7 - Fatigue Failure of Track - Crack Initiation Flashcards
Meaning of fatigue failure
Crack growth in rail taking place over many thousands/millions of wheel passes (i.e. many load cycles)
Caused by repeated wheel contact, hence ‘rolling contact fatigue’ (RCF)
Fatigue failure of rail joints
Lots of problems at bolted joints
Stress concentration ‘bolt hole cracking’
Mostly in older track
Electrically insulated joints still needed in continuously welded track
Fatigue failure of rail foot
Often caused by corrosion
Case studies of consequence of no maintenance
6th February 2017 - 660mm gap with sulphuric acid tankers going over it, near NYC, slow speed freight
17th October 2000 - Hatfield crash, London to Leeds derailed at 115mph, 4 people killed
What are ‘squat’ defects?
Found on running surface of rail head
Cracks typically grow 25-50mm at shallow angle beneath rail surface
May branch towards rail surface, causing a spall, or downwards into rail, leading to a transverse rail break
Usually below a dark spot on rail surface, with widening of rail/wheel contact band, and by formation of small depression on rail surface
Typically occur on straight track or in shallow curves
What are ‘head check’ defects?
Consist of series of surface breaking cracks in gauge corner of rail
Typically form on high (outer) rail in curves
Don’t often propagate to form deep cracks or transverse rail breaks
Often result in chips of rail gauge corner breaking away
Lots of repeated small surface cracks can make rail ‘ultrasonically untestable’ (i.e. impossible to prove using ultrasonic non-destructive testing/NDT that there are no larger cracks below)
Results in rail replacement
Consequences of fatigue cracking
Maintenance required - rail grinding to remove small cracks and redistribute load
Lubrication to control rail surface friction levels to prevent crack formation
Rail replacement if severe cracking
Regular non-destructive inspection (e.g. eddy current, ultrasonic or visual) to spot early cracking stages
Management prevents safety problems but expensive
Is cracks are missed/grow quickly there’s potential for rail break and derailment
Plastic flow at rail surface
RCF usually originates from plastic flow
Same plastic flow generates wear (delamination mechanism) and causes crack initiation
Extension of cracks deeper, below plastic damage, depends on different propagation mechanisms
Surface traction from driving wheel ‘pushes back’ steel near rail surface
Modelling rail RCF using bearings/gears
Rail RCF has similarities to bearings and gear teeth surface pitting (issue for wind turbines)
Surface hardness (correlates with yield) - rails: 250-450Hv, bearings/gears: high, >700Hv
Lubrication - rails: dry or water, bearings/gears: full film oil
Surface plasticity - rails: lots, bearings/gears: no
Laboratory/academic definitions of crack initiation
Dry contact, 1500MPa maximum Hertzian contact pressure
Cracks much larger, dislocations are visible with EM after just 125 passes
Industry definition of crack initiation
Not exact, but usually refers to cracks up to size they can first be found during rail inspection
Ultrasonic non-destructive inspection can find cracks from 4mm upwards
By 4mm, mechanisms driving crack development have moved beyond pile-up of dislocations, so are now large relative to steel microstructure
2 ways to understand crack initiation
Strain accumulation under high loads of Hertzian contact (materials focused, considers shear stress internal to rail steel)
Energy input to rail steel (sliding work at rolling-sliding contact on rail surface, correlated with observed damage)
Rail internal shear stresses
Plotting absolute value of peak shear stress experienced with depth below a contact reveals that at low surface traction (no driving/braking traction), it has a sub-surface peak
With little traction applied, this peak can produce plastic damage below the surface
If traction/braking occurs (higher traction coefficient), peak sub-surface shear stress can rise towards rail surface
Peak in shear stress at surface is more damaging than below as the stressed steel has reduced support from surrounding lower stressed material
Stresses in rail for low rail-wheel load
As wheel rolls past, rail steel experiences some elastic deformation, but this is recovered when wheel moves away
No permanent deformation but loads need to stay low
Stresses in rail for very high rail-wheel load and contact pressure (much higher than elastic limit)
Every wheel pass, rail material experiences some permanent plastic deformation and also leave some residual stress in rail steel
Leads to ‘incremental plastic collapse’ or ‘plastic ratcheting’
Large strains can accumulate over thousands/millions of wheel passes
Stresses in rail for high rail-wheel load, contact pressure exceeds elastic limit but is not high enough to cause ratcheting
First few wheels cause some plastic flow and accompanying residual stress in rail
But residual stresses are ‘protective’, so make yielding less likely in second wheel pass than in first etc…
Initial plastic flow has taken place but then stopped for the same wheel loads - called ‘shakedown’
Allows rail-wheel system to safely operate above rail yield point without permanent accumulating plastic damage
What do the shakedown limit and ratcheting threshold depend on?
Friction level at rail surface
Because, using plot, shear stress peak moves towards surface with increase in friction
At surface, material under highest stress level in ‘uncontained’ by surrounding less stressed material, so can yield and ratchet more easily
Define shakedown limit
Highest applied load possible without continuing plastic deformation
Advantage of operating at shakedown rather than elastic limit
Assess yield of ductile materials using Tresca criterion
Shows that in pure shear, a material yields at a shear yield stress equal to half the tensile yield
Using shakedown plot, can find maximum value without causing continued plastic flow, so can find when initial yield followed by shakedown to stable state will happen
Shakedown limit complications
Materials which strain harder can take even higher loads - almost double non-strain hardening cases
Strain hardening behaviour of different steels varies - experiments/modelling are needed to understand how new rail steel will behave
Even if designing for operation below shakedown limit, doesn’t always work: poor suspension design/maintenance; wheel faults; difficulty calculating shakedown limits for real contacts; dynamic forces; bridges under tracks; welds and other features producing dynamics loads higher than designed for
Crack initiation - energy input and ‘damage index’ method
Sacrifice link to mechanisms and material properties but gain ability to quickly predict likely damage sites
Energy based wear number can be correlated with fatigue crack initiation
Fatigue crack initiation is referred to here as ‘damage’
Wear removes rail surface including any cracked material (so can reduce damage)
Damage index quantifies process
Damage index
Defined with a value of 1, indicating fatigue cracking, just visible on rail head
More hands-on, industry definition than stress analysis method
For particular combination of wear number and rail material the damage index is a non-dimensional number equal to the proportion of fatigue life of the material exhausted by 1 wheel pass
Converting wear number to damage index
Correlations are non-linear
At very low wear numbers, energy input is too low drive crack initiation
Mild wear regime - crack initiation can take place driven by energy available, but low levels of wear
Severe wear regime - large amounts of rail material is removed including any initiating cracks, resulting in negative or zero damage
Major issue - hoe energy is really divided between wear, crack initiation, heat, noise, vibration etc. is not known
Using the damage index
Damage index application depends on vehicle dynamics calculations to predict contact patch size, location and stress
Need rail and wheel cross-sectional profiles, typically measured every 500mm along track for detailed analysis
Measure rail and wheel profiles -> vehicle dynamics calculations -> sum up damage from all wheels passing each location
Typically use mix of 25% new, 50% mid-worn and 25% worn wheels
Damage index outputs
Damage is summed up for all vehicle types considered
Damage maps produced, showing where fatigue crack initiation is expected
Model has been compared to real track inspection data and damage function modified to improve predictions
Damage index research issues
Reliance on correlation with past cases of fatigue cracking rather than modelling actual mechanisms of fatigue crack initiation
Works until there are major changes (e.g. new rail steels)
How to ‘sum up’ damage from different traction directions isn’t fully resolved - some create damage, others don’t
Vertical load/weight of train isn’t considered
Common underpinning of both approaches to crack initiation
Describe the same underlying physical events
Accumulation of plastic strain cycle cannot go on indefinitely
At a certain strain (limit of ductility) heavily deformed steel loses its integrity
Surface material breaks away as wear debris
Sub-surface, steel stays in place but can’t support a shear or tensile load (i.e. has begun to behave like a crack)
Limit of ductility can be found by experiments
Wear and fatigue crack interaction
Experimental evidence - only way to understand if fatigue or wear will donate on a new type of rail steel
Rail grinding is artificial wear - re-profiles the rail, also removing near surface damage and cracks
