Over-run Task Force: Improvement of test method accuracy - - PowerPoint PPT Presentation

Over-run Task Force: Improvement of test method accuracy

EDEN-group meeting called by Statens vegvesen Vegdirektoratet Oslo, 25th of September 2015

Steven MICHELS Head of Fleet Testing Goodyear S.A.

History of the Over-run test

The legislator wanted to characterize road wear caused by studded tires.

• Overrun test method has been used since 1986 for determining the road

wear of studded tires

• Test method was developed by VTT
• Many different test samples and materials were tested like cylindrical

stones, various metals, asphalt mixture stones glued on epoxy plates

• Granite of Kuru was chosen because its quality is very uniform and the

crystal size is small.

• The matrix-like surface was introduced to imitate the crushed stones used

in asphalt mixture.

• VTT studies found that the test correlates very well with actual road wear.
• The test is done in real life conditions on a wet surface, representative of

conditions while studded tires are used

2

Over-run test description

Repeat steps 1 to 6 Issue report 1 2 3 4 5 6 7 8 9a 9b

3

Trafi study presented at EDEN meeting in 2014

There seems to be: systematic variation in averages systematic variation in deviation Random variation may exceed the confidence limits given in data Reproducibility: Individual over-run test results and the related 95% confidence limits in measurements taken in 2013 Repeatability: 5 measurements in autumn 2014

Reference: Variation in over-run test results based on measurements in 2013 – 2014, EDEN expert meeting in Helsinki, 27.11.2014 (Riikka Rajamäki, Trafi)

4

Mandate & Objective:

One of the outcomes of the 2014 EDEN meeting in Helsinki was that Trafi invited all accredited laboratories to join a Task Force with the objective to further develop the accuracy of studded tires road wear testing. recommendations to be submitted to Trafi by end 2016

5

Members

All Laboratories currently accredited by Trafi as recognized experts for Stud Type approval are welcome to join the Task Force. Currently participating in this joint effort:

• Tikka Spikes, Continental

J Rautiainen, T Becherer

• Nokian Tyres plc

M Liukkula

• BD Testing

I Halén

• TestWorld

M Hilli

• Goodyear

S Michels

• Trafi (observer)

M Loponen

6

Potential sources of variation

• Interpretation of current test method description & calculations
• Repeatability of test itself:
• Testing operation
• Test conditions
• Driving style
• Weighing precision
• Reproducibility between different accredited laboratories
• Stones
• Geometry
• Origin
• Influence of vehicle

(Other sources might be identified during the test analysis)

Total variation Process variation Measurement system variation Repeatability Reproducibility

7

Approach

Increase repeatability

• Reduce sources of variation
• Increase total wear in order to reduce impact of variation

Increase reproducibility

• Identify discrepancies between test laboratories
• Align recognized experts on test details that have, so far, not yet been

described in the test procedure

8

Test plan

Stage 1. A1 - Sample geometry: grove depth A2 - Sample geometry: grove width and block size A3 - Stone preparation and weighing Stage 2. B1 - Round Robin test B2 - Vehicle influence (propulsion type: FWD, RWD, 4x4) B3 - Different stone manufacturers B4 - Confirmation of A2 Stage 3. C1 - Influence of sprinkling the entire surface vs. samples only: coming up. Same test tire model used for entire test campaign

9

A1 – grove depth

Geometries: Labra-0058 (5 mm grove depth, std dimension) K05 (3 mm grove depth)

Labra-0058 K05

10

A1 – grove depth

Average mass loss of normal stones was 1,043 g Average mass loss of K05 was 0,844 g (23.58% less, will be used as linking ratio later) The lower wear of K05 was probably caused by the fact that block edges are less fragile when they are supported better.

11 0.00 0.20 0.40 0.60 0.80 1.00 1.20 Labra-0058 K05

Test A2. Final result (average row wear) [g]

A2 – influence of test stone geometry

Geometries: K05, K01, K03 (all with 3 mm grove depth) Objective is to verify the influence of the total edge length per sample

1 set of K05, 1274 mm total edge length, 144 corners 2 sets of K03, 1680 mm, 224 corners 2 sets of K01, 1800 mm, 360 corners

12

A2 – influence of test stone geometry

Link between mass loss, total edge length and number of corners?

1274 0.82 1680 0.93 1800 1.35 0.00 0.20 0.40 0.60 0.80 1.00 1.20 1.40 1.60 1200 1300 1400 1500 1600 1700 1800 1900

edge length vs. mass loss

edge length 144 0.82 224 0.93 360 1.35 0.00 0.20 0.40 0.60 0.80 1.00 1.20 1.40 1.60 100 150 200 250 300 350 400

number of corners vs. mass loss

# of corners 13 0.82 0.93 1.35 0.00 0.20 0.40 0.60 0.80 1.00 1.20 1.40 1.60 1.80 K05 K03 K01

Test A2. Final result (average row wear) [g]

A3 – Stone preparation and weighing

Geometry: K05 Laboratory test - each laboratory performs the initial measurements, sends the stones to Lab E to over-run test. After Over-run test, Lab E sends the stones back to same laboratories to make the final measurements.

14

A3 – Stone preparation and weighing

ABCDE ABCDE Avg: 0.749 g 0.733 g Reproducibility: 0.073 g 0.033 g

15

We found out that Lab D did not fill the oven with “dummy” stones, but only the 15 + 5 test stones. This practice was so far not specified in the test method description. The oven capacity should always be fully used. Any free places at the oven should be filled with wet “dummy” stones Significant potential for increased reprodicibility.

0.72 0.71 0.73 0.81 0.76 0.00 0.20 0.40 0.60 0.80 1.00 1.20 A B C D E

Test A3. Final result (average row wear) [g]

B1 – Round Robin test

Geometry: K05 Vehicle: VW Golf VII 1.4 TSI – manual gearbox 2 samples per laboratory

16

ABCD ABCD Measurement system variation: 0.234 g 0.152 g Reproducibility: 0.198 g 0.088 g

B1 – Round Robin test

0.75 0.71 0.83 0.96 0.00 0.20 0.40 0.60 0.80 1.00 1.20 A B C D Test B1. Final result (average row wear) [g]

Lab D is watering the full track Significant potential for improving reproducibility through test conditions alignment between Labs. This practice was so far not specified in the test method description.

17

B2 – Influence of test vehicle drive axle

Geometry: K05 Vehicle type: FWD, RWD, 4x4 Test vehicle was the same during the 3 sessions (Toyota Hilux).

Remark: During the 4x4 test, we were lacking K05 stones so only 1 of 5 sets was composed of K05 stones, the other 4 were std stones. Therefore, we applied a correction factor of 1.2358 (found during A1) to the 4 std sets.

18

B2 – Influence of test vehicle drive axle

The propulsion type seems to influence the total mass loss.

19

Propulsion type stdev (g) FWD 0.024 RWD 0.041 4x4 1 sample 4x4, using wear ratio found in A2 0.028

0.989 0.938 0.831 0.845 0.00 0.20 0.40 0.60 0.80 1.00 1.20 FWD RWD 4X4 4x4 (A2 ratio) Test B2. Final result (average row wear) [g]

B3 – Influence of test stone supplier

Geometry: Labra-0058 (std stones) Material: Kuru grey granite Suppliers: 2 different suppliers Test vehicle: VW Passat Conditions: wet track

Remark: 1 set of samples has been compromised and was therefore excluded from the data set.

20

B3 – Influence of test stone supplier

• 4.3% difference between results from samples sourced from Lab A and from Lab B.
• Wear appearance after visual inspection seems to be slightly different (size of lost grains).
• Influence of test stone suppliers seems to be within normal test variations.

21 1.099 1.053 0.00 0.20 0.40 0.60 0.80 1.00 1.20 Lab A Lab B

Test B3. Final result (average row wear) [g]

B4 – Influence of stone geometry

Geometries: K01, K03 Confirmation of A2

2 sets of K03, 1680 mm, 224 corners 2 sets of K01, 1800 mm, 360 corners

22

Link between mass loss, total edge length, number of corners and net surface? Data added to the A2 data set.

B4 – Influence of stone geometry

23 1.61 0.96 0.00 0.20 0.40 0.60 0.80 1.00 1.20 1.40 1.60 1.80 K01 K03

Test B4. Final result (average row wear) [g]

144, 0.82 224, 0.94 360, 1.48 0.00 0.20 0.40 0.60 0.80 1.00 1.20 1.40 1.60 100 150 200 250 300 350 400

mass loss vs. number of corners

# of corners 1274, 0.82 1680, 0.94 1800, 1.48 0.00 0.20 0.40 0.60 0.80 1.00 1.20 1.40 1.60 1200 1300 1400 1500 1600 1700 1800 1900

mass loss vs. edge length

edge length 2812, 0.82 3150, 0.94 2250, 1.48 0.00 0.20 0.40 0.60 0.80 1.00 1.20 1.40 1.60 1200 1700 2200 2700 3200 3700

mass loss vs. net surface

net surface

• A1: The grove depth of the samples seems to influence the mass loss.
• A2: The grove width of the samples seems to influence the mass loss.
• A3: Weighing operations seems to be consistent once details were aligned
• B1: Reproducibility can be further improved through technical alignment

and more specific description of test method. Potential source of variation still to be confirmed, ongoing.

• B2: The propulsion system of the vehicle seems to influence the mass loss.
• B3: influence of sample sourcing seems to be minor, to be continued.
• B4: Confirmation of A2, link between mass loss and geometrical parameters
• f test samples

Conclusions, 1/2

24

Conclusions, 2/2

• Watering of the complete track seems to increase the weight loss. When

the tires are cooler, the stud impact increases as the layer under the stud is stiffer.

• During the Over-run test, the studs are causing wear most commonly at

sample edges.

• Drying conditions influence mass loss measurements (ventilation, capacity).

We are convinced the current test method’s accuracy can be significantly improved by:

• More specific description of test method and test conditions,
• Elimination of identified sources of variation.

25

Already implemented

(currently agreed between Labs, not yet reflected in regulation)

• New standardized report format proposed by Trafi.
• Trafi supplied also a template for the calculation of the correction in order

to prevent different interpretations.

• All recognized experts agreed to always use the full capacity of their drying
• vens.
• All samples have to be positioned in the oven in a way that unites

maximum ventilation and a stable positioning.

26

Next steps

2015:

• Dry vs. wet test (C1), scheduled 3rd quarter 2015
• In-depth analyses of test results from stages A & B
• EDEN meeting, 25th of September
• Task Force to decide about next test campaigns, based on findings of stages A & B

2016:

• Continuation of Task Force test sessions
• End of the year: Task force to submit recommendations to Trafi

2017:

• Trafi to analyze and adopt recommendations

27

Recommendations

• Focus on 100% technical alignment between test laboratories regarding the test

method and test conditions.

• Once more influencing factors are confirmed, the test method description should

be further elaborated to specify test conditions in an even more detailed manner.

• Continue open dialogue between Task Force members and Trafi. The task force is
• pen for other administrations to join as observers.
• Harmonisation of legislation related to stud type approval for all Nordic

countries.

• VTI study shows link of road wear and PM particle emission
• Russian customs union is going to implement a new legislation regarding road wear by

studded tires, based on the Over-run test principle.

• In case there are signficiant changes in legislation or test method, the tire

industry will need a sufficient lead time to implement the required changes.

28

Thank Thank Thank Thank you you you you! ! ! !

29

Annex

• Repeatability or test–retest reliability[1] is

the variation in measurements taken by a single person or instrument on the same item, under the same conditions, and in a short period of time.

• Reproducibility is the ability of an entire experiment or study to be

duplicated, either by the same researcher or by someone else working

• independently. Reproducing an experiment is called replicating it.

30