1 Wheel-Rail Contact: GETTING THE RIGHT PROFILE Simon Iwnicki, Julian Stow and Adam Bevan Rail Technology Unit Manchester Metropolitan University The Contact The contact patch between a wheel and a rail is typically the size of a 5p piece and its shape depends on the geometry of both bodies. The theory developed by Hertz (when he was a 24 year old research assistant!) can be used as a good approximation and this gives an elliptical shape and distribution of the normal force. 2a 2b
21
Embed
GETTING THE RIGHT PROFILE - pudn.comread.pudn.com/downloads151/ebook/656142/Wheel-Rail_Contact_GE… · FE model of wheel rail contact Courtesy Tom Kay Corus Rail technologies. 3
This document is posted to help you gain knowledge. Please leave a comment to let me know what you think about it! Share it to your friends and learn new things together.
Transcript
1
Wheel-Rail Contact:GETTING THE RIGHT PROFILE
Simon Iwnicki, Julian Stow and Adam Bevan
Rail Technology UnitManchester Metropolitan University
The ContactThe contact patch between a wheel and a rail is typically the sizeof a 5p piece and its shape depends on the geometry of both bodies.
The theory developed by Hertz (when he was a 24 year old research assistant!) can be used as a good approximation and this gives an elliptical shape and distribution of the normal force.
2a
2b
2
Alternative contact modelsIn practice the geometry at the contact patch is often more complicated and the contact patch isfar from elliptical. Possible alternative models are:
• Hertz theory in strips
• Finite elementsCan give good results but very slow
where: Wn = the wear numberA = the contact patch areaµ = the coefficient of friction between the
wheel and railTx,Ty = longitudinal and lateral creep forcesγx γy = longitudinal and lateral creepages
[ ]xxyy γTγT0.6µ
A1Wn +=
Wear models
• Archard wear model:V = k1Fns/H (m3)
where k1 is a non-dimensional wear coefficient, H is the hardness of the softer material, Fz is normal contact force, s is the sliding distance in the contact patch.
11
-a
0
a
-b
0
b
0
0.5
1
1.5
2
x 10-11
xy
δz
Local wear rate in the contact patch(Archard model)
Simulated and measured wheel profile and wear distribution example 2 - after 128000 km
An Example:
The WRISA2 anti-RCFwheel profile
(development funded by RSSB)
13
Contact position simulation for c2cEast Ham Depot - Contact Stress vs Position
0
500
1000
1500
2000
2500
3000
3500
4000
4500
-790 -780 -770 -760 -750 -740 -730 -720 -710
Distance /mm
Con
tact
Str
ess
/ M
N/m
^2
class 357class 312S i 2
Contact Conditions – P8 Profile
14
New P1 Wheel
on New BS113a Rail
No contact in these positions
Heavy Worn P1 Wheel
on Worn BS113a Rail
Contact Conditions – P1 Profile
P1 wheel on c2c
15
The WRISA2 anti-RCF profile
RCF Sensitive Region
Anti-RCF ‘Relief’
WRISA2 profile designed by NRC to:
•Include an Anti-RCF ‘relief’ in flange root area of the profile
•Reduce the contact stress in mid-gauge region of rail
•Reduce the rolling radius difference when curving to reduce longitudinal creepages and tangential forces
•Include a smoother tread run-off to eliminate geometric stress raisers caused by transition.
New P1 (Red), New P8 (Green), Light Worn P8 (Blue), WRISA2 (Purple)
WRISA2 Rolling Radius Difference
16
WRISA2 - Shakedown Exceedence
How does Nature solve these problems ?
(Which is the best beak ?)
17
Genetic Algorithm
Basic Method - (simulates natural evolution)• Profile digitised and converted to a binary ‘gene’• 2 parent profiles selected and their genes ‘mated’
to produce ‘child’ genes• Child genes converted back to profiles• Simulation used to evaluate each child profile• Best profiles selected as parents and process
repeated • More details in Persson and Iwnicki paper
presented at 18th IAVSD conference 2003
The genesx: 12.459 mm
y:5.223mm
x: 12.909 mm
y:6.201 mm
X: .....11010101
01110101
10011011
11001100
... 1101010101110101100110111100...
18
Mating
1101011010111000110101110101100110111100...
1010101011000101010101110101100110111100...
Bits are copied with random switches between the parents
1101011011000101010101110101100110111100...=
+
Reconstructing the new profile
x: 12.459 mm
y:5.223mm
x: 12.909 mm
y:6.201 mm
X: .....
11010101
01110101
10011011
11001100
... 1101010101110101100110111100...
19
Survival of the fittest• Each profile is assessed by running a vehicle simulation over a typical track case.
•The best profiles become the parents for the next generation.
• Mutations can be added to avoid local minima.
Some interim results
20
Assessing the new profiles
The Penalty Index - the sum of sub factors:
• Ride comfort penalty factor • Lateral track shifting force penalty factor• Maximum derailment quotient penalty factor• Wear penalty factor• Maximum contact stress penalty factor
(Other factors are possible and each can be weighted according to importance)
Results - ‘stiff bogie’
21
Results - ‘soft bogie
Conclusions• The design of wheel profiles for the conflicting requirements of
stability and derailment resistance as well as low wear and resistance to rolling contact fatigue is a significant engineering challenge
• Rules of thumb related to tread cone angle and flange angle havelimited usefulness
• Various tools are available to predict contact conditions and possible problems such as RCF and wear
• A novel method using a genetic algorithm together with a penaltyindex and a dynamic simulation can help to optimise profiles for a particular vehicle or track case.