force-based assessment of weld geometry · 14 ecs presentation on weld geompresentation on weld...

1

ECSECS Presentation on Weld Geometry Standards and RAILPROFPresentation on Weld Geometry Standards and RAILPROF

www.esveld.comwww.esveld.com www.rail.tudelft.nlwww.rail.tudelft.nl

FORCE-BASED ASSESSMENTOF

WELD GEOMETRY

FORCE-BASED ASSESSMENTOF

WELD GEOMETRY

Coenraad EsveldCoenraad EsveldDelft University of Technology

Esveld Consulting ServicesDelft University of Technology

Esveld Consulting Services



TRACK LOADSTRACK LOADS • Wavelength λ• Frequency f• Wavelength λ• Frequency f λ =

vf

λ =vf

λ[m]λ[m]

Wav

eleng

th

Wav

eleng

th

Rollin

g de

fects

Rollin

g de

fects

Balla

st an

d Fo

rmat

ion

Balla

st an

d Fo

rmat

ion

Wel

dsW

elds

Hertzian spring

Hertzian springW

heels

Wheels

BogieBogie

Sprung mass

Sprung mass1000-100 Hz

1000-100 Hz

100-20 Hz

100-20 Hz

20-5 Hz

20-5 Hz

5-0.7 Hz

5-0.7 Hz

0.30.3 33 1010 120

120

Dynamic ForcesDynamic Forces Passenger comfort (Track Recording Cars)Passenger comfort (Track Recording Cars)

3



Tonnage [MGT]

% per 20 MGT

UIC 54 CWR 100% = 1700 km

0 50 100 150 200 250 300 350

10

20

30

NP 46 CWR 100% = 1300 km

TONNAGE BORNE ON NS PER 01-01-1988 TONNAGE BORNE ON NS PER 01-01-1988

25 - 30 years25 - 30 yearsInfraspeed contract ~ 750 - 900 MGTInfraspeed contract ~ 750 - 900 MGT

4



TONNAGE BORNE ON UNION PACIFIC TONNAGE BORNE ON UNION PACIFIC

5



DAMAGE DUE TO POOR WELD GEOMETRYDAMAGE DUE TO POOR WELD GEOMETRY

6



EXISTING WELD GEOMETRY STANDARDSEXISTING WELD GEOMETRY STANDARDS

For exampleVersine: 0 < p < 0.3 mmFor exampleVersine: 0 < p < 0.3 mm

p < 0.3 mmp < 0.3 mm

Grind off topGrind off top

7



Wheel follows rail irregularities;Dynamic part of contact force is governed by:Wheel follows rail irregularities;Dynamic part of contact force is governed by:

z

v

u(t) = z(t) M

K

dynF (t) = Mz(t)&&dynF (t) = Mz(t)&&

22

dyn 2

d zF = Mv

dxα

22

dyn 2

d zF = Mv

dxα

ACCELERATION APPROACHACCELERATION APPROACH

8



with:with:

dyn eF (t) M z(t)= &&dyn eF (t) M z(t)= &&

VELOCITY APPROACH (1)VELOCITY APPROACH (1)Assumption: Equivalent wheel mass is proportional to wavelength:Assumption: Equivalent wheel mass is proportional to wavelength:

e0 0

1 1 vM ML M

L L f= =e

0 0

1 1 vM ML M

L L f= = 2 v

z zLπ

=&& &2 v

z zLπ

=&& &and:and:

9



The dynamic contact force as a function of the first time derivative:The dynamic contact force as a function of the first time derivative:

dyn0

2 vF M z

Lπ

= &dyn0

2 vF M z

Lπ

= &

VELOCITY APPROACH (2)VELOCITY APPROACH (2)

The dynamic contact force in terms of the spatial derivative, including calibration factor β:The dynamic contact force in terms of the spatial derivative, including calibration factor β:

2dyn

0

M dzF v

L dxβ= 2

dyn0

M dzF v

L dxβ=

10



QI ≤ 1: Accepted QI > 1: RejectedQI ≤ 1: Accepted QI > 1: Rejected

max, actual max,actual

norm

norm

dzF dx

QI 1 OKdzFdx

= = ≤ ⇒max, actual max,actual

norm

norm

dzF dx

QI 1 OKdzFdx

= = ≤ ⇒

QUALITY INDICES (QI)QUALITY INDICES (QI)

11



0dz 2 2

z 0.15 0.3 mraddx 3

π πλ

= = ≈0dz 2 2

z 0.15 0.3 mraddx 3

π πλ

= = ≈

RAIL MANUFACTURINGRAIL MANUFACTURING3mλ = 3mλ =

02z 0.3mm=02z 0.3mm=

02 x

z z sinπλ

= 02 x

z z sinπλ

=

12



2tot stat

0

02

M dzQ Q v

L dxdz 1 L 1

Qdx M v

β

β

= + ⇒

< Δ

2tot stat

0

02

M dzQ Q v

L dxdz 1 L 1

Qdx M v

β

β

= + ⇒

< Δ

EXTENSION TO HEAVY HAUL AND HSL (1)EXTENSION TO HEAVY HAUL AND HSL (1)

Total wheel load versus velocity:Total wheel load versus velocity:

Qmax is approximately 450/2 kN = 225 kNM = 2,000 kgQmax is approximately 450/2 kN = 225 kNM = 2,000 kg

13




14



Intervention values for Heavy Haul and HSL lines:Intervention values for Heavy Haul and HSL lines:


0.50.01985280/2High-Speed

1.40.05630100/2Heavy Haul

1.80.07040225/2Conventional

Norm value[mrad]

v [m/s][kN]

0.50.01985280/2High-Speed

1.40.05630100/2Heavy Haul


Norm value[mrad]

v [m/s][kN]

0

dz Mdx L

β⋅

0.50.01985280/2High-Speed

1.40.05630100/2Heavy Haul


Norm value[mrad]

v [m/s][kN]

0.50.01985280/2High-Speed

1.40.05630100/2Heavy Haul


Norm value[mrad]

v [m/s][kN]

0

dz Mdx L

β⋅QΔ

15



FORCE-BASED STANDARDSFORCE-BASED STANDARDS

1.4 mrad50 kN100 km/h


InclinationFDynVelocity




3.2 mrad5 kN40 km/h

QI=1QI=1

Con

vent

iona

lC

onve

ntio

nal

HS

LH

SL

HH

HH

Impl

emen

ted

in R

AIL

PR

OF

Impl

emen

ted

in R

AIL

PR

OF

Tota

l for

ce in

prin

cipl

e 22

5 kN

Tota

l for

ce in

prin

cipl

e 22

5 kN

16



NEW VERSUS OLD NORMNEW VERSUS OLD NORM

1.00.30Old Norm

0.70.21300 km/h

Inclination [mrad]

Versine[mm]

Velocity

0.90.27200 km/h

1.80.54140 km/h

2.40.7280 km/h

3.20.9640 km/h

0dz 2

zdx

20.3 1.0 mrad

2

πλπ

=

= ≈

0dz 2

zdx

20.3 1.0 mrad

2

πλπ

=

= ≈

02 x

z z sinπλ

= 02 x

z z sinπλ

=

2mλ = 2mλ =

02z02z

0z 0.3mm=0z 0.3mm=

For 80 km/h the new norm is 2.4 times more favorable than the old norm, provided short waves have been ground off.For 80 km/h the new norm is 2.4 times more favorable than the old norm, provided short waves have been ground off.

17



LATERAL GEOMETRY STANDARDSLATERAL GEOMETRY STANDARDS

0.5 mm300 km/h

VersineVelocity

0.5 mm200 km/h

0.5 mm140 km/h

0.7 mm80 km/h

1.0 mm40 km/h

QI=1QI=1

Impl

emen

ted

in R

AIL

PR

OF

Impl

emen

ted

in R

AIL

PR

OF

18



ASSESSMENT OLD AND NEW ON PRORAILASSESSMENT OLD AND NEW ON PRORAIL

RP002432RP002432

RP002945RP002945

RP002949RP002949

RP003125RP003125

Old norm: Rejected, New: OK

Old norm: Rejected, New: OK

Old norm: OK, New: Rejected

Old norm: Rejected, New: Rejected

19



SELECTION ON PRORAILSELECTION ON PRORAIL

0

0.2

0.4

0.6

0.8

1

Cum

ulat

ive

Freq

uenc

y

CDFMoerdijk - Dordrecht (VSRT)Delft - Den HaagLage Zwaluwe - Hollands Diep

0 1 2 3 4 5 6 7 8 9Weld Quality Index [-] (140 km/h)

81%

60%

31%

1.8 mrad (140 km/h)1.8 mrad (140 km/h)

Limit at 80 km/hLimit at 80 km/h

100 welds per group100 welds per group

20



OLD VERSUS NEW STANDARDSOLD VERSUS NEW STANDARDS

0

0.2

0.4

0.6

0.8

1

Cum

ulat

ive

Freq

uenc

y

0 1 2 3 4 5 6 7 8 9 10 11 12 13Maximum Absolute Inclination of Weld Geometry (25 mm base) [mrad].

46%

New Standards300 140 80 40 km/h

Old Norm (0 – 0.3 mm)Independent of line speed16 % passed

3%

58%

73% Population 239 welds

21



DYNAMIC FORCEDYNAMIC FORCE

y = 18,62x + 20,93R2 = 0,09

0

50

100

0 0,25 0,5 0,75 1 1,25

versine [mm]

max

. dyn

. con

tact

forc

e [k

N] y = 4,33x

R2 = 0,91

0

50

100

0 5 10 15 20

max. discretised gradient (5 mm basis) [mrad]m

ax. d

yn. c

onta

ct fo

rce

[kN

]

Low correlationforce and versineLow correlation

force and versineHigh correlation

force and QIHigh correlation

force and QI

22



The value of 0.5 mrad as max. inclination for HSL was changed to 0.7 mrad based on 100 measurements of new HSL rails;97 % of rails is better than 0.7 mrad → QI = 1;Standard can only be achieved by QI via Electronic StraightedgeIn new tracks apply grinding train (Plasser GWM).

The value of 0.5 mrad as max. inclination for HSL was changed to 0.7 mrad based on 100 measurements of new HSL rails;97 % of rails is better than 0.7 mrad → QI = 1;Standard can only be achieved by QI via Electronic StraightedgeIn new tracks apply grinding train (Plasser GWM).

HSL STANDARDHSL STANDARD

23



RAIL & WELD GEOMETRY HSLRAIL & WELD GEOMETRY HSL

0.00

20.00

40.00

60.00

80.00

100.00

0.05

0.35

0.65

0.95

1.25

1.55

1.85

2.15

2.45

2.75

3.05

3.35

3.65

3.95

4.25

QI for 300 km /h

Cum

ulat

ive

dist

ribu

tion

%

W eldsRails

QI=

10.

7 m

rad

QI=

10.

7 m

rad

97 %97 %

24



0

0.2

0.4

0.6

0.8

1C

umul

ativ

e Fr

eque

ncy

(%)

CDFbefore grinding trainafter grinding train

0 1 2 3 4 5Weld Quality Index [-]

12%

64%

WELD GRINDING HSL-SOUTH WITH GWMWELD GRINDING HSL-SOUTH WITH GWM

0.7

mra

d0.

7 m

rad

25



WELD STRAIGHTENING VIA STRAITWELD STRAIGHTENING VIA STRAIT

26



F ~ 300 kNF ~ 300 kNSTRAITSTRAIT

27



WELD GRINDING VIA PLASSER GWMWELD GRINDING VIA PLASSER GWM

28



PLASSER GWM EXAMPLESPLASSER GWM EXAMPLES

Esveld, C.: ‘STRAIT: Innovative Straightening of Welds, Rail International, Schienen der Welt, July 1983.

Esveld, C.: ‘STRAIT: Innovative Straightening of Welds, Rail International, Schienen der Welt, July 1983.

29



Procedure:Sample weld geometry with digital straightedgeFilter measured signalDetermine 1st derivative (inclination)Normalize with intervention value for line speed Calculate QI.QI < 1: OK, otherwise: grinding.

Procedure:Sample weld geometry with digital straightedgeFilter measured signalDetermine 1st derivative (inclination)Normalize with intervention value for line speed Calculate QI.QI < 1: OK, otherwise: grinding.

PRACTICAL IMPLEMENTATION (1)PRACTICAL IMPLEMENTATION (1)

30



PRACTICAL IMPLEMENTATIONPRACTICAL IMPLEMENTATION

31



EXAMPLES OF PDA SCREENS (1)EXAMPLES OF PDA SCREENS (1)

32




33




Now seconds addedNow seconds added

34



PDA SCREENPDA SCREEN

V = 140 km/hQI = 1.06V = 140 km/hQI = 1.06

QI uniquely shows where to grindQI uniquely shows where to grind

35



RAILPROF INTERIORRAILPROF INTERIOR

36



Data transfer to PC

Data transfer to PC

37



DES

KTO

P S

OFT

WA

RE

DES

KTO

P S

OFT

WA

RE

All data and graphs can be shown on a PC;Results in pdf-format can directly be emailed to customer.All data and graphs can be shown on a PC;Results in pdf-format can directly be emailed to customer.

38



Exam

ple

of H

SL-S

outh

Exam

ple

of H

SL-S

outh

39



CONCLUSIONS (1)CONCLUSIONS (1)1. Theory based on first derivative works fine in practice;

2. Steel straightedge is absolutely inadequate;

3. Instead electronic straightedges with QI (RAILPROF);

4. High correlation of force and QI, low correlation with versine;

5. With RAILPROF QI measurement: You see what you do;Higher quality;Less rejections provided short waves are ground properly (also negative welds allowed); Extension of life cycle.

1. Theory based on first derivative works fine in practice;

2. Steel straightedge is absolutely inadequate;

3. Instead electronic straightedges with QI (RAILPROF);

4. High correlation of force and QI, low correlation with versine;

5. With RAILPROF QI measurement: You see what you do;Higher quality;Less rejections provided short waves are ground properly (also negative welds allowed); Extension of life cycle.

40



CONCLUSIONS (2)CONCLUSIONS (2)6. New, high quality rails have a first derivative < 0.7 mrad;

7. Mechanical grinding (GWM) is inevitable to achieve such an accuracy for weld geometry;

8. The presented concept is very well applicable to heavy haul tracks and high-speed tracks.

6. New, high quality rails have a first derivative < 0.7 mrad;

7. Mechanical grinding (GWM) is inevitable to achieve such an accuracy for weld geometry;

8. The presented concept is very well applicable to heavy haul tracks and high-speed tracks.

41



CONCLUSIONS (3)CONCLUSIONS (3)9. Validation will be carried out early 2006 by TU Delft:

Dynamic track force measurements at welds for different trains atdifferent speeds;Axle box acceleration measurements;RAILPROF measurements;Statistical analysis to determine relationships.

9. Validation will be carried out early 2006 by TU Delft:Dynamic track force measurements at welds for different trains atdifferent speeds;Axle box acceleration measurements;RAILPROF measurements;Statistical analysis to determine relationships.

RAILPROF Geometry

RAILPROF Geometry

Force Measurements with Gotscha

Force Measurements with Gotscha

Axle Box Accelerations

Axle Box Accelerations

42



SOME EXAMPLES OF HSL SOUTHSOME EXAMPLES OF HSL SOUTH

START OF DESKTOP SOFTWARESTART OF DESKTOP SOFTWARE

RP430320051017133533.xmlRP430320051017133533.xmlSerialSerial year, month, dayyear, month, day hh,mm,sshh,mm,ss

43



www.esveld.comwww.esveld.com

force-based assessment of weld geometry · 14 ecs presentation on weld geompresentation on weld...

Documents