M.

GRDAMAND MOHAMMED
[email protected]

Lecture 4
Summery of Lecture 3
 SUB-BALLAST AND SUBGRADE
1. Effect on Ballast Properties (Confinement)
2. What is Sub-ballast?
3. Alternatives
i. Geogrid
ii. A Stabilised or Bound Foundation
iii. Improvements to Existing track
iv. Mini-piling
4. Design of Ballast/Sub-ballast thickness
 INHERENT TRACK QUALITY
1. Track Variability
2. Drainage
3. Critical Velocity
 ‘BEAM ON ELASTIC FOUNDATION’ ANALYSIS

2
Lecture Contents
 SUB-STRUCTURE STRESS ANALYSIS
 TRAIN DYNAMICS

3
 TRACK ANALYSIS

• Substructure Analysis
Analysis of ballast, subgrade etc. is extremely complex and
therefore some sort of simplification is required for practical
purposes.
A sensible simplification - use a load spread angle

Suggest use 𝜽 = 35 degrees [see below]

This is reasonably conservative for a

structure which is stiffer at the top than the
bottom.


4
 TRACK ANALYSIS
• Don’t forget that below a certain depth, the material can ‘feel’
stress from more than one sleeper,
• So it will be necessary to combine in some way.
• There are no rules; the whole analysis is approximate, but it is
useful to get some idea of the stresses at top of subgrade level.

5
 TRACK ANALYSIS
Substructure Analysis
Boussinesq’s solution for displacement under a circular plate:

P
Suggest use  = 35 degrees
r
Boussinesq: E = P (1 - 2) / 2 r y
or: y = P (1 - 2) / 2 r E
where P = load, r = radius, y = deflection
At a load spread

angle of :
Elastic half y =  (P/(.(r+z.tan)2.E).dz
0 
space = (P/E).[-1/(tan.(r+z.tan))]
0
E, ν = (P/E)/(tan.r)
Equate: tan = 2 / ((1 - 2). )  0.7
   35 degrees
TRACK ANALYSIS

Substructure Analysis

Suggest use  = 35 degrees

e.g. if stress at base of sleeper = 150 kPa

and if sleeper is 2.5m by 0.28m
then stress at 300mm depth
= 150  (2.5  0.28)
 (2.5+2.3 tan35)  (0.28+2.3 tan35)
= 52 kPa
TRACK ANALYSIS

Substructure Analysis

Suggest use  = 35 degrees

Estimating deflection:
Convert area of sleeper to equivalent circular
area and then use integral
e.g. 2.5m × 0.28m  r = [2.5×0.28/π]0.5
= 0.472m

 
y = 0  (P/(.(r+z.tan)2.E).dz 
= (P/E).[-1/(tan.(r+z.tan))]
= (P/E)/(tan.r) 0

e.g. if P=100kN, E=80MPa: y = 1.204mm

DYNAMICS – Horizontal

1. Wheel and Rail Shape

The top of the rail is horizontal, matching

the shape of the wheel. The flange of the
wheel contacts the side of the rail and
Flange provides the required guidance.

This system is fine for low speed travel, but problems were
encountered with impacts between the wheel flange and the
rail as soon as faster trains were developed.

9
 DYNAMICS – Horizontal
• The solution:
The top of the rail is inclined and the face of the wheel is
conical to match. The flange still performs its guidance function
but the result is a much better ride than before Klingel Motion.

Conicity 

Conicity 𝛾(= due to tangent of angle)

10
DYNAMICS – Horizontal

Doesn’t work; too much banging from side to side

Flange
Conicity 

Much better; a nice smooth ride

A real wheel shape

 DYNAMICS – Horizontal
2. Klingel Motion
• The distance between the inside faces of the rail is slightly greater
than the distance between the faces of the wheel flanges.
• When the axle is slightly offset to one side, the points of contact
between wheel and rail will be at slightly different positions on the
wheel.
• This gives two different wheel radii.

Wheel 1 Wheel 2
r1 r2

s
12
 DYNAMICS – Horizontal
If: r = average wheel radius
 = conicity (≈ angle of wheel face to horizontal)
s = gauge (≈ distance between contact points)
If the axle is displaced to the side a distance y from its average
position:
And: The axle is displaced sideways by y
Then: Wheel 1 has an effective radius r1 = r – y
Wheel 2 has an effective radius r2 = r + y

Both wheels obviously still rotate at the same rate  one goes
faster than the other  curved motion.

13
DYNAMICS – Horizontal

The radius of curvature of the motion R can be

found by equating the ratio of the radius of the outer
wheel (R + s/2) to that of the inner wheel (R – s/2)
to the ratio of the two speeds of travel.
R
DYNAMICS – Horizontal

From geometry: (R + s)/R = (r + 2y)/r

Therefore: s/R = 2y/r

R
But: 1/R = K = curvature of motion = – d2y/dx2

So: d2y/dx2 = – 2y/rs

Solution: y = C sin { √(2/rs) × x } KLINGEL MOTION

 DYNAMICS – Horizontal
• So Klingel Motion produces a sinusoidal oscillation about the
mid position. The amplitude can be anything and is given by the
constant in the equation. The wavelength is:

Wavelength = 2π /√(2/rs)

• The distance between the rails is generally 10-20mm greater

than the distance between wheel flanges. This sets the limit to
the amplitude of Klingel Motion. The result is something like:

Limits

16
 DYNAMICS – Horizontal
Typical track parameters: s = 1.5m
r = 0.4m
𝛾 = 0.05
Wavelength ≈ 15m
At a speed of 30 m/sec (108 km/hr) the frequency is about 2 Hz.
• Now, what happens when the train reaches a corner?
 No problem at all. The mid point of the oscillation simply shifts
a short distance transversely.

17
 DYNAMICS – Horizontal
• Now the average position is slightly off centre, meaning that
one wheel will, on average, have a greater diameter than the
other.
• Result: perfect cornering with no contact between flange and
rail.

• e.g. For a curve of radius 1km, the outer wheel set should have
an effective radius which is proportionally s/1000 greater than
the inner set. The inner one might be at 0.3997m radius and
the outer one at 0.4003m [.3997 × (1+1.5/1000)]. At a conicity
of 0.05, this requires a transverse shift in position of 0.003m ÷
.05, which is 6mm – quite possible.

18
 DYNAMICS – Horizontal
3. Real Motion
In reality, both rails and wheels deform with millions of repeated
loads. This means that, after a time, rails will not have flat
surfaces inclined at a steady 0.05 slope neither will wheels have a
perfectly conical face.

Worn rail and wheel shape:

19
 DYNAMICS – Horizontal
3. Real Motion
This means that the effective conicity, the conicity at the point of
contact, varies continually depending on just where the wheel is in
relation to the rail.

This actually doesn’t upset

Klingel motion too much as
long as the effective conicity
never gets too big.
In fact it may give a greater
range of corners which can
be handled without flange to
rail contact.
20
 DYNAMICS – Horizontal
• Hunting is the phenomenon produced when the amplitude of
Klingel motion becomes excessive and the flanges start to make
contact with the rails.
• This might be produced by a tight bend, uneven track, worn rails
etc.
• The result is a relatively violent motion with much higher
frequency oscillation.
• This is a nuisance for passengers and is damaging to both track
and train components.
Limits

Hunting movement Path of Klingel motion if

limits were not present
21
 DYNAMICS – Horizontal
• Hunting
Limits

Hunting movement Path of Klingel motion if

limits were not present

In the above sketch, the frequency of hunting movement would be

about twice that of Klingel motion, but it would also be much less
smooth, with relatively violent decelerations at each flange / rail
contact. Hunting is bad news. It can be thought of as Klingel motion
with very high values of conicity.

22
 DYNAMICS – Horizontal

Wavelength = 2π /√(2/rs)
Limits
e.g. 2π/√(2×0.05/0.4×1.5)
Klingel Motion ≈ 15m

Represents contact between

wheel flange and side of rail
Hunting Limits

Hunting movement Path of Klingel motion if

limits were not present
DYNAMICS – Vertical
• These are important and
become worse if the track is
in poor condition.

Suspension system

An uneven track leads to uneven forces and therefore uneven

deterioration
24
DYNAMICS – Vertical

The wheels follow the track, whereas the main mass of the wagon tends to
continue uninterrupted. This will result in variable compression of the
suspension system, variable force transmitted to the track and, eventually,
vertical oscillation of the main mass of the wagon as well.
25
DYNAMICS – Vertical

26
 DYNAMICS – Vertical
• Poor track can lead to contact forces double those expected due
to dead weight.
Simplify it to a single spring with a mass

M
(= body of vehicle)
Suspension y
system
F = k(y-z) z
(= rail level)

27
DYNAMICS – Vertical

The resulting movement of the mass has two components, one due to the
suspension system itself and the other due to the imposed vibration from
beneath.
28
DYNAMICS – Vertical

8
Vertical Displacement of Vehicle (mm)

Track defect 4

2
frequency and
0
suspension system -2
frequency can -4
interact -6

-8
0 0.5 1 1.5 2 2.5 3

Time (seconds)
29
 DYNAMICS – Vertical
• NOTE: If MC1 2 /k = 1; i.e. C1 = √(k/M); in theory there would be
an infinitely large vibration amplitude!
• In an ideal case, a smooth track would lead to zero vertical
dynamic forces and zero consequential damage.

• In a real case, unevenness in track causes a dynamic response

from the train as shown in the previous section.
• The problem here is one of resonance. Fortunately real defects
are less regular, so this will tend to avoid actual resonance, but
there will still be a worst wavelength for any given train speed
and suspension hardness.
• In practice, this wavelength is typically 20 - 30m for goods
wagons, more for passenger wagons.
30
 DYNAMICS – Vertical

If some sort of resonance occurs, the result is repeated heavy

loading at particular points on the track.
250

Displacement / Acceleration
In theory you could 200
maximum
displacement (mm)
maximum
get resonance acceleration(m/sec2)

150

100

50

0
The result is 0 10 20 30 40
Rail defect wavelength (m)
uneven load:

Load
 DYNAMICS – Vertical
These high load points will tend to settle more quickly than others.
Track becomes more uneven – just at the wrong wavelength!
Dynamic loads become even worse.

Result:

Accelerating
deterioration as
deformation leads
to increasing
Rapid deterioration dynamic loads
immediately after
tamping Time
 DYNAMICS – Vertical

Modelling effects

Primary suspension
Secondary suspension

Secondary suspension (+
damping)
Primary suspension (+
damping)
DYNAMICS – Vertical

Modelling effects

Varying level
Varying subgrade modulus
Varying deformation susceptibility

Each vehicle pass delivers a pattern of load

 Individual settlement of each sleeper
 Changed set of levels
DYNAMICS – Vertical

Modelling effects
3
Perfect
Standard deviation (mm)

2.5 Very good

Good
2 Moderate
Poor
1.5

0.5

0
0 100000 200000 300000 400000 500000 600000
Number of passes
 DYNAMICS - Corrugation
Corrugation Creating Forces
• The other dynamic effect to consider is a very high frequency
one, due to oscillations within the wheels and rails themselves.

36
 DYNAMICS - Corrugation
• This is a very complex effect.
• Calculations are possible but not reliable.
• Certain lengths of track are susceptible to this sort of oscillation;
others not.
• The result is a frequency of oscillation in the order of 1000 Hz or
more.
• This means that as the train travels along, the wheel / rail
contact force changes at very high frequency, as well as any
much lower frequency effects due to track unevenness.

• e.g. A train at 60 m/sec; oscillation at 2000 Hz –> stress pulses

every 30mm

37
 DYNAMICS - Corrugation
• Since rails deform under load (see later in notes), the result is a
minute additional indentation every 30mm. Once the pattern is
established, subsequent wheels tend to apply the stress pulses
in exactly the same places and eventually the indentations
become noticeable.
• They are called CORRUGATIONS and are a maintenance
headache.

38
 DYNAMICS - Corrugation
Solution: Grid the rail

39
 DYNAMICS - Corrugation

Solution: Grid the rail

40
 DYNAMICS - Corrugation
Solution: Grid the rail

41
DYNAMICS - VEHICLE/TRACK
INTERACTION
1. The Effect of Longitudinal Track Geometry on the Vehicle
Treatment in the previous section was highly idealised. In reality,
variation in rail level is never nice and sinusoidal. There is also a
considerable degree of damping in the system.
Reality:

42
DYNAMICS - VEHICLE/TRACK
INTERACTION
There is also the issue of the two rails doing different things:

That’s bad enough, but when the two rails do different

things:

43
 DYNAMICS - VEHICLE/TRACK
INTERACTION
Train swaying from side to side
= a common cause of
derailment

The worst scenario of combined forces is the case where a high horizontal force
(due to hunting) occurs at the same time as a low vertical force (due to poor track
geometry). This is where derailment is most likely.
DYNAMICS - VEHICLE/TRACK
INTERACTION
2. The Effect of non-uniform Trackbed Stiffness
A track can be perfectly aligned with no irregularities in geometry
BUT still generate dynamic problems because of non-uniform
stiffness.
Hanging sleeper Gap begins to appear

Soft Soil

Unloaded rail profile

Path through space of wheel rail contact point

45
DYNAMICS - VEHICLE/TRACK
INTERACTION
• As the train load goes by, the rail deflects.
• It will deflect more where the ground is soft; the result is an
irregular motion.
• The result is exactly the same as if the track was uneven.
• Greatest loads will occur at the points where the track deflects
most since that is where the vertical acceleration will be
greatest.
• This will simply make the situation worse.

46
DYNAMICS - VEHICLE/TRACK
INTERACTION
3. Role of Pads

47
…next lecture
 OTHER FORCES ON THE TRACK
1. Thermally Induced Forces
2. Cornering Force
3. Wind
4. Traction and Braking
5. Wheel Flats
 WHEEL / RAIL CONTACT

48