Optimisation of Railway Crossings

Chang Wan

2

Research problem

Research methods

Challenges and solutions

Conclusion

Outline

6% of the train delays due to turnouts in 2010, are

responsible for 55% of the total disruption time.

Impact of broken crossings is 28% of the

mentioned 55% of the disruption time.

Source from: I.Y. Shevtsov. Fracture problems of the crossings: Most common fractures in 2009-2012, presentation of workshopon defects in railway frogs, 2014, Kedichem, the Netherlands.

Some facts in Netherlands

4

Lipping-plastic deformation with cracks on the crossing nose

Head checksDamaged crossing noseSpalling

Damage at crossings

5

How to solve the problem?

6

Two approaches

7

Approach I

Tune the crossing geometry

8

After grindingBefore grinding

Accelerations of nose rail (example of a single train passing at speed of 138.5 km/h)

5 10 15 200

20

40

60

80

100

120

140

Accele

ration (

g)

Wheelset No.

Vertical acceleration

Lateral acceleration

Total acceleration

5 10 15 200

20

40

60

80

100

120

140 Vertical acceleration

Lateral acceleration

Total acceleration

Wheelset No.

Acce

lera

tio

n (

g)

Impact along the crossing (collected data during 2 hour of train passages)

Effect of grinding

9

Changing rail profiles

Crossing

Crossing

Wing rails

Crossing nose

Adjust crossing geometry

10

Approach II

Tune the track elasticity

11

Rail pads

Elastic baseplates

Under Sleeper Pads(USP)

Adjust track elasticity

12

How ???

13

Case 1

Case 2

…

Case n

Predefine n design cases

Best case optimum

Evaluate n cases

Traditional way in railway design optimisation

14

How to choose the n cases ???

15

( ) 1, 1, ,jF j Mx

, 1, ,i i iA B i Nx

Minimise 0( ) min, NF x x R 0( ) min, NF x x R

Subject to

Initial design

Update design using optimisation method

Analyse design

Does the design satisfy convergence criteria? Done

Y

N

Numerical optimisation problem

16

Evaluation of crossing design based on

Dynamic vehicle-turnout interaction

Crossing geometry optimisation

18

Wheel-rail normal contact pressure S

Energy dissipation in the contact patch W Wear

RCF damage

0 1 2

( ) ( )min

( ) ( )

S WF w w

S W

X X

XX X

Optimise crossing geometry to reduce damages at wheel-rail interface!

Optimisation problem

19

No risk of derailment

Sufficient support surface when impact occurs

Constraints of nose rail geometry

• Smooth-convex cross-sectional shape

• Monotonically increased height profile

Constraints

Track elasticity optimisation

21

High frequency force on rail (P1) Accounts for damage at the wheel/rail interface

Low-frequency forces on ballast (Fbl) and sleepers (Fsl) Accounts for damage of the sleeper and ballast bed.

Reduction of dynamic forces on turnout crossing

1

10

1

minbl sl

bl slP F F

bl sl

F FPF W W W

P F F X

Optimisation problem

22

Maximum displacement of rail

Maximum displacement of sleepers

Maximum deflection of rail pads

Constraints

23

Challenges

24

1.

Parameterisation of crossing design

25

Parameterisation of crossing geometry

26

701:

+2 -1

55 25x 4

3¡À1

¡ À5 43

¡À1 +

2 -16

0

+2 -1

90

14

mm

10*

I II III IV

25

¡À1

33

¡À1

min

. 4

5

min

R10

25

¡À1

33

¡À1

min

. 4

5

+1 -0.5

10 +

1 -0.5

5

+1 -0.5

3.5

10x ¡ À5 10x ¡ À5 50x ¡ À5

I II III IV

min R10

0

4:1 4:1

4:1 4:1

4:1 4:1

20* 70*

+1 -0.5

10 +

1 -0.5

5

+1 -0.5

3.54¡ À1

TP

(a)

(b)

(c)

10xα±5 10xα±5 50xα±5

25xα±5

4±1

33±1

25±1

33±1 25

±1

43±1

43±1

Step I: Choose control cross-sections

27

Tuning cross-sections III, V

Fixed cross-sections I, II, IV

70

A B C

D

7014 m

m

2xa1

b1

2xa 2

b 2

A B C D

25 mm

55

4

1:

10x 10x 50x

20xαE

I II III VIV

Step I: Choose control cross-sections

28

Longitudinal height profile of nose rail

Step II: Determine design parameters

0 200 400 600 800 1000 1200-15

-10

-5

0

5

10

15

fixed point

movable point

reference profile

adjusted profile

II

Hei

ght

[mm

]

I

h1

h2

Distance from TP [mm]

III

V

IV

Hei

ght w

.r.t t

he t

op o

f no

rmal

rai

l [m

m]

0

-14

TP

h1h2

e e 2e

Wing rail Nose rail

I II III V IV

3e

29

Step II: Determine design parameters

Distance from nose rail centre [mm]

Hei

ght

[mm

]

0 5 10 15 20 25-4

-2

0

2

4

6

8

10

12

1:4

1:25

fixed control points

movablecontrol points

P0

P1 P2

P3

P4

P5

P

M

Transvers profile of nose rail (B-spline)

30

Parameterisation of track elasticity

31

3 6

1 2 3 4 5 6 7 8 9 10 11

Design parameters: elastic properties of rail pads & USP

Parameterisation of track elasticity

32

Design parameters: elastic properties of rail pads & USP

3 6

1 2 3 4 5 6 7 8 9 10 11

M RF

transition part

Rail pad

USP

F M R

Parameterisation of track elasticity

33

2.

Uncertainties in crossing design

34

Robust Optimisation

f(X)

X

optimal: non-robust

suboptimal: robust

35

3.

Balance between model accuracy and

computational cost

36

Experimental investigation

MBS simulation

Rail surface

FEM simulation

37

Optimum solution

38

Nose rail shape

Longitudinal height profile of nose rail

0 200 400 600 800 1000 1200-15

-10

-5

0

5

10

15

Distance from TP [mm]

Heig

ht

[mm

]

ref

opt

II

III

V

IV

0 5 10 15 20 25 300

5

10

15

Distance from nose rail centre [mm]

Heig

ht

[mm

]

ref

opt

Optimum crossing geometry

39

Shift of impact location

Reduction of contact pressure

2000

2500

3000

3500

4000

4500

5000

Conta

ct

pre

ssure

[M

Pa]

opt

ref

C2C1B2B1A2A1

200

300

400

500

600D

ista

nce f

rom

TP

[m

m]

opt

ref

A2 B1 B2 C1 C2A1

A1-C2 represent different vehicle-track conditions

Robust dynamic behaviour

40

Softer rail pads combined with the USPs

Rail pads with different properties

Relatively less soft rail pads before and after the crossingnose than under the crossing nose

Optimum track elasticity

Reduction of dynamic forces--Compared to original design

Optimum designs

42

How to implement ?

43

Crossing before grinding Crossing after grinding

Grinding/welding maintenance

44

source from Network Rai Engineering education video: https://www.youtube.com/watch?v=qsCoJJLhS68

Manufacturing process

45

A methodology of improving the dynamic crossing performance by tuningcrossing designs is proposed, including:

• Adjusting crossing geometry

• Adjusting track elasticity

The methodology combines modern optimisation techniques and dynamic train-turnout interaction, and accounts for

• Multiple assessment criteria in design optimisation

• Uncertainties in design

• Realistic parameterisation of crossing designs

• Balance between accuracy and computational cost

Concluding remarks

46

Details of the research can be found in

C. Wan. Optimisation of vehicle-track interaction at railway crossings. PhD thesis; TU Delft.

https://www.researchgate.net/publication/307977713_Optimisation_of_vehicle-track_interaction_at_railway_crossings

http://repository.tudelft.nl/islandora/object/uuid:8dac8f02-fe9e-4baa-9503-e9b3d79dd1aa?collection=research