Developing an Objective Measure for Flushing on Chipseals

Dave Whitehead – Senior Asset Manager

NZTA National Office

Developing an Objective Measure for Flushing on Chipseals

Background to Study

Flushing is a major reason for resurfacing in NZ.

Flushing = Low texture.

Flushing = Low skid resistance.

Chipseals account for around 90% of SH surfacing

Background to Study - cont

Good to identify an objective measure of texture that correlates well with the reduction in skid resistance, so that:

Quantify flushing.

Targets for flushing can be established

Strategies can be evaluated

Appropriate levels of service for texture values can be set to maintain skid resistance and safety on the network

Methodology – Data Collection

Data from our annual SCRIM+ survey of the SH network.

Seasonal correction sites used for the study data.

Skid Resistance – SCRIM coefficient (SC)

Texture – survey collects 6 different measures.

– Mean Profile Depth (MPD)– modified MPD– Average Peak Count– Material Ratio – 10m average Sensor Measured Texture Depth (SMTD) – Texture Variability

Methodology - Visual Evaluation

0.00

0.10

0.20

0.30

0.40

0.50

0.60

0.70

0.00

0.50

1.00

1.50

2.00

2.50

0 500 1000 1500 2000 2500 3000 3500 4000 4500 5000

SC

Textu

re M

easu

re

Distance (m)

MPD SC

Double dip

Methodology – Evaluating the Data

Methodology – Selecting Sites for Analysis

Only sites with low texture selected.

>10% of site has low texture defined as

•≤0.5mm SMTD•≤0.7mm MPD and modified MPD•Material Ratio >40

32 sites meet these criterion and these formed the study dataset.

Methodology – Statistical Evaluation

Regression is a common test for determining the correlation between variables.

Needs to be a relationship between SC and texture for all values.

Therefore an alternative test, Pearson’s Chi squared test, was used to study dependency.

The Chi squared (X2) is used to assess independence by testing paired observations on two variables.

Methodology – Statistical Evaluation

In the X2 test the expected and the observed frequency is compared as shown in the equation.

Where:Χ2 = the test statistic that asymptotically approaches a X2 distribution. Oi = an observed frequency; Ei = an expected (theoretical) frequency, asserted by the null hypothesis;

n = the number of possible outcomes of each event.

Methodology – Statistical Evaluation

The table value for the null hypothesis is 10.827 so this can be rejected and we can conclude there is a dependency between the variables.

Wheel PathTexture Measure

MPD Mod MPD

SMTD Material Ratio

X2 X2 X2 X2

Left Wheel Path 16759 4333 20549 11080

Right Wheel Path

10143 3431 16982 7351

Methodology – Setting Texture Threshold Levels

Optimum texture thresholds were evaluated by determining % of 10m lengths below a specified SC at specific texture values in each wheel path.

SMTD(mm)

SCRIM Coefficient (SC)

0.30 0.325 0.35 0.375 0.4

LWP RWP LWP RWP LWP RWP LWP RWP LWP RWP

0.3 39.4%

30.4%

47.4%

34.6%

56.0%

39.1%

60.8%

42.5%

67.0%

45.3%

0.4 29.2%

24.3%

39.5%

31.0%

50.5%

38.8%

57.4%

44.0%

64.2%

48.9%

0.5 18.8%

15.4%

26.8%

21.4%

36.6%

28.7%

43.9%

34.2%

51.2%

39.6%

Detailed Visual Evaluation

0.00

0.10

0.20

0.30

0.40

0.50

0.60

0.70

0.00

0.20

0.40

0.60

0.80

1.00

1.20

1.40

1.60

1.80

2.00

2.20

2.40

3500 3600 3700 3800 3900 4000 4100 4200 4300 4400 4500

SC

Textu

re M

PD

an

d S

MTD

(m

m)

Length metres

MPD LWP SMTD LWP SC LWP

A

Detailed Visual Evaluation

Point A from Graph

Methodology – Quantifying Flushing

Texture or SC alone is no guarantee that flushing exists but together should produce a satisfactory measure.

Challenge to set measures at levels with acceptable levels of false positives and negatives.

SC Value of <0.35 with SMTD ≤0.4mm and MPD ≤0.7mm appear an appropriate initial values to test using video.

Methodology – Quantifying Flushing

Various thresholds for both texture measures were tested against SC to check the false positives and negatives.

Final assessment using SC<0.40 and SMTD ≤0.5mm expanded sites to 370 with 10% false positives.

Applying above to LWP of data set were able to quantify % meeting the criteria.

Summary of Conclusions

Objective of the study was to determine if texture data could be used to identify flushing.

Flushing is normally associated with a significant reduction in skid resistance.

Three years of seasonal site data was evaluated on 6 texture measures with 4 found to show correlation at low texture and skid resistance.

Using the video, correlation was also found between low texture and flushing.

Sites without low texture were removed from the study.

Summary of Conclusions

4 texture measures were evaluated to ensure a statistical dependency between them and low skid resistance.

SMTD and MPD found to be best and these were used for the rest of study.

Quantitative evaluation to determine proportion of SC at particular levels in bands of the SMTD and MPD.

Since texture or SC alone is no guarantee of flushing both were used to set threshold levels.

Review of the videos found that SC<0.40 and SMTD ≤0.5mm were most useful identifying areas starting to flush.

Recommendations

Threshold

Texture SC CommentsSMTD MPD

1 ≤0.50 ≤1.0 ≤0.40 To identify areas that are starting to flush

2 ≤0.40 ≤0.7 ≤0.35 To identify areas that are fully flushed

The following are recommended threshold levels for texture and skid resistance to identify flushed areas.

Also recommended that SMTD be used as a texture measure rather than MPD for identifying flushing.

Recommendations

Further recommendations

Validation of the criteria derived from this study is undertaken using a significant sample from the 2010/11 annual survey of the State Highway network.

Subject to validation consider a threshold report identifying flushed areas as a future standard deliverable.

Developing an Objective Measure for Flushing on Chipseals

THANK YOU