You are on page 1of 18

ESCHEDE TRAIN DERAILMENT

BHATT RAJAT RAJENDRAPRASAD-2017H1410156P

-
JAIN AKSHAY PAVAN 2017H1410147P
-

INTRODUCTION
Fracture of structures and components due to fatigue crack growth in many cases cannot
solely be explained by fatigue or service strength investigations. Not only in those cases, but
even more generally fracture mechanics provides valuable approaches and methods in order
to predict and to prevent crack growth or at least in order to understand and investigate
already occurred failures. Fracture mechanical evaluations also contributed for the
objectification of the discussion concerning the failure of the ICE-wheel tyre, which caused
the disastrous accident of Eschede . The ICE Wilhelm-Conrad-Rontgen was partially
equipped with rubber-sprung wheels.

One of those wheels broke down about 6 km before crossing a track switch near Eschede. At
this track switch parts of the train put aslant and crashed into a bridge, which collapsed from
this.Rubber-sprung wheels in general consist of a wheel tyre, a wheel rim and a number of
rubber blocks, that are clamped in between. The highest stress in the wheel tyre can be found
on its inner side. As consequence of the rotation of the wheel a cyclic loading can be
observed in the wheel tyre. This loading led to an extended fatigue crack growth and in the
end to the fracture of the wheel tyre.

The finally broken ICE-wheel in new condition had a diameter of 920 mm. Due to wear and
thus necessary re-profiling the diameter of the broken wheel had decreased to 862 mm (the
limiting size of the diameter was set to 854 mm).As result of the decrease of the diameter
(respectively the thickness of the wheel tyre) the cyclic circumferential stress, which is the
relevant stress component for fatigue, is increasing. The stress and fracture mechanical
analyses described in the following especially are performed for the broken wheel s diameter
of 862 mm.

To mitigate noise and vibration, monolithic (made entirely of one material) steel wheels were
replaced by what are known as resilient wheels. These wheels feature a rubber dampener
around the centre hub with a steel rim around the outside of the rubber. The rubber is less
stiff than the steel, and effectively dampens vibrations, shocks and noise. These wheels had
proven track records on street cars and light rail systems, and were very effective at reducing
noise and creating a smoother ride. These wheels are still used today (quite effectively and
appropriately) in such systems. However, such wheels had never before been used on a high
speed intercity train.
OVERVIEW OF THE ACCIDENT

On the early morning (5:45 am) of 3 rd June 1998, ICE-884 Wilhelm Conrad Rntgen with
a capacity of around 800 passengers, set off its journey from Munich to Hamburg with 287
passengers on board. ICE-884 was equipped with 18 Rubber-Sprung (type BA 064) and 38
Monobloc wheel sets. After a quick stop in Hanover at 10:30am, the train continued its trip
northwards. About 6 km away from village of Eschede, in Lower Saxony, at a speed of
125mph, the wheel tire of the last rubber-sprung wheel on the right side of first carriage
broke down. Accompanying a big noise and violent vibrations, a passenger Jrg Dittmann
noticed a piece of metal went through an armrest between his wife and son and he reported
this to the conductor who was in coach 3. Rather than applying emergency brakes, conductor
insisted to see the damage himself, owing to the company policy. Metal piece was the wheel
tire peeled away from the wheel body, punctured the floor of first carriage and kept
embedded for almost 6km back from the accident site. The embedded wheel tire was hung up
directly under the carriage screeching along the rail track making constant damage to it.
There were two sets of points (use to change the track of train) on the rail track coming
shortly in the way of ICE-884. Just about 3.6sec before the crash, as the train reached the first
set of point, the end of the broken wheel tire scooped up the check rail (check rails used to
guide railways safely through set of points) and the check rail smashed into the floor of first
coach. The strike was massive enough to cause derailment of the two wheels in the rear of
coach 1. At that stage if the train still could have been stopped by emergency brakes, the
derailment might not had turned into a big catastrophe. Train continued to move on with the
speed of 125 mph and one of the derailed wheels hit the switch of the second set of point and
opened them. That caused the switching off the coaches, following carriage 1, to the local
branch line parallel to main line. Sudden switching off the track of train at a high speed
resulted in disconnection of the front power head from rest of the train. As a consequence of
this 3 rd carriage smashed into a fragile pier of a nearby 300 tons road bridge due to which
the bridge collapsed and fell onto the rear half of coach 5 and restaurant coach 6, destroying
them completely. The coach 4 derailed by the big vibrations of coach 3, went down the track
and killed 2 Deutsche Bahn workers nearby. Seconds before this collision, Jorg Dittmann was
just going to show the conductor that a big metal piece has punctured through the floor of
first passenger coach; but they already took precious time coming from coach 3 to the coach
1. Conductor did not see the damage and the brakes were not applied and eventually the
accident occurred at 10:59 am local time. The collapsed bridge completely blocked the track
and subsequent remaining coaches 7 (service car), 8th and 9th, first class coaches (10-12) and
rear power head all derailed and jack knifed into the pile. By 11:07am, the police declared a
major emergency and more than 1000 rescue workers were sent to the crash site. [7][9]
Eschede train disaster is the worst high speed train accident in German history, after World
War 2, with 101 people killed and 88 people injured.
Description of the actual reason of the failure
Although many critical issues, explained later in this case study, were involved in the incident
which led to the catastrophic failure of a first generation ICE 884 Wilhelm Conrad Rontgen
on 3 rd June 1998, but, the main reason of that failure, after the trial of the incident was found
as metal fatigue; and is still considered as the root cause of the failure. At the speed of 125
mph ICE 884, while on its way to Hamburg from Munich, was passing through the village of
Eschede, when the last wheel on the right side of the first bogie suffered from a fatigue crack,
causing devastating results. ICE 884 was equipped with mono bloc wheels as well as rubber-
sprung wheels. The fractured wheel was a dual bloc rubber-sprung resilient wheel type BA
064. Generally these Rubber-Sprung wheels consisted of a steel wheel disc (inner wheel rim
having a hole for shaft), a steel wheel tire (outer wheel rim) and a number of rubber pads,
sandwiched between wheel disc and wheel tire, to minimize the noise problem. It was the
outer rim, the wheel tire made of steel, which was broken when a fatigue crack finally
propagated through the wheel tire causing the deadliest fast speed train accident in the world.
The crack was originated near the center of the inner surface of the wheel tire which contacts
the rubber block and where the highest stress level in the wheel tire was located. Fatigue
failure occurred, even at low mean level stress, because of the extra thinning of the wheel
tire. [1] Fatigue is known as a single cause which can lead to failure of metals, polymers and
some ceramics as well, with glass as the only exception. In metals particularly iron based
alloys it is a common issue for the fractured surfaces. The applied stress which can lead to
fatigue failure may be axial (tension-compression), flexural (bending), or torsional (twisting
in nature). Different stress-time modes are possible regarding to fatigue. In a reversed stress
cycle amplitude is symmetrical about a mean zero stress level, for example, alternating from a
4 maximum tensile stress ( max) to a minimum compressive stress ( min) of equal
magnitude. In another type, known as repeated stress cycle the maxima and minima are
asymmetrical relative to the zero stress level. For this case the stress level alternates about a
mean stress m, defined as the average of the maximum and minimum stresses in the cycle.
m = ( max + min )/2

There is a range of stress r which is just the difference between max and min;

r = max - min
Stress amplitude a is just one half of this range of stress, or a = r /2 Another term stress
ratio R is just the ratio of min and max.
R = min / max
Issues linked with catastrophic failure:

Inadequate design of Rubber-sprung resilient wheels:

Initially the 1st generation ICE trains were equipped with the single cast mono bloc wheels.
These wheels are a single assembly made of steel and are much stronger and long life wheels.
On the other hand the rubber-sprung wheels, of type BA 064, which were used for ICE 884
later on, are a dual bloc wheel. This wheel consists of a wheel tire, 34 rubber pads (20 mm
thick), an inner wheel disc and a solid shaft. Wheel rim for this type of wheel is divided into
two parts, a central wheel disc and an outer detachable wheel tire. Rubber pads are pressed
between wheel tire and disc, while they are both bolted together. Due to the presence of
rubber pads the wheel tire has to be made thin. Highest stress level on the tire can be found
on the inner side of the wheel tire where it makes contact with rubber pad. Design of this new
type of wheels was clearly having flaws in it. It was made to overcome the problem of
fatigue, noise and vibrations at cruising speeds; but this design resulted in fatigue failure. As
a result of rotation of the wheel, cyclic loading can be observed in the wheel. The thickness
of already thinner wheel tire reduced much quickly because of the intense cyclic loading
caused by resilient rubber pads sandwiched between a two parts steel rim. As the train
travelled, the inner wheel disc also worn out and its diameter reduced which caused further
severe dynamic and cyclic loadings on the wheel tire making it thinner and weakened it. This
resulted in increasing the stress amplifications on a single point on the inner side of wheel
tire, from where the crack initiated and finally propagated to cause failure. This fracture
occurred at relatively low fatigue mean stress level as the cross section of wheel tire was
reduced up to 7 dangerous limits. Also thermal expansion co efficient of rubber is much less
than steel, hence the rubber could not have expanded while heated. This added the stress on
wheel tire.
Poor Operational Testing/Inspection Procedures:

During the operational modes of ICE 884 WCR methods employed to test the rubber-sprung
wheels were inadequate and poor. The engineers of DB used to check the macro level cracks
with a flash light. For micro level cracks no useful method was employed that time other than
a high tech technique coming up with results having constant errors in them (ultrasound
technique). Hence, engineers were unable to detect the cracks inside the wheel tire or disc.
Also no equipment was present that time to perform the complete fatigue test (crack
initiation, propagation and failure analysis) while the train was operational.

Tread Diameter Minimum Limit Set:

Experts of Fraunhofer Institute expressed their concern that metal fatigue could lead to wheel
rim failure. Experts warned that wheels should not be operated if worn out to levels below
880mm of their tread diameter. Even then DB set the minimum limit for the tread diameter as
854mm. New wheels had a tread diameter of 920mm and the wheel that fractured was having
a tread diameter of 862mm. [4].Given below is the table containing data related to that
fractured wheel:

Different Notifications Pointing Out Defected Wheel :

In July 1997, the company that runs the network of Trams in Hanover, detected dangerous
metal fatigue cracks occurring in their dual bloc resilient wheels; even though trams were
operated at the speeds of only 24 km/h .Only months before the incident Tram company
notified Deutsche
8 Bahn with this problem. But DB did not perform any checks for they were not having any
problems of metal fatigue in the wheels of their trains. When the maintenance report was
downloaded from the crash train onboard computer, another, vital information was revealed.
It was found that, two months before the incident, conductors of train forwarded about eight
separate complaints about the unusual noise and vibration coming from the carriage which
had the wheel that fractured. Just week before, the wheel that got fracture was highlighted as
being defected in three separate automated checks. In spite of all these warnings the fractured
wheel was not replaced and it took the lives of 101 people after getting a sudden fatigue
failure.
Strict Company Policy:

Had the emergency brakes been applied immediately after the derailment of train, or tearing
off the floor of first bogie by broken tire, ICE 884 might had stopped before hitting into the
bridge; and many lives could have been saved. But, owing to the strict policy of DB
Company emergency brakes can only be applied after the conductor/manager views the
damage or know the cause of stopping the train himself. The man who saw the metal piece
tearing the floor of carriage went to call train conductor rather than applying emergency
brakes himself and hence time was wasted. Conductor was about to see the damage in first
carriage, when the train collided into the bridge with a catastrophic incident.

Poor Bridge Design & Set Point Location :

The Bridge into which ICE-884 hit at the speed of 125 mph was not supported by strong
spans anchored to solid abutments on either side, it was in fact supported by just two thin
piers. Also the location of railway track set points (used to change the track of train) was in
proximity to the bridge that caused the accident turning into a big disaster.
Other Related Issues:

Rescue workers found it very difficult to find a way to trapped passengers inside the train, as
they took a lot of time to cut through the rigid aluminum frames and pressure proof glass
windows. Also the passengers were not able to come out by breaking the windows. The
welded parts of carriages were unzipped in crash and resulted in almost complete destruction
of some of the carriages of ICE-884 WCR.

THE RUBBER SPRUNG RAILWAY WHEEL:

Rubber sprung wheels have worldwide been commonly used in vehicles of tramways,
subways and city railways for many decades. Their benefit especially comes from
minimisation of wear and noise development, resulting, e.g., from small radii of the rails and
other speciality of those railway systems. Due to oscillations and noise in the ICE-high speed
trains rubber-sprung wheels were advanced to standard-gage railways and were permitted to
be used for high speed trains by the Deutsche Bahn in autumn of 1992.
A rubber-sprung wheel of ICE-type consists of a wheel tyre, 34 rubber blocks, a wheel rim
and a solid shaft . For the purpose of simple assembly the wheel rim is divided into two parts,
a wheel centre
and a detachable ring. At first the rubber blocks are put in to the gap between the wheel tyre and the
wheel centre equidistantly around the perimeter. Then the detachable ring is bolt together with the
wheel centre. Thereby the rubber blocks are distorted, which mean s they are pre-stressed in axial and
radial direction and can extend in circumferential direction into the existing free spaces. The rubber
blocks are protected against shifting by friction between them and on the wheel tyre respectively the
rim .
So the fully assembled rubber-sprung wheel can be seen as a supporting structure. The
essential load transmission takes place between the point of the wheel contact and the wheel
shaft. Thereby the wheel contact force is transmitted via the wheel tyre into the rubber blocks
and from there into rim and thus into the shaft. The newly developed wheel set type (mar ked
as BA 064) in new condition has a diameter of 920 mm. By wear and the subsequently
necessary re-profiling this diameter notably decrease during service. The limiting
size was appointed 854 mm, whereas the fractured wheel had a remaining diameter of 862
mm. This means, that the thickness of the wheel tyre had decreased from 60 to 31 mm.
According to the UIC-directive 510-5 resp. DIN EN13979-1 especially the loading cases
straightforward driving, rolling turn and track switch crossing can be
distinguished. For the traction operations the contact wheel force is given by
Q =1.25 Q0 =98 kN
where Q0 = 78 kN is the static wheel load, that can be derived from the weight of the wagon.
The side forces are given by
YB = 0.6 Q0 = 47 kN
for the rolling turn and
YW=0.36 Q0 = 28 kN
for track switch crossing. The highest circumferential stress, which is most crucial for crack
propagation, can be found at the inner side of the worn off wheel tyre (i.e., with minimum
diameter) for the loading case straightforward driving.

Numerical stress analysis:


For an accurate numerical stress analysis for the wheel tyre a three-dimensional finite-
element analysis is necessary. This enormous effort is required due to the geometry of the
wheels, the multi-dimensional pre stressing of the rubber blocks and the spatial loading
situation, especially when regarding all loading cases that might occur in service. So, e.g., the
highest stresses develop at the side shoulders of the wheel tyre for rolling turn and track
switch crossing. But even for straightforward driving the stress state becomes three
dimensional.
All following investigations refer to the accident s wheel with a diameter of 862 mm. Fig
shows the finite-element mesh of the ICE-wheel. For reasons of symmetry it is sufficient to
just take the half of the wheel into consideration. The symmetry conditions in the cross
section and the bearings are fulfilled by appropriate kinematic boundary conditions. The
forces defined in act in the upright symmetry plane and have to be cut in halves, as only the
half of the wheel is calculated. In the finite-element analyses of the loading cases the rubber
blocks are fixed at their positions by the form and frictional fit resulting from contact with the
wheel tyre and the rim.
The finite-element model of the rubber-sprung wheel basically consists of the solid shaft, the
wheel centre, the detachable ring, 17 rubber blocks and the wheel tyre. The finite-element
meshes for the wheel tyre, wheel centre, detachable ring and solid shaft are built up with
hexahedral brick elements with 8 nodes and linear functions for the displacements, which are
recommended for contact analyses. The material constants for steel are E = 210000 MPa and
= 0.3. The finite-element meshes for the rubber blocks consist.

of hexahedral brick elements also with 8 node s but hybrid ansatz functions. For the rubber
the material law of MOONEY- RIVLIN with the material constant s C10 = 2.90 MPa and C01
= 0.726 MPa is applied.

The finite-element mesh of Fig. 6 incorporates approximately 130,000 elements with more
than 150,000 nodes. In the FE-system rubber-sprung wheel a triple non-linearity is
inherent. Non-linearity due to the non-linear material behaviour of the rubber blocks.
Geometrical non-linearity due to the big deformations of the rubber blocks. Structural non-
linearity due to the contact between the parts of the wheel. For the finite-element analysis a
two-stage approach is necessary. At first the wheel has to be assembled, and the resulting
stresses of this process have to be determined. Subsequently the stress analysis for the
loading cases can be performed. As consequence of the non-linearity of the system rubber-
sprung wheel the resulting linear equation system with more than 450,000 equations has to
be solved approximately 60 times for the assembly process and about 50 times for each
loading case. By the assembly process a circumferential stress occurs, that is constant in
circumferential direction (except from some minor fluctuations induced by the rubber blocks)
and variable in axial direction due to the unsymmetrical shape of the wheels cross section. At
the roof-ridge of the wheel tyre the circumferential stress is about 50 MPa. The analysis of
the loading case straightforward driving yield positive circumferential stresses that
heavily vary in circumferential as well as axial direction how the stress distribution in the
area of the wheel/rail contact .The highest circumferential stress can be observed at the
roof-ridge in the symmetrical plane of the loaded wheel. For the loading case
straightforward driving for Q = 98 kN a maximum stress Rmax = 220 MPa will develop. As
can be seen in the circumferential stress, which is responsible for the fatigue crack growth,
rapidly decreases in circumferential direction. A minimum can be observed for u = 45 with
Rmin = 6 MPa . In the range between u = +45 and u = +90 the stresses are approximately
equal to the assembly stresses. Shows the circumferential stresses at the inner side of the
wheel tyre for a half turn of the wheel. It can be seen, that for any full turn of the wheel one
loading cycle is passed through. For Q = 98 kN thus a stress amplitude ra = 107 MPa, a mean
stress rm = 113 MPa and an R-ratio = +0.03 is effective. Those maximum stresses occur at the
roof-ridge (y = 67.5 mm). In the loading case straightforward driving the stresses decrease
towards the shoulders of the wheel tyre. Also the R-ratio R = rmin/rmax depends on the y-
coordinate. For the purpose of evaluating the fatigue fracture especially the stress amplitude,
but also the mean stress as well as the R-ratio, are of relevance.

Fig: Circumferential stress at the inner side of the wheel tyre at y = 67.5 mm (roof-ridge) in dependence of the angle u for the
loading case straightforward driving (assembly + Q-load) for a wheel diameter of 862 mm.

Fig: Circumferential stress divided into stress amplitude ra(y), mean stress rm(y) and R-ratio at the inner side of the wheel tyre in
dependence of the y-coordinate.
Fig: Stress amplitude ra in dependence of the wheel contact force Q in comparison to the fatigue strength of the material.

The dependence of the maximum stress amplitude Ra from the contact wheel force Q can be
gathered from Fig. It becomes obvious, that there exists a linear relation describable by
a =c1Q
with c1 = 1.07 MPa/kN and Q in kN . Taking into consideration the assembly process as well
as the wheel contact force, the maximum stress is given by

rmax =c2*c3* Q

with c2 = 49 MPa, c3 = 1,73 MPa/kN and Q in kN.


Furthermore, presents a comparison of the maximum stress amplitude with the fatigue
strength of the wheel tyre, which is determined for the wheel tyre material B5 according to
the FKM-directive . It becomes apparent, that for a wheel tyre with diameter 862 mm and
wheel contact force Q = 98 kN according to the UIC-data sheet a sufficient security against
fatigue failure is given. Even up to a wheel contact force Q = 141 kN (i.e., 1.8Q0) the wheel
tyre is fatigue endurable. This shows, that just on the basis of fatigue strength criteria the
fracture of the wheel tyre cannot be explained. Since at the location of the crack initiation no
material defect could be found, the majority of experts is of the opinion, that the crack
initiation is to be seen as incidental event, which was unforeseeable. However, the fatigue
crack growth can excellently be explained by means of fracture mechanical methods.

Fracture mechanical investigations


All fracture mechanical investigations assume, that a crack already exists within a component
or structure. Such cracks disturb the existing stress field and reduce the load capacity and
lifetime of a structure.
The following investigations will show, that in the case of the ICE-wheel of type BA 064 a
semi-elliptical crack with a depth of just 1.3 mm starts to grow.
Basics of fatigue crack growth:
The criticality of a crack under constant loading can be determined by the stress intensity
factor K, whereas in case of a temporally changing loading situation the so called cyclic
stress intensity factor K is of relevance. It can be calculated by

Thereby is the loading inserted into the structure, a is the crack length and Y(a/w) is a
value, which depends on the geometry of the structure and the type of load application.
.

If a component is subjected to fatigue loading , following questions arise:


1)Is the crack able to propagate?
2)What is the velocity of a propagating crack?
3)At what crack length respectively under which loading will the crack propagation become
unstable?
4)What is the lifetime or the remaining lifetime of the component?
Those questions can be answered by the da/dN-curve, which describes uniquely for any
material the correlation between the crack growth rate da/dN (increase of the crack length per
number of cycles of the oscillating load and the cyclic stress intensity factor DK. So crack
propagation is possible,

if K is in the range between Kth and Kc:


Kth < K < Kc
In this formula Kth is the Threshold-value of fatigue crack growth. For K < Kth a crack
is unable to propagate. Kc is the cyclic stress intensity, at which fracture, i.e., unstable crack
propagation will occur. The crack growth rate, that results from an actual local crack loading
situation K can directly be derived from the diagram in fig The remaining lifetime might be
obtained by integration of the da/dn-curve.
Fatigue crack growth under cyclic load with constant amplitude is fundamentally different
from that under cyclic load with variable amplitudes. Under constant amplitude loading an
extremely continuous crack propagation behaviour can be observed, whereas the interaction
effects resulting from variable amplitudes produce a very discontinuous crack growth.
In an event of damage the crack surfaces give valuable information about the precedent crack
growth and the actual loading situation of the structure. In this way it is possible to clearly
distinguish between the area of stable crack propagation and unstable final rupture. If one
observes a small area of fatigue and large area of rupture, it can be assumed, that the structure
was subjected to a very high loading and vice versa. Additionally on the fatigue surface it can
be determined, whether the loading was applied with constant or variable amplitude, since as
consequence of variable loading arrest marks as well as colour effects are visible.

Analysis of the fracture of the wheel tyre:


Under cyclic loading (fatigue loading) the overall lifetime of components and structures is
assembled from the period of crack initiation and the period of crack growth. For an equally
loaded structure with a smooth surface in general a long period of crack initiation and a small
period of crack propagation can be observed. If otherwise a structure with a rougher surface
(e.g., by material failures, etc.) is subjected to a more unequal stress distribution, the period
of crack propagation becomes relatively larger in an overall reduced lifetime.
As for the rubber-sprung wheel the inner side of the wheel tyre does not contain any notches
in circumferential direction, every point of the wheel along the perimeter is subjected to the
same cyclic loading, which repeats in every turn of the wheel. Systematic investigations of
the broken wheel tyre have proven, that no surface or material failures existed at its inner
side. Therefore it can be assumed, that the period of crack initiation was very long. However,
this does not mean, that the period of crack propagation was short, but just, that it presumably
was shorter than the initiation phase. The crack propagation of the ICE-wheel started near the
location, where the maximum stresses are observed in the loading case straightforward
driving at inner side of the wheel tyre. In the very beginning of the propagation the crack
predominantly extended into the depth, while later on propagated with a half-elliptical shape.
In this particular phase the crack growth rate along the surface was notably higher than into
the depth. Final rupture of the wheel tyre did not occur, until approximately 80% of the cross
section already was damaged.
On the fracture surface one can observe arrest marks, colour effect and further fracture
structures, that indicate a very discontinuous crack growth. So, e.g., phases of slow
propagation or even temporary crack arrest alternate with phases of fast propagation.
Altogether a very extended period of stable crack propagation is most likely. The fracture
surface clearly shows, that the material of the wheel tyre is very tolerant against crack
growth. In many components and structures just after relatively small fatigue cracks unstable
crack growth will occur. Different from this in the ICE-wheel the stable crack surface covers
a cross section of approximately 80% before the crack became unstable. Those considerations
are even supported by the fact , that the fracture toughness of the wheel tyres material Kc =
86.8 M Pa m1/2 is quite high. On the basis of numerical and experimental simulations now
the fatigue crack growth of the wheel tyre is further investigated.

Numerical simulation of the fatigue crack growth:

The fatigue crack growth in three-dimensional structures can be simulated with the help of
the finite-element-method. Especially the program system ADAPCRACK3D has proven its
value for such simulations. It consists of three modules, which in combination comprise a
fully automatic crack propagation simulation. As input objects basically a description of the
un cracked structure in terms of a 3DFE-mesh, a description of the initial crack and some
material data for the definition of the crack propagation behaviour are required.

For the purpose of simulating the crack grow thin the ICE- wheel tyre with a diameter of 862
mm a segment of the un cracked wheel tyre was discretised with a number of 52,000
elements. In a first step initial cracks of different length and depth were introduced into the
model in order to determine an appropriate initial crack size for the propagation simulation.
By those pre-test a semi-circular initial crack with a radius r = 1.5 was identified to be best
suited for the automatic simulation, as for this crack size and shape the stress intensities
exceed the Threshold-value Kth = 8.2 MPa m1/2 along the whole crack front. Because the
fatigue crack growth of the broken wheel tyre did not start at the roof-ridgebut about 13
mm shifted, the initial crack is positioned accordingly. The simulation is performed for a
wheel contact force Q = 98 kN. Thereby for any loading cycles, which equals one full turn of
the wheel, a certain crack growth can be identified. The whole crack growth simulation
consist of 26 simulation steps and is carried out until the fracture toughness Kc = 86.8 MPa
m1/2 is reached. The crack fronts, that are calculated from step to step, are shown in fig. In the
beginning the crack keeps its semi-circular shape, but in the following it rather develops a
semi-elliptical shape with a higher crack growth rate along the surface. Immediately before
occurrence of instability the crack has a depth of 31.7 mm and a maximum length at the
surface of 71.7 mm. Fig presents a comparison of the.

simulated crack front s with the real crack surface. It becomes apparent, that the numerical
simulation is excellently qualified to predict such events of da mage. Also from the sim
ulation it can be concluded, that final fracture of the wheel tyre happened only after an
extended stable crack growth with just a very small remaining rupture surface.
Under assumption of an initial crack length of 1.5 mm and a cyclic constant amplitude load
with a wheel contact force Qmax = 98 kN the fatigue crack growth lifetime of the wheel tyre
can be calculated under consideration of the crack growth rate curve of Fig to be
approximately 1.4 million cycles. Because during the simulation a constant amplitude loading
is assumed, this lifetime is much smaller than in reality. The wheel is exposed to variable
amplitude loadings, which cause a higher number of cycles due to retardation effects. This is
also observable on the fracture surface, which shows distinct signs of a discontinuous crack
growth. So the real crack propagation lifetime can rather be determined by experiments.
FACTORS OF FAILURE:

There were four factors in the use of this wheel on the high speed train that contributed to its
failure in this particular application:

1. Higher speed, resulting in a higher strain rate. Even though the effective yield strength,
tensile strength, ductility and fatigue life (number of cycles to failure) would most likely have
been slightly increased, the number of cycles per second was much higher, resulting in a
shorter time to failure.
2. Higher load, leading to greater strain. A larger, heavier train meant that there was a greater
amount of flexure of the wheel rim. So the initial strain at peak load on the wheel during each
cycle was higher so the initial strain and stress amplitudes were higher.
3. Faster wear, resulting in increasing strain amplitude over time. As the wheel rim wore
more quickly, it became thinner and less stiff, resulting in even greater flexure, stress and
strain with each cycle.
4. Fatigue crack initiation and growth in a spot that could not be detected in visual
inspections. The most important takeaway is that even if design is proven to work at low
speed, it may not work at high speed without extensive modifications. Higher speeds can
greatly accelerate the failure process, at a rate disproportionate to the increase in speed.

SUMMARY AND CONCLUSION:

The fracture of a wheel tire which had brought about the accident of Eschede clearly was
caused by a fatigue crack at a relatively low mean stress level. The final fracture only
occurred after the remaining cross-section had dropped to about 20%. The material
corresponded to all specifications; no traces or pre-existing flaws which could have initiated
a fatigue crack could be detected. Service loads had never been determined for this type of
wheel, train and track. The impact factors which had been measured at a so-called comfort
measuring point can only serve as approximate indicator.Before the wheel was commissioned
there were performed strain measurements on a prototype of the wheel, fatigue tests on
complete wheels and extended service tests on rail tracks. This procedure corresponded to the
UIC-recommendations in force at that time. Seen from todays view-point facilities for
design would be better. For this type of rubber-sprung wheel no individual UIC-guideline
existed. The recommendations for monobloc wheels nevertheless could correspondingly also
be applied here.After the accident fatigue tests have been carried out on whole wheels and on
segments of tires which gave as result that a fatigue load of nearly four times the nominal
wheel load would have been necessary to produce a fatigue crack without pre-existing flaw,
which is practically impossible because the tire comes into contact with the wheel disc
already at wheel loads of 22.5-times the nominal wheel load. When the segments were
tested, a fatigue crack could be produced only with an initial flaw at least nearly 1 mm deep
and a load 2.5-times the nominal load, without notch only at about 7times the nominal load.
Since no pre-existing flaw at all had been found at the crack origin, the fatigue crack must
have started from other reasons.
A calculation of stresses in the tire using finite elements showed the stresses to be not
significantly high at nominal loading . Because such calculations would have been useful
primarily for optimisation of design details and a validation of their results would have been
necessary anyway by measuring strains, which had been done in this case before
commissioning of the wheels, the design process cannot be criticized in this point.
A calculation of in-service integrity based on the stresses calculated with loads according to
UIC has proven a sufficient safety of fatigue cracking. Nevertheless it was disputed whether
such a calculation would have be en relevant at all be cause the railway companies demand a
fatigue proof design. On the approximately 100 wheel s which had been taken out of service
before the accident no cracks or flaws have been reported when they were disassembled an d
scrapped. On the approximately 5000 remaining wheel s remove d after the accident several
cracks were found in tires but no fractures . The cracks had occurred also in wheel s with
larger diameters than that of the accident an d without pre-existing flaws. This supports the
presumption of a rare or singular event which could have initiated the crack. It also supports
the assumption of a generally quite low mean loading level be cause these cracks, which were
detected late r, had grow n only very slowly or not at all and never reached a critical depth.
The nature of such a rare event can only be specula ted about; in the official investigation of
the case such topics were not brought up and be cause this wheel type is no longer used for
high-speed trains no further research was done in this direction.
The case has shown as topics for improvement that modernisation of de sign rules or
international standards is necessary, that realistic measurement of service loads and impact
factors should be carried out previous to design work, that the weak spots need identification
by tests on roller type test stands until fracture and non-destructive test of these weak spots
ought to be done in the regular service checks.