05 Britto's Line 2 3 Update to MSP 2013-05-06

Transportation Technology Center, Inc., a subsidiary of the Association of American Railroads

© TTCI/AAR, 1/10/2012. filename, p1

Update on TTCI/MSP Vehicle Dynamics

ProjectsMarch 26, 2015

Ruben PeñaStan GuruleRuss Walker

© TTCI/AAR, 1/11/2012, Filename p2

® Introduction

♦ Line 3 both Cobrasma (MTE) and (Mafersa/Tokyu)●Objective: To learn how the age of the primary suspension affects performance. This will allow MSP to decide on maintenance intervals

♦ Line 2 – Millenium Fleet Cars (Mafersa/Tokyu)●Characterize and measure the performance of the Line 2 Millenium fleet and make recommendations for any necessary changes to optimize suspension performance.

♦ Line 5 (CAF/Alstom)●Characterize and measure the performance of the Line 5 fleet and make recommendations for any necessary changes to optimize suspension performance.


® Primary Suspension Line 3 MTE/Cobrasma


®

The Cobrasma/MTE French Design Primary Suspension


® The Cobrasma/MTE French Design Complete Truck


® Primary SuspensionLine 2 and 3 Tokyu/Mafersa


®

Exploded View of Primary Chevrons of Mafersa/TokyuTruckLines 2 and 3


® Alstom/CAF Line 5 Car


®

Bogie Arrangement ViewLine 5 Car


® Bogie Arrangement Line 5 Car


® Method

♦ Characterize the track♦ Characterize the cars♦ Truck load equalization (dQ/Q) testing of cars♦ Test the cars on Jabaquara curve and on perturbations

at Belem yard, or at Capao Redondo for Line 5♦ Use computer modeling of the car to estimate wheel

loads and L/V Ratio in a variety of conditions


® What We Need to Prevent!


® Performance Criteria

♦ The performance measures selected for this computer modeling study are based on the United States (US) Code of Federal Regulations (CFR) Title 49 Part 213.333 Vehicle Track Interaction safety limits.

♦ The wheel L/V safety limit is a function of flange angle. The G-29 flange angle is 64.753 degrees corresponding to a wheel L/V safety limit of 0.79.

Vehicle/Track Interaction Safety Limits Parameter Safety Limit Filter/Window Requirements

Wheel/Rail Forces1

Single Wheel Vertical Load

Ratio

≥0.1 5 ft

No wheel of the equipment shall be permitted to unload to less than 10% of the static

vertical wheel load. The static vertical wheel load is defines as the load that the wheel

would carry when stationary on level track. The vertical wheel load limit shall be

increased by the amount of measurement error.

Single Wheel L/V Ratio

.51 .5

5 ft

The ratio of the lateral force that any wheel exerts on an individual rail to the vertical force exerted by the same wheel on the rail shall be

less than the safety limit calculated for the wheel’s flange angle ( ).

Net Axle L/V Ratio ≤ 0.5 5 ft

The net lateral force exerted by any axle on the track shall not exceed 50% of the static

vertical load that the axle exerts on the track.

Truck Side L/V Ratio ≤ 0.6 5 ft

The ratio of the lateral forces that the wheels on one side of any truck exert on an individual rail to the vertical forces exerted by the same

wheels on that rail shall be less than 0.6 Acceleration

Carbody Lateral2

≤ 0.5 g peak-to-peak

10 Hz 1 sec window

The peak-to-peak acceleration, measured as the algebraic difference betweeen the two

extreme values of measured acceleration in a one second time period, shall not exceed 0.5

g.

Carbody Vertical

≤ 0.6 g peak-to-peak

10 Hz 1 sec window

The peak-to-peak acceleration, measured as the algebraic difference betweeen the two

extreme values of measured acceleration in a one second time period, shall not exceed 0.6

g.

Truck Lateral3

≤ 0.4 g RMS mean-

removed

10 Hz 2 sec window

Truck hunting4 shall not develop below the maximum authorized speed.

1. The lateral and vertical wheel forces shall be measured with instrumented wheelsets with the measurements processed through a low pass filter with a minimum cut-off frequency of 25 Hz. The sample rate for wheel force data shall be at least 250 samples/sec.

2. Carbody lateral and vertical accelerations shall be measured near the car ends at the floor level.

3. Truck accelerations in the lateral direction shall be measured on the truck frame. The measurements shall be processed through a filter having a pass band of 0.5 to 10 Hz.

4. Truck hunting is defined as a sustained cyclic oscillation of the truck which is evidenced be lateral accelerations in excess of 0.4 g root mean square (mean removed) for 2 seconds.


® Vehicle Measurements


® Wayside Instrumentation

♦ Bill, Mazza, and Mike placed strain gages on the rails to measure vertical and lateral wheel forces.

♦ Capao Redondo♦ Belem♦ Jabaquara

Curve♦ 2 Sites on Line

2♦ 1+ Site on

Line 5


® Wayside Instrumentation

♦ Wayside measurements provide:●Vertical Wheel Forces●L/V Ratio

♦ These can be compared directly to the metrics

♦ Only available at specific points on the track

Example of Wayside Data


® Vehicle Dynamics Model

♦ Modeling vehicle using NUCARS®

●Multibody vehicle dynamics model.♦ Inputs

●Masses and mass moments of inertia●Stiffness●Damping●Friction

♦ Outputs●Wheel/rail forces●Body motions●….


® Includes Track Model

♦ Nucars can model the rail and ties (or slab) as masses with many degrees of freedom

♦ Requires that we characterize the track in addition to the vehicle


® Track Characterization

♦ Track characterization provides inputs for the computer model

♦ Done with 2 types of tests●Track Hammer Test

▲Allows us to determine the stiffness of rail connections to the slab

●Instrumentation on slab to measure displacements under traffic▲Allows us to determine the stiffness of the slab to ground


® Track Hammer Test

♦ Accelerometers are placed on the rail and on the tie or slab

♦ The rail is struck with an instrumented hammer♦ The vibration of the track helps identify the stiffness of

the fasteners


®

62m

31m

124m slab in curve suspended on coil springs

Adjacent slab, not suspended on coil springs

DY03DZ08DZ07

DZ06DZ05

DZ02DY01DY02

DZ04DZ03

DZ01

Slab Track Instrumentation

Vertical Displacement Measurement

Lateral Displacement Measurement

♦ Sections of track were instrumented with rail force circuits and slab displacement measurements


® Example of Slab Track Data


® Vehicle Characterization Test

♦ These tests are only to identify model inputs●Suspension Stiffness Test●Car body modal test – Used to estimate the car center of

gravity height and car body mass moments of inertia●Bogie modal test – Used to identify the bogie mass moments of

inertia


® Stiffness Testing

Longitudinal Stiffness Test Lateral Stiffness Test

Vertical Stiffness Test

y = 5.7293x + 0.1249R² = 0.9817

‐0.5

0

0.5

1

1.5

2

2.5

3

3.5

‐0.2 0 0.2 0.4 0.6

Force (kN)

Displacement (mm)

Primary Lateral (Loaded)

Linear (Primary Lateral (Loaded))


® Carbody Modal Testing

♦ Car is instrumented with enough accelerometers to identify the five modes of vibration

♦ Car is excited by hand or by quickly releasing a jack to let the car settle


® Bogie Modal Test


® Truck Load Equalization

♦ The truck must distribute the load to the track equally on all the wheels

♦ This is key to preventing flange climb derailment

Flexible body Stiff Body


® Truck Load Equalization

♦ Raise 1 wheel of a truck while measuring load on all 4 wheels

♦ APTA SS-M-014-06●Class G

▲Used where track standards allow a difference in crosslevel of no more than 76.2mm in 18.9m

▲35% wheel load at 63.5mm and >0% wheel load at 76.2mm

●Class R – Using this level – MSP can advise if it is appropriate▲Used where track standards allow difference in

crosslevel of no more than 76.2mm in 18.9m and no more than 57.2mm in 3.0m

▲35% wheel load at 50.8mm and >0% wheel load at 63.5mm

This standard needs to be evaluated considering MSP track maintenance practices.


® Curve Testing

♦ Curve tests were performed either at Jabaquara on a 154m curve or at Capao Redondo on a 300m curve

♦ Wayside gages recorded L/V ratios and onboard measurements recorded suspension displacements

00.10.20.30.40.50.60.70.80.91

0 5 10 15 20 25 30 35

L/V Ratio

Speed (kph)

Empty‐South

Empty‐North

00.10.20.30.40.50.60.70.80.91

0 5 10 15 20 25 30 35

L/V Ratio

Speed (kph)

Loaded‐South

Loaded‐North


® Perturbed Track Tests

♦ Track geometry (deviations in the track) may excite the car in some mode of rigid body vibration


® Perturbed Track Tests

♦ This sketch shows how the perturbed track zones are set up at Belem Yard

♦ Similar zones are installed at Capao Redondo for line 5 tests

Car 3144 Car 3139

N

Yaw Bounce Roll

Roll tests were performed westbound with car 3144 leading

Yaw tests and bounce tests were performed eastbound

with car 3139 leading


® Outline of Line 3 Results

♦ Review results for line 3 cars Cobrasma (MTE) and Maferasa (Tokyu)

♦ Model Validation♦ Test result from recently re-profiled wheel and the same

wheel after re-profiling marks had worn away.♦ Simulations

●Turnout simulations▲Worn wheels and rails▲Gage clearance variations▲Turnout defects

●Primary suspension stiffness variation


® Model ValidationLine 3

♦ Truck load equalization predictions match test results showing that the model correctly represents the primary suspension

0%

50%

100%

150%

200%

0 20 40 60 80

Percent V

ertical Load

Wheel 1L Height (mm)

Test 1LTest 1RTest 2LTest 2RModel 1LModel 1RModel 2LModel 2R

G‐0%R‐35% G‐35%R‐0%

35% Wheel Load

0%

50%

100%

150%

200%

0 20 40 60 80

Percent V

ertical Load


Test 1LTest 1RTest 2LTest 2RModel 1LModel 1RModel 2LModel 2R

G‐0%R‐35% G‐35%R‐0%

35% Wheel Load

Cobrasma Mafersa


® Model Validation – Line 3♦ Models reproduce dynamic test results (Example: Em T&R)

Cobrasma Mafersa

0

10

20

30

40

50

10 30 50 70

Displacement (mm)

Speed (kph)

Test LL Vert

Test LR Vert

Test TL Vert

Test TR Vert

Model LL Vert

Model LR Vert

Model TL Vert

Model TR Vert0

10

20

30

40

50

60

10 30 50 70

Dsiplacement (mm)

Speed (km/h)

Test LL Vert

Test LR Vert

Test TL Vert

Test TR Vert

Model LL Vert

Model LR Vert

Model TL Vert

Model TL Vert


® Key Test Result – Cobrasma Car

♦ The wheels on the lead truck of the Cobrasma car were re-profiled just before the first empty curving test. The measured L/V ratios were very high.

♦ The test was repeated after some wear in. The L/V ratio had dropped significantly.

0

0.2

0.4

0.6

0.8

1

1.2

0 10 20 30 40

L/V Ra

tio

Speed (kph)

Immediately After Reprofiling ‐ Outer RailImmediately After Reprofiling ‐ Inner RailRepeat After Wear In ‐ Outer RailRepeat After Wear In ‐ Inner Rail


® Line 3 Turnout Simulations – Line 3

♦ Covers conditions in RFP♦ New and worn wheel profiles♦ New and curve worn rail

profiles♦ Diverging route simulation of

1:8, 1:9, 1:11.4, and 1:14 turnouts

♦ Switch embedding for 1:8, 1:9, 1:11.4, and 1:14 turnouts

♦ Speeds of 20, 40, and maximum civil speed for the turnout. Maximum civil speeds are 44kph for 1:8 and 1:9, 63kph for 1:11.4, and 73kph for 1:14.

♦ Track and wheelset gauge variations

Input Wheel Profile

Rail Profile Speed System File Num. of Sim.

1:8 New and Worn

New or Measured and Worn

20, 40, 44kph

Empy Cab Leading and Trailing

24

1:9 New and Worn


20, 40, 44kph


24

1:11.4 New and Worn


20, 40, 63kph


24

1:14 New and Worn


20, 40, 73kph


24

1:8 New New or Measured 20, 40, 44kph

Loaded Cab Leading and Trailing

6



6

1:11.4 New New or Measured 20, 40, 63kph


6



6

1:8 narrow gage

New Wide B-B

New or Measured 20, 40, 44kph


6

1:9 narrow gage

New Wide B-B



6

1:11.4 narrow gage

New Wide B-B



6

1:14 narrow gage

New Wide B-B



6

1:8 wide gage

Worn Narrow B-B

Worn 20, 40, 44kph


6

1:9 wide gage

Worn Narrow B-B

Worn 20, 40, 44kph


6

1:11.4 wide gage

Worn Narrow B-B

Worn 20, 40, 63kph


6

1:14 wide gage

Worn Narrow B-B

Worn 20, 40, 73kph


6

1:8 with embedding

New New or Measured 20, 40, 44kph


6

1:9 with embedding



6

1:11.4 with embedding



6

1:14 with embedding



6

Total Number of simulations for Cobrasma (and the same number for Mafersa) 192


® Line 3 Turnout Simulations

♦ Embedding – Switchpoint defect cased

Ramp Length = 1mDefect Length Varies 0 to 2.1m

Mainline Stock Rail

Defect

Switch Rail


® Line 3 Primary Stiffness Variations

Body Longitudinal Primary Stiffness

(kN/mm)

Lateral Primary Stiffness (kN/mm)

Vertical Primary Stiffness (kN/mm)

Nominal Stiffness (Long and Lat Estimated)

7.0 4.0 1.735

Measured Stiffness 15.0 9.0 3.6 2 x Increase 23.0 14.0 5.5 3 x Increase 31.0 19.0 7.4

Body Longitudinal Primary Stiffness

(kN/mm)

Lateral Primary Stiffness (kN/mm)

Vertical Primary Stiffness (kN/mm)

Nominal Stiffness (except longitudinal)

17.0 9.5 5.9

Measured Stiffness 30.4 28.9 9.9 2 x Increase 57.1 48.3 13.9 3 x Increase (longitudinal is nominal)

83.8 67.7 17.9

Primary suspension stiffness values being simulated for the Cobrasma/MTE car.

♦ Simulating vehicle performance with a wide range of primary suspension stiffness.

♦ Measured stiffness of worn suspension is normally higher that the nominal stiffness given on drawings or procurement standards

♦ Higher levels of suspension stiffness will be simulated as well

Primary suspension stiffness values being simulated for the Mafersa/Tokyu car.



♦ Simulation cases:●Truck load equalization

(Jacking Test) with airbags inflated

●Truck load equalization with airbags deflated

●Nominal 1:9 turnout●1:9 turnout with truck

twist track perturbation●1:9 turnout with truck

twist perturbation with double amplitude



♦ Specific cases of asymmetric suspension conditions●One case investigates performance of a suspension if 1

element is stiffer than the other 3.▲Three of the primary springs of the lead truck will use the

measured stiffness, and the lead inside wheel will use 2x the measured stiffness.

●Two cases investigate performance of an improperly shimmed truck.▲All 4 primary springs of the lead truck will use the Measured

Stiffness values.▲Lead outside spring will be offset 5mm or 10mm below the

other 3 springs of that truck (to take less weight).●These conditions will be simulated for the 1:9 turnout with the

track twist perturbation.


® Wheel/Rail Profile Simulation PredictionsLine 3

♦ Simulations using new and worn wheel profiles show that worn profiles are worse, but both cars meet the FRA 213 safety criteria for these combinations of wheel/rail profiles on nominal inputs

Cobrasma Mafersa

0

0.2

0.4

0.6

0.8

1

1.2

15 35 55 75

Whe

el L/V

Speed (kph)

New Wheel, New RailNew Wheel, Worn RailWorn Wheel, New RailWorn Wheel, Worn RailLimit

1:9 Turnout

0

0.2

0.4

0.6

0.8

1

1.2

15 35 55 75

Whe

el L/V

Speed (kph)

New Wheel, New RailNew Wheel, Worn RailWorn Wheel, New RailWorn Wheel, Worn RailLimit

1:9 Turnout


® Gage Clearance Simulation PredictionsLine 3

♦ Simulations using narrow and wide gage clearance show that nominal gage clearance perform better than either narrow or wide gage clearance. ●Cobrasma is close to but meets FRA 213 wheel L/V safety

criteria●Mafersa is close to but does not meet FRA 213 wheel L/V

safety criteriaCobrasma Mafersa

0

0.2

0.4

0.6

0.8

1

1.2

15 35 55 75

Whe

el L/V

Speed (kph)

New Wh and Ra St Ga and B‐BNew Wh and Ra Nar Ga Wide B‐BWorn Wh and Ra St Ga and B‐BWorn Wh and Ra Wide Ga Nar B‐BLimit

1:9 Turnout

0

0.2

0.4

0.6

0.8

1

1.2

15 35 55 75

Whe

el L/V

Speed (kph)


1:9 Turnout


® Turnout Defect Simulation PredictionsLine 3

♦ The FRA 213 wheel L/V criteria with a 5-foot window does not properly indicate the tendency for flange climb derailment for turnout defect simulations.●The high angle of attack and the large lateral displacements

cause derailments in an extremely short distance.●Because of this an output that indicates the position of flange

contact was substituted.



♦ Simulations show that long defect lengths create potential for flange climb derailment on 1:8 and 1:9 turnouts.●Cobrasma: Defect lengths should be limited to 1m or less for 1:8 and

1:9 turnouts●Mafersa: Defect lengths should be limited to 1.4m and 0.7m for 1:8

and 1:9 turnouts respectively.●Because both cars use the same lines the most restrictive case should

be used.

Cobrasma Mafersa

0

0.005

0.01

0.015

0.02

0.025

0.03

15 25 35 45

Rolling

Rad

ius

(to indicate flan

ge con

tact position

)

Speed (kph)

New

0.7m Defect

1.0m Defect

1.4m Defect

1.6m Defect

1.8m Defect

2.0m defect

2.1m Defect

Limit

1:9

0

0.005

0.01

0.015

0.02

0.025

0.03

15 25 35 45

Rolling

Rad

ius

(to indicate flan

ge con

tact position

)

Speed (kph)

New

0.7m Defect

1.0m Defect

1.4m Defect

1.6m Defect

1.8m Defect

2.0m defect

2.1m Defect

Limit

1:9


® Primary Stiffness Simulation PredictionsLine 3

♦ Primary stiffness has a significant effect on truck load equalization simulation predictions.

♦ The truck load equalization test uses very large displacements (up to 75mm) compared to MSP’s track twist limit (12mm).

Cobrasma Mafersa

0%

50%

100%

0.0 20.0 40.0 60.0 80.0

Percent V

ertical Load


NominalMeasured2 x Increase3 x Increase35% Load

G‐0%R‐35% G‐35%R‐0%

35% Wheel Load

Empty Airbag Deflated

0%

50%

100%

0.0 20.0 40.0 60.0 80.0

Percent V

ertical Load



G‐0%R‐35% G‐35%R‐0%

35% Wheel Load




♦ While the Cobrasma car is below the limit for all primary suspension stiffness values checked, the Mafersa car is very close to the limit in all cases.

♦ The Nominal stiffness for the Mafersa car is greater than the 2 x Increase case for the Cobrasma car.

Cobrasma Mafersa

0.0

0.2

0.4

0.6

0.8

1.0

1.2

10 20 30 40

Whe

el L/V

Speed (kph)

NominalMeasured2 x Increase3 x IncreaseLimit

Perturbation

0.0

0.2

0.4

0.6

0.8

1.0

1.2

10 20 30 40

Whe

el L/V

Speed (kph)


Perturbation


® Special Cases Simulation PredictionsLine 3

♦ The softer primary stiffness of the Cobrasma car makes it more tolerant of conditions that cause asymetricalloading of the truck, such as improper shimming or variations in primary suspension stiffness.

Cobrasma Mafersa

0

0.2

0.4

0.6

0.8

1

0 10 20 30 40 50

Whe

el L/V

Speed (kph)

Measured StiffnessHigher stiffness on one axleboxImproper shim on one axlebox (5mm)Improper shim on one axlebox (10mm)

0

0.2

0.4

0.6

0.8

1

0 10 20 30 40 50

Whe

el L/V

Speed (kph)

Measured StiffnessHigher Stiffness on one axle boxImproper Shim on one axlebox (5mm)Improper Shim on one axlebox (10mm)


® Summary Final Report Recommendations for Line 3 Cobrasma

♦ The line 3 Cobrasma car performed well with the highest stiffness simulated, 7.4 kN/mm, which is twice the thestiffness measured on the worn suspension. MSP should evaluate the risks of extending the current interval of suspension replacement.


® Summary Final Report Recommendations for Line 3 Mafersa

♦ The maximum primary stiffness of the Mafersa car should be maintained at 10kN/mm at each axle box. This is consistent with the stiffness measured at MSP’s current maintenance interval. ●MSP should continue to a maintain track twist of 12mm in 4 m

on all of their tracks.●This high level of primary stiffness makes the vehicle more

susceptible to deterioration in performance caused by asymmetric primary suspension setup. Use of a truck twist index computed from wayside force measurements can identify trucks with abnormal primary suspension setups. Such a system could be used on Line 3 to identify cars with this problem.


® Additional Recommendations

♦ MSP should refine surface finish requirements for wheel reprofiling to minimize the effects seen during the line 3 Cobrasma empty curving test.

♦ MSP should limit the size of turnout defects to 1.0m and 0.7m on 1:8 and 1:9 turnouts respectively.

♦ MSP should consider pursuing a Wheel/Rail profile study. A wheel profile with a steeper flange angle over a longer distance on the flange would increase the L/V ratio and distance to climb required for flange climb derailment. Using a profile design based on the worn shape may also reduce wheel wear, although care must be taken to avoid rolling contact fatigue of wheels and rails.


® Outline of Line 2 Results

♦ Review results for line 2 Millennium Cars♦ Model Validation♦ Simulations

●Turnout simulations▲Worn wheels and rails▲Gage clearance variations▲Turnout defects

●Primary suspension stiffness variation●Possible suspension improvements – Reducing primary

suspension vertical stiffness●Analysis of wayside data


® Line 2 Model Validation

♦ Truck load equalization simulation predictions match test results very closely.

0%

50%

100%

150%

200%

0 20 40 60 80

Percent V

ertical Load


Axle 1LAxle 1RAxle 2LAxle 2RModel 1LModel 1RModel 2LModel 2R35% LoadLimit

G‐0%R‐35% G‐35%R‐0%

35% Wheel Load

0%

50%

100%

150%

200%

0 20 40 60 80

Percent V

ertical Load


Axle 1LAxle 1RAxle 2LAxle 2RModel 1LModel 1RModel 2LModel 2R35% LoadLimit

G‐0%R‐35% G‐35%R‐0%

35% Wheel Load

Empty Loaded


® Line 2 Model Validation

♦ Pitch and Bounce Model vs Test

0

5

10

15

20

25

10 30 50 70

Displacem

ent (mm)

Speed (kph)

Model LL Vert

Model LR Vert

Model TL Vert

Model TR Vert

Test LL Vert

Test LR Vert

Test TL Vert

Test TR Vert0

5

10

15

20

25

10 30 50 70

Displacem

ent (mm)

Speed (kph)

Model LL Vert

Model LR Vert

Model TL Vert

Model TR Vert

Test LL Vert

Test LR Vert

Test TL Vert

Test TR Vert

Empty Loaded


® Wheel/Rail Profile Simulation PredictionsLine 2

♦ Simulations using new and worn wheel and rail profiles show that worn profiles are worse, but all combinations of wheel/rail profiles meet the FRA 213 safety criteria for nominal inputs.

0

0.2

0.4

0.6

0.8

1

1.2

15 35 55 75

Whe

el L/V

Speed (km/h)

New Rail, New WheelNew Rail, Worn WheelWorn Rail, New WheelWorn Rail, Worn WheelLimit

1:9 Turnout


® Gage Clearance Simulation PredictionsLine 2

♦ Simulations using narrow and wide gage clearance show that nominal gage clearance performs better than either narrow or wide gage clearance. All combinations of gage clearance meet the FRA 213 safety criteria for nominal inputs.

0

0.2

0.4

0.6

0.8

1

1.2

15 35 55 75

Whe

el L/V

Speed (km/h)


1:9 Turnout



♦ Simulations of turnout defects show 1:8 and 1:9 turnouts with defects longer than 1.6m and 1.0m respectively have a tendency for flange climb derailment. For 1:11.4 and 1:14 turnouts defect lengths up to 2.1m (the longest simulated) do not show a tendency for flange climb derailment.

0

0.005

0.01

0.015

0.02

0.025

0.03

15 25 35 45

Rolling

Rad

ius

(to indicate flan

ge con

tact position

)

Speed (km/h)

New

0.7m Defect

1.0m Defect

1.4m Defect

1.6m Defect

1.8m Defect

2.0m defect

2.1m Defect

Limit

1:9



♦ Simulation predictions show that while primary stiffness has a significant effect on truck load equalization results, the effects are smaller for simulations using track twist perturbations based on the normal MSP track twist maintenance limits.

0.0

0.2

0.4

0.6

0.8

1.0

1.2

10 20 30 40

Whe

el L/V

Speed (km/h)


Perturbation

0%

50%

100%

0.0 20.0 40.0 60.0 80.0

Percent V

ertical Load



G‐0%R‐35% G‐35%R‐0%

35% Wheel Load



® Special Case Simulation PredictionsLine 2

♦ Line simulations to investigate asymmetrical primary suspension configuration were based on the condition of a car involved in a yard derailment in May 2011.

♦ One axle of the derailed truck had new primary suspension elements while the other axle had worn ones.

♦ The wheels of the derailed truck had recently been re-profiled

♦ Static loads on the derailed truck showed asymmetrical wheel loads

Location Side 1 Side 2First Axle 33.5 kN 60.5 kNSecond Axle 60.1 kN 30.7kN


® Special Cases Simulation PredictionsLine 2

♦ Simulation predictions show that asymmetry in the primary suspension can cause performance to degrade. Simulations designed to replicate the static wheel loads measured on a Line 2 car that derailed in May 2011 showed that the asymmetrical loading caused lower vertical loads and higher L/V ratios

0.0

0.2

0.4

0.6

0.8

1.0

10 20 30 40

Maxim

um W

heel L/

V

Speed (km/h)

Nominal Condition

Derailed Condition

Limit

0%

20%

40%

60%

80%

100%

120%

10 20 30 40

Vertical Loa

d

Speed (km/h)

Nominal ConditionDerailed ConditionLimit


® Suspension Improvements – Reducing Primary Stiffness, Line 2

♦ Reducing primary stiffness 30% below nominal improves truck load equalization performance but has little effect on simulations that use track twist perturbations similar to the 12mm maintenance limit followed by MSP.

0%

50%

100%

150%

200%

0.0 50.0

Percent V

ertical Load


Original Model 1L

Original Model 1R

Original Model 2L

Original Model 2R

Revised Model 1L

Revised Model 1R

Revised Model 2L

Revised Model 2R

35% Load

Limit

G‐0%R‐35% G‐35%R‐0%

35% Wheel Load

0

0.2

0.4

0.6

0.8

1

0 10 20 30 40 50

Whe

el L/V

Speed (km/h)

Reduced Primary Stiffness

Nominal

12mm Perturbation CL


® Data from Line 2 Track Characterization

♦ Use wheel-rail force data for a vehicle performance survey.●Wheel L/V ratio indicates curving performance

▲Criteria = 0.79●Defined a car body twist index to indicate secondary

suspension performance. ▲(L1-R1+L2-R2-L3+R3-L4+R4)/Car Weight▲No established criteria. Use Quartile analysis to find outliers.

●Defined a truck twist index to indicate primary suspension performance.▲(L1-R1-L2+R2)/Truck weight.▲No established criteria. Use Quartile analysis to find outliers.


® Analysis of Wayside Data – Line 2

♦ Wayside force measurements can be used to identify trucks with poor primary suspension condition. If not already doing so, MSP should consider using a truck twist index to identify poorly performing primary suspensions.

♦ For comparison, the truck twist index for car that derailed in May 2011 was 0.31

0

0.2

0.4

0.6

0.8

1

12:00 14:00 16:00 18:00

Max L/V

Ratio by Truck

Time ‐May 24, 2011

L/V RatioOuter Fence 0.49Wheel L/V Safety Limit

0

0.02

0.04

0.06

0.08

0.1

0.12

12:00 14:00 16:00 18:00

Car T

wist Ind

ex

Time ‐May 24, 2011

Car Twist IndexOuter Fence 0.104

0

0.05

0.1

0.15

0.2

12:00 14:00 16:00 18:00

Truck Tw

ist Ind

ex

Time ‐May 24, 2011

Truck Twist IndexOuter Fence 0.13


® Track Modeling and Characterization

♦ What is the NUCARS Track model and how does it work●Multi-body representation of track structure

▲Flexible rails and ties (euler-bernoulle beams)─Up to 89 modes each: vertical, lateral, twist

▲Standard NUCARS connections for track fasteners and ballast

▲Multiple layers possible●W/R Penetration model

▲Rail profiles can vary along track



♦ Single Layer Model



♦ Two Layer Model



♦ Flexible rail modes



♦ Line 2 Coil Spring Slab♦ Simulating 123m slab with 20 modes



♦ Hammer testing results. Frequency match is good.

♦ Indicates correct stiffness of rail-slab fastener.

♦ About 80kN/mm

‐15

‐10

‐5

0

5

10

15

20

25

0.49 0.51 0.53 0.55 0.57 0.59

Acceleratio

n (g)

Time (s)

Line 2 Coil Springs ‐ 1/2 Way Between Fasteners

ModelTest

‐15

‐10

‐5

0

5

10

15

20

25

0.49 0.51 0.53 0.55 0.57 0.59

Acceleratio

n (g)

Acceleration (g)

Line 2 Coil Springs ‐ 1/4 Way Between Fasteners

ModelTest

‐20

‐15

‐10

‐5

0

5

10

15

20

25

0.49 0.51 0.53 0.55 0.57 0.59

Acceleratio

n (g)

Time

Line 2 Coil Springs ‐ Over Fastener

ModelTest



♦ Slab displacement shows good agreement between model and test. About 10 kN/mm.

‐0.004

‐0.0035

‐0.003

‐0.0025

‐0.002

‐0.0015

‐0.001

‐0.0005

0

0.0005

‐2 3 8Displacemen

t (m)

Time (s)

Line 2 Coil Springs

DZ07model


® Line 2 Conclusions

♦ The maximum primary stiffness in the worn condition should be maintained at 10 kN/mm at each axle box or less for Line 2 Millennium Fleet cars. This is consistent with the worn stiffness measured at MSP’s current primary suspension maintenance interval. To ensure safety at this stiffness level further recommendations are offered:●MSP should continue to maintain a maximum track twist of 12

mm in 4 m on all their tracks.●This high level of primary stiffness makes the vehicle more

susceptible to deterioration in performance caused by asymmetric primary suspension setup. Use of a truck twist index computed from wayside force measurements can identify trucks with abnormal primary suspension setups.


® Additional Recommendations

♦ Simulations of turnout defects show that 1:8 and 1:9 turnouts with defects longer than 1.6m and 1.0m respectively have a tendency for flange climb derailment.

♦ MSP should refine surface finish requirements for wheel reprofiling to minimize the effects seen during the line 3 Cobrasma empty curving test.

♦ MSP should consider pursuing a Wheel/Rail profile study. A wheel profile with a steeper flange angle over a longer distance on the flange would increase the L/V ratio and distance to climb required for flange climb derailment. Using a profile design based on the worn shape may also reduce wheel wear, although care must be taken to avoid rolling contact fatigue of wheels and rails.

05 Britto's Line 2 3 Update to MSP 2013-05-06

Documents

Transcript of 05 Britto's Line 2 3 Update to MSP 2013-05-06