monitoring Ballast issues - ballast migration - flying ballast Conclusions
Background and Aims
Background Increased train speeds on both new and classic railway lines Lack of detailed analytical understanding of track system / sub-soil behaviour, even for traditional speed railways Not sure how well past experience and observation (empiricism) will extrapolate to higher speed lines Application of recent advances in soil mechanics and instrumentation improved understanding of ballast and sub-base performance better whole life cost modelling Is ballasted track the best choice for high speed rail?
Aims To assess the effects of train speed on the load/deformation response of the track foundation, by analysis and field measurements To investigate some additional problems associated with the use of ballasted track for high speed railway lines
Numerical modelling of track-bed displacements Effect of train speed
Method Dynamic, 2D finite element analysis using ABAQUS Elastic, undrained response Interpretation in terms of total stresses Loading magnitude and geometry based on Spoornet COALlink line ( Cape gauge, 1067 mm) 130 kn maximum wheel load Analysis modelled the vertical centre plane along the track, and passage of a three-wagon train Model validated with reference to field data
2D dynamic FE mesh Pad (0.01m thick) Rail 0.38 0.27 Sleeper 0.2 Ballast 0.3 V1 V2 V3 V4 SSB SB 0.2 0.2 E1 A B 0.2 0.2 E2 31.51 Natural ground E3 109.2
Typical results: deflections
Variation of maximum displacement with speed
Field monitoring: methods
Measurement techniques: remote video monitoring Webcam captures digital video images of a target, from which displacement is calculated using computer algorithm. Digital camera frame rate up to 170fps
Measurement techniques: geophones Geophones: LF24, 1 Hz natural frequency, logged at 500Hz Mounted on sleeper or positioned in borehole at different depths in the ground
Geophone data a b c d Geophone produces a voltage proportional to velocity of the sensor (a). Knowing the response characteristics of the geophone the velocity can be computed (b). Integration of data leads to calculated displacement (c). Dominant axle and bogie frequencies can be obtained (d).
Comparison: PIV and geophone data 0.1 0 Geophone Video Geophones and PIV data are in agreement. Displacement (mm) -0.1-0.2-0.3-0.4-0.5-0.6 6 6.5 7 7.5 8 8.5 9 9.5 10 Time (s) Video frame rate of 30fps gives an image every 880mm of travel compared with 53mm for 500 Hz geophones. Both methods capture displacements due to individual axles.
Field monitoring: results What happens in reality?
Effect of train speed: HS1 Standard gauge (1435 mm) Same trains (Class 390 Eurostar sets) Static axle load 15.36 tonne (wheel load ~ 75.3 kn) Speeds ~ 120 km/hr and 270 km/hr
Vertical displacement vs speed for similar trains on HS1
Calculate subgrade modulus from Beam on Elastic Foundation (BOEF) model
Track modulus from sleeper displacements assuming a constant (static) axle load
Track modulus vs displacement for all sleepers; track modulus assumed constant for an individual sleeper
Trend lines for different speeds at constant modulus
Increase in dynamic load with speed Dynamic FE analysis suggest that at o.5v c (train speed = 400km/h) dynamic load increases by less than 10% of static, so at 260km/h dynamic load static load Field monitoring suggests dynamic load at 260km/h is around 1.2 1.3 static At 260km/h, Li and Selig (1998) suggest dynamic load increases to around 2.45 static
Ballast issues (1) Ballast migration (ballast circles)
Ballast migration
Ballast migration investigation: instrument layout Vertical, lateral and longitudinal sleeper velocities measured using geophones
Measured sleeper displacements Vertical displacement of high rail end of sleeper is about 2.2 times that of low rail end
h Y e Q e α mv 2 /R mgsinα s mg Y i h c For a Pendolino traversing a curve of radius 1230m radius at 180km/h, quasi static analysis gives Q e = 85.0 kn and Q i = 49.2 kn, i.e. a ratio of loads of 1.73 Q i
Train Run Inner rail δ mm Outer rail δ mm Ratio δ inner /δ outer Ratio k outer /k inner 1 0.390 0.853 2.19 1.27 2 0.414 0.900 2.17 1.25 3 0.402 0.903 2.25 1.30 4 0.387 0.908 2.35 1.36 Ratio of deflections is greater than the ratio of quasistatic loads implying difference in support stiffnesses
Dynamic analysis using Vampire gives a maximum value of Q e of >100kN and a load ratio of up to at least ~2.5
Proposed mechanism Idealised movement of sleeper During loading sleeper rotates about the low rail end and moves towards the high rail end. Due to shape of sleeper ballast falls vertically during loading and is pushed up (down slope) during unloading.
WCML: vertical movements due to loco + coaches vs Pendolino trainset Class 87 locomotive produces displacements comparable with the Class 390 Pendolino train; displacements for Mk3 coaches are considerably less.
Ballast issues (2) Ballast flight (flying ballast)
Flying ballast Geotechnical and aerodynamic investigation in collaboration with University of Birmingham (aerodynamics: Professor Chris Baker, Dr Andrew Quinn) and Network Rail HS1 (David Hutchinson, Mick Hayward)
Background During passage of a high speed train, ballast particles become detached from the ballast bed ( flying ballast ) Cause rail defects which require grinding to repair Causes damage to undercarriages and exposed equipment unless protection is provided At present the cause is not understood Is it mechanical, aerodynamic or a combination?
Measurements Geophones to measure velocities of sleepers during train passage Accelerometers to measure accelerations of the ballast High speed camera to observe air turbulence near sleeper (Professor Chris Baker, University of Birmingham)
Instrumentation layout Aerodynamic equipment installed in between sleeper Accelerometer position in ballast
Instrumentation view
Sleeper velocity vertical At the site monitored, sleeper vertical velocity was typically around 20mm/s
Sleeper displacements - vertical
Accelerations Filtered accelerations from ballast are similar in magnitude to accelerations of sleeper obtained by differentiating sleeper velocity
Air turbulence: visual observation
Does the track see this turbulence? Slight increase in voltage is observed just before train arrives (4.2 m ahead of first wheel). Is this caused by the turbulence?
Ballast flight: findings Sleeper velocity was reasonably consistent for all train passages and sleepers ~ 20 mm/s Maximum ballast accelerations were ~ 3 m/s Geotechnical effects (ground accelerations) alone are insufficient to cause ballast flight: the cause is probably a combination of aerodynamic and ballast acceleration effects Video recording showed pulse of air, which is quite turbulent, travelling in front of train, which may give rise to downward force into ballast
Conclusions
Conclusions Stresses and deflections increase with train speed - perhaps more than analysis of perfect track would suggest, but less than current empirical rules allow Differential forces on rails when curving at high cant deficiency together with sleeper geometry and trainset operation ( 10 the number of high load events per train pass) can cause ballast migration Combined aerodynamic and ground vibrational effects can lead to ballast flight
Journal papers (1) Monitoring the dynamic displacements of railway track. D Bowness, W Powrie, A C Lock, J A Priest and D J Richards. Proc I Mech E Part F, J Rail and Rapid Transit 221 (F1), 13-22, March 2007. Awarded IMechE John F Alcock Memorial Prize and Thomas Hawksley Gold Medal Stress changes in the ground below ballasted railway track during train passage. W Powrie, L A Yang and C R I Clayton. Proc I Mech E, Part F, J Rail and Rapid Transit 221 (F2), 247-261, May 2007 Dynamic stress analysis of a ballasted railway track bed during train passage. L Yang. W Powrie and J A Priest. J ASCE Geotechnical and Geoenvironmental Engineering 135(5), 680-689, May 2009 Determination of dynamic track modulus from measurement of track velocity during train passage. J A Priest and W Powrie. J ASCE Geotechnical and Geoenvironmental Engineering 135(11), 1732-1740, November 2009
Journal papers (2) A full-scale experimental and modelling study of ballast flight under high-speed trains. A D Quinn, M Hayward, C J Baker, F Schmid, J A Priest and W Powrie. Proc I Mech E, Part F, J Rail and Rapid Transit 224 (F2), 61-74, 2010 Measurements of transient ground movements below a ballasted railway line. J A Priest, W Powrie, L Yang, P J Gräbe and C R I Clayton. Géotechnique 60(9), 667-677, September 2010 Contribution of base, crib, and shoulder ballast to the lateral sliding resistance of railway track: a geotechnical perspective. L M Le Pen and W Powrie. Proc I Mech E, Part F, J Rail and Rapid Transit 225(F2), 113-128, 2011 An assessment of transition zone performance. B Coelho, P Hölscher, J A Priest, W Powrie and F Barends. Proc I Mech E, Part F, J Rail and Rapid Transit 225(F2), 129-139, 2011
Acknowledgements EPSRC Daren Bowness, Chris Clayton, Tony Lock, Louis le Pen, David Richards, Liang Yang; University of Southampton David Hutchinson, Mick Hayward; Network Rail CTRL Chris Baker, Andrew Quinn; University of Birmingham Patric Mak, Mark Burstow, James Dean; Network Rail
Thank you for listening