Prediction of wheel profile wear and rolling contact ... · Prediction of wheel profile wear and rolling contact fatigue for the Stockholm commuter train Babette Dirks 1, Roger Enblom

117th Nordic Seminar on Railway Technology

3-4 October 2012

Prediction of wheel profile wear and rolling contact fatigue for the Stockholm

commuter train

Babette Dirks1, Roger Enblom1,2

1Royal Institute of Technology, Stockholm - Sweden2Bombardier Transportation, Västerås - Sweden

2

Overview

• Project goals• Scope of this presentation• Methods• Results• Conclusions

3

Project

• The main goals of the project are:o To create one model for prediction of the total

expected life of wheels and railso Selection of reference vehicles, lines and curves

for validation of the modelso Perform/collect measurementso Validation of the modelo Apply the model to investigate the influence of

different parameters (wheel/rail profiles, vehicle suspension, axle load, track condition etc.)

4

Scope of this presentation

• To study the behavior of two reference vehicles with respect to wear and RCF of the wheels.o Two wear and two RCF prediction models

have been used in combination with vehicle dynamics simulations.

o Multi-body simulations in Gensys provided the input to the wear and RCF models

5

Methods• Two reference vehicles, running on the Stockholm

commuter network, have been selected

•2,6 m

Vehicle B

Vehicle A

•2,4 m•2,7 m

6

Methods – RCF models

• Two RCF initiation prediction models have been studied and compared:o 1) based on the shear stress (SI-model)o 2) based on the energy dissipation (DI-model)

7

Methods – RCF models

• Surface initiated RCF index (SI) of the form:

0)( >−= kSI τ

•τ is the shear stress [N/m2]

•k is the yield stress in shear [N/m2]

•p is the contact pressure [N/m2]

0)(>

−=

p

kFI

τ•Shakedown map

8

Methods – RCF models• Rail RCF model based on Tγ (DI-model)

yyxx TTT γγγ ⋅+⋅=

0 100 200 300-15

-10

-5

0

5

10

15

Tγγγγ [N]

RC

F D

am

age Index (*1

0-6

)

•Tx is the longitudinal creep force [N]

•Ty is the lateral creep force [N]

•γx is the longitudinal creep [-]

•γy is the lateral creep [-]

9

Methods – wear models

• Two wear prediction models have been studied and compared:o 1) wear model according to Pearce and Sheratt

(based on the energy dissipation)o 2) Archard’s wear model

10

Methods – wear models

γ⋅⋅= TCAW

0 50 100 150 200 250 3000

2

4

6

8x 10

-10 Wear according to PSH, one wheel revolution

T γγγγ [N]

Wear [m

2]

•Tγγγγ [N]

•AW[m

2]

•Wear for one wheel revolution

AW=worn-off area per wheel revolution

T=creep force

γ =creepage

• Pearce and Sheratt (PSH) wear model

11

Methods – wear models

• Archard wear model (AR)

H

sNkVw

⋅⋅=

Vw = wear volume s = sliding distance N = normal forceH = hardness k = wear coefficient ∆z = wear depthpz = contact pressure

H

spkz z ∆

∆⋅

⋅=

12

Methods – wear models

Wear coefficient, k (dry) [10-4]

0,0

0,5

1,0

1,5

2,0

2,5

3,0

0 0,2 0,4 0,6 0,8 1Sliding velocity [m/s]

Pre

ssure

[GPa]

k3 = 30-40k 2

= 1

-10

k 4 =

1-1

0

k1 = 300-400

13

Results - curving

0 200 400 600 800 1000-2

0

2

4

6

8x 10

-3

Position [m]

Angle

of att

ack [ra

d]

Long transition curve

A axle 1

B axle 1

B axle 3

•(R=∞ -> R=300 m)

•V

14

Methods - simulations

Rm [m] Vvehicle [km/h] Rail profiles % of Ltot

338 60 1, 2, 3 2.4

432 74 1, 2, 3 2.7

574 92 1, 2, 3 8.0

676 98 1, 2, 3 6.7

895 113 1, 2, 3 6.0

1204 120 1 13.5

2035 120 1 15.5

Straight 120 0 44.9

15

Results - RCF

-30 -20 -10 00

1

2

3

4x 10

12

Ywheel

[mm]

SI [N

/m2]

A

B

B3

-30 -20 -10 00

0.05

0.1

0.15

Ywheel

[mm]

DI [-]

A

B

B3

• Calculated RCF damage on wheel profile

16

Results - RCF

• Limitation of the creep forces for high creepages (full slip) for SI-model.

0 0.002 0.004 0.006 0.008 0.010

0.2

0.4

Inner wheel

Creep [-]

µµ µµ=F

T/F

Z [-]

00 >−⋅= kpF

FSI

z

T

yyxx TTTDI γγγ ⋅+⋅=:

17

RCF inspections vehicle B

•High lateral creep forces

18

Results - wear

• Calculated wear depth on wheel profile

-30 -20 -10 00

0.5

1

1.5

x 10-4

Ywheel

[mm]

dz A

R [m

]

A

B

B3

-30 -20 -10 00

0.5

1

1.5

x 10-4

Ywheel

[mm]

dz P

SH

[m

]

A

B

B3

19

Results - wear

• Wear map for single contact in curve

0 0.5 10

0.5

1

1.5

2

2.5

vslip

[m/s]p [G

Pa]

R676, vehicleB

0 0.5 10

0.5

1

1.5

2

2.5

vslip

[m/s]

p [G

Pa]

R676, vehicleA

•k3=30-40

•k2=

1-1

0

•k3=30-40

•k2=

1-1

0

20

Results - wear

• Calculated wear depth dz for single contact

21

Conclusions

• The following main conclusions can be drawn for the RCF prediction models:o Both RCF models predict more damage for vehicle B

than for vehicle A due to the better steering performance of vehicle A

o Under poor adhesion conditions, however, the models behave differently: • The SI-model predicts less damage for high

creepages, due to the independence on creepage• Previous research, however, has also shown that

high creepage has no effects on RCF life.

• The RCF inspections of the wheels of vehicle B show that the steering of the axles under certain circumstances can be poor.

22

Conclusions

• The following main conclusions can be drawn for the wear prediction models:o Both wear models predict more wear for vehicle B

than for vehicle A due to the better steering performance of vehicle A

o The Archard’s wear model predicts more wear due to the large influence of the sliding velocity in the wear map, therefore, especially for vehicle B.

23

Thank you for listening!