A Study on Prediction of Wear A Study on Prediction of Wear Progress of Railway WheelProgress of Railway Wheel

“Long-term endurance test” running at 240 km/h(from Sept. 1983 till Sept. 1984)

Morioka

Omiya

on Tohoku Shinkansen line

(465 km long)

Open on 1982(JNR)

Ueno

1

Background

Katsuya Tanifuji Niigata University

Japan

Test trainTest train((Type 925 electric and track inspection train)

q

2

6-car train

Test runningPeriod: 1 yearRunning distance:

total 268 000 km

Bogie features○ with side bearers○ wheel of 1/40 conical tread

Niigata University

Wheel shape and object parameters Wheel shape and object parameters

Tread wear

Flange wearδ, (WF)Flange angleαF

New

Neutral point

Worn WT Tread shape

3

Niigata University

Comparison of flange wear Comparison of flange wear among wheelset positionsamong wheelset positions

4

=268x103 km

30x103 km

Niigata University

Tread wearWT

Flange wear

ρT

Radius of curvatureWF

New

Flange wearFlange wear

WF at the height of HF=21 mm

Flan

ge w

ear

WF

mm

5

δ

Niigata University

Tread wearWT

ρT

Radius of curvatureFlange angleαF

New

Flange angleFlange angle

Mean value of 48 wheels

6

Niigata University

Profile of Profile of tread weartread wear

ρTFlange angleαF

Tread wear shape

ρT

Radius of curvatureFlange angleαF

New

Mean value of 48 wheelsW

T

7

Niigata University

ρTFlange angleαF

Tread wearWT

ρT

Radius of curvatureFlange angleαF

New Tread wearTread wearat neutral pointat neutral point

WT

8

Niigata University

Flow chart for wear predictionFlow chart for wear prediction

Start

End

SIMPACK

Vehicle model

Simulation onmain line & cross-over

Wear process fortread & flange

Reach to target distance?

Update ofwheel profile

Data ofvehicle & track

Data ofwheel profile

Yes

No every 10000 km

9

Niigata University

268000km走行後

新品形状NewWorn

( )1 1 2 2T TWI k T Tγ γ= +・ Tread wear（Elkins & Eickoff）

T1,T2： Creep force： Creepage1 2,γ γ

Wear indicesWear indices

・ Flange wear（Heumann）

( )F F F FWI k Fμ ψ= ： Coefficient of friction ： Flange force： Attack angle

FμFFψ

10

Niigata University

Wear process on treadWear process on tread

Running simulationwith SIMPACK

① Contact position and wear index

30

40

50

60

70

80

90

100

-5 0 5 10 15 20 25 30Time [s]

Posi

tion o

n w

heel [m

]

Contact position

0

5

0 5 10 15 20 25 30Time [s]

WI FW

I T

Wear index

Whe

el

11

Niigata University

[ ][ ]

( ) [ ]

0.25 / : 100

25.0 / :100 200

1.19 154 / : 200

T

T

T

F D F

F D F

F D F

δ

δ

δ

= <

= ≤ <

= − ≥

Ｎ

Ｎ

Ｎ

where F＝T1γ1+T2γ2， D： Wheel diameter [mm]

: Interim wear thickness [mm] Tδ

Interim wear based on Pearce, et al.

Wear process on treadWear process on tread

Accumulationinto 5 mm partitions

Hi

Position on wheel [mm]

② Conversion to wear & accumulation

12

・ Wear thickness Ti T icΔ Η= ⋅

Tc ：Coefficient determined from measured data at running distance 60 000 km

・ Smoothing of distribution

(2) Smoothing ofby polynomial

TiΔ ′

Wear process on treadWear process on tread

( )1 1 3Ti Ti Ti TiΔ Δ Δ Δ− +′ = + +(1) Averaging among 3 partitions

Position on wheel [mm]95～10085～9075～8065～7055～6045～50

バック面からの距離 [mm]45 50 55 60 65 70 75 80 85 90 95 100

③ Distribution of wear thickness

13

Niigata University

Wear process on Wear process on flangeflange

Wear index

F F FWI Fμ ψ=

100 5 15Time [s]

0

50

100

150

WI F

100 5 15

0

10

-10

[mra

d]

Left wheelRight wheel

Simulation on cross-over section

14

Wear process on flangeWear process on flange

Position on wheel X

① Wear index in unit area wF

F Fw WI A=

A：Contact area

Accumulation of wF into partitions

FiWConversion to wear thickness

Fi F Fic WΔ = ⋅

:Fc Coefficient determined from measured data at running

distance 60 000 km

Wheel

FWI

Fw

Δx mm

Fw

Contact area

i②

15

③ Distribution of wear ΔFiNiigata University

Wear process on flangeWear process on flange

30

Wear distribution of WFi

30 35 45

25

20

15

Position on wheel [mm]

Hei

ght f

rom

flan

ge to

p [

mm

] Updated worn shape

Before update

Wear distributionaround flange root

Envelope of arcs with the radius of rail edge

representing ΔFi

Wear distribution of WFi

R13

16

Envelope of arcsrepresenting ΔFi

・With alignment irregularities・New rail of Type 60-kg

Main line

Sharp curve (cross-over section)

・ Straight track ･･･85%・ Curved track (R 4 000 m) ･･･15%

Running speed 40 km/h

5 m

R200 m

R200 m

65 m

Track conditions for simulationTrack conditions for simulation

Running speed 240 km/h

17

Niigata University

Comparison of worn profiles between Comparison of worn profiles between prediction and measurement (wheel tread) prediction and measurement (wheel tread)

18

29.5

30.0

30.5

31.0

31.5

32.0

40 50 60 70 80 90 100Position on wheel [mm]

Heig

ht

from

fla

nge

top [

mm

]

1/40 Initial Profile

270 000 km180 000 km120 000 km 60 000 km

MeasuredCalculatedMeasuredPredicted

268 000

Initial profile

Comparison of worn profiles between Comparison of worn profiles between prediction and measurement (wheel flange) prediction and measurement (wheel flange)

25 30 35 40 45Position on wheel [mm]

25

30

20

15

10

5

0

Hei

ght f

rom

flan

ge to

p [m

m]

Initial profile60000km120000km268000km

PredictedMeasured

19

Composition of wear profiles in equal scalesComposition of wear profiles in equal scales20

30

0

29.5

30.0

30.5

31.0

31.5

32.0

40 50 60 70 80 90 100Position on wheel [mm]

Height from flange top [mm]

1/40 Initial Profile

270 000 km180 000 km120 000 km 60 000 km

MeasuredCalculated

20

10

32

25 35 45

55 65 8575 10595

Position on wheel X [mm]Hei

ght f

rom

flan

ge to

p [m

m]

Flangewear

Treadwear

PredictedMeasured

Predicted flange shape in case ofPredicted flange shape in case ofworn outer rail in sharp curveworn outer rail in sharp curve

25 30 35 40 45Position on wheel [mm]

25

30

20

15

10

5

0

Hei

ght f

rom

flan

ge to

p [m

m]

60000 km

PredictedMeasured

New Worn

Outer rail in curved track

80°R10R30

Initial profile

268000 km

120000 km

21

A wear progress of railway wheel was compared with measured data in a field test.

(1) For the tread wear, the general tendency is predictable although the wear is limited within the narrower region. It is thought that the difference comes from the rail profile and switch structure which are not considered in the simulation.

(2) For the flange wear, the predicted profile greatly depends on the inner side shape of the outer rail and tends to follow the worn rail shape in sharp curves.

Concluding remarksConcluding remarks22

Niigata University

(3) This prediction can be useful, however, if we are interestedin the comparison of influences between bogie structures or between railway line conditions such as the rate of contained curved section.

Such examples are as follows:

Concluding remarksConcluding remarks23

Niigata University

As mentioned above, it is perceived that practical rail profile is necessary to predict the real wear profile of wheel.

30

25 30 35 40 45

25

20

15

10

5

0

Position on wheel [mm]

Hei

ght f

rom

flan

ge to

p [m

m] Truck with bolster

and side bearers

Bolster-less truck

270000kmInitial profile

Initial profile(Arc profiled)60000km120000km270000km

Comparison of flange wears Comparison of flange wears between bogie structuresbetween bogie structures

(Conical)

24

29.5

30.0

30.5

31.0

31.5

32.0

40 50 60 70 80 90 100Position on wheel [mm]

Heig

ht

from

fla

nge

top

[mm

]

0%15%30%

Initial Profile

270 000 km

180 000 km

60 000 km1/40

・ Rate of contained curved section (R4000 m )

Comparison of tread wears Comparison of tread wears between the rates of contained curved section between the rates of contained curved section

Initial profile

25

30 %

15 %0 %

Thank you for your kind attention

Niigata University