Portorož, Slovenia • Alex Wright Alex Wright • TRL Infrastructure Division TRL Infrastructure Division • Group manager, Technology Development Group manager, Technology Development • [email protected][email protected]Developing the automatic measurement of surface condition on local roads
29
Embed
Portorož, Slovenia Alex WrightAlex Wright TRL Infrastructure DivisionTRL Infrastructure Division Group manager, Technology DevelopmentGroup manager, Technology.
This document is posted to help you gain knowledge. Please leave a comment to let me know what you think about it! Share it to your friends and learn new things together.
Developing the automatic measurement of surface condition on local roads
Portorož, Slovenia
Measuring condition at traffic-speed in the UK
o UK condition surveys measure• Longitudinal profile• Transverse profile• Texture profile• Cracking (automatic)• Geometry
o Annual coverage • TRACS: 40,000km motorway and trunk
roads • SCANNER: 80,000km local road network
o Surveys carried out to an end result specification
Portorož, Slovenia
“UK” Systems
• Accredited Systems:• Jacobs
– Ramboll RST26, RST27• WDM
– RAV1, RAV2, RAV3, RAV4• DCL
– Roadware ARAN1, ARAN2
Portorož, Slovenia
UK trunk roads - TRACS
Portorož, Slovenia
UK local roads (rural) - SCANNER
Portorož, Slovenia
UK local roads (urban) - SCANNER
Portorož, Slovenia
Use of the Data
o Local use• Parameters reported over 10m lengths for local use
o Network use• For trunk roads total length of poor values reported
• Single HA performance indicator (PI)
0
10
20
30
40
50
60
70
80
90
A
Road category
Pro
port
ion
(%)
Red
Amber
Green
• For local roads a Road Condition Index (RCI) is produced every 10m
• Reports “overall” condition score• Distribution of RCIs over the local
authority defines network condition (LA Indicator)
• Potential use in allocation of funding across authorities
Portorož, Slovenia
o Local roads differ from trunk roadso New methods required to maximise value of local road datao Research to improve the use of the survey data
• Measuring ride quality on local roads using shape data• Using texture to assess surface deterioration on local roads• Measuring edge deterioration on local roads
o Work concentrated on the use of shape datao Began with consultation to find out what users needed in
practice
Enhancing the use of data from local roads
Portorož, Slovenia
“Shape” data collected at traffic-speed
6566676869707172737475
Chainage (m)
Portorož, Slovenia
o Consultation with engineers found that• Little importance placed on longitudinal profile data• Key structural measure is cracking and rutting• Engineers desire a reliable assessment of general ride
quality (functionality)• But engineers key concern is defects giving rise to
bumps (user complaints)o Concluded that methods needed to
• Reliably identify lengths with poor ride quality• Identify general locations giving rise to bumps
Measuring ride quality on local roads - consultation
Portorož, Slovenia
Measuring ride quality - data collection
o A practical investigation to relate surface profile to user opinions on local roads
o Several routes surveyed, including sections known to be pooro Profile data provided by HARRIS1 profilometer
• Measurements in both wheel tracks (and across survey width)o User surveys:
• Car surveys• Motorbike survey• Utilising on-board data collection
with GPS referencing• Reported on ride and bumps• Repeat surveys for consistency
Portorož, Slovenia
Considering general ride quality
3m 5m0.000000001
0.000001
0.001
1
1000
1.00E-01 1.00E+00 1.00E+01 1.00E+02
Wavelength
Po
wer
100m lengths where dial >2100m lengths where dial <=2
0.37
5916
0.49
5487
0.65
3091
0.86
0827
1.13
4638
1.49
5543
1.97
1245
2.59
8258
3.42
471
4.51
4041
5.94
9865
7.84
2396
10.3
3690
2
42400
42410
42420
4243042440
42450
42460
42470
42480
4249042500
42510
42520
42530
42540
4255042560
42570
42580
42590
42600
4261042620
42630
42640
42650
Wavelength (m)
Sit
e C
hai
nag
e (m
)
1.38-1.4
1.35-1.38
1.33-1.35
1.3-1.33
1.28-1.3
1.25-1.28
1.23-1.25
1.2-1.23
1.18-1.2
1.15-1.18
1.13-1.15
1.1-1.13
1.08-1.1
1.05-1.08
1.03-1.05
1-1.03
0.98-1
0.95-0.98
0.93-0.95
0.9-0.93
o Wavelet Decomposition
o PSD
1m – 5m
o IRI, Ride Number, Profile Indexo MA and enhanced varianceo Coefficient de planeiteo Waveband Energyo Standard Deviation
Portorož, Slovenia
General ride quality - wavelength response
o IRI
o 3m Variance
Portorož, Slovenia
Parameter for general ride quality
o Predicting general ride quality on local roads• 1-5m wavelength features cause the users most
discomfort. • 3m enhanced variance agreed best with user opinion of
underlying ride quality. Other measurements agreed no better with the user’s opinion.
• 10m enhanced variance showed some agreement (effects of longer wavelengths on truck drivers).
• Wavelengths over 20m - little or no agreement with usero Effect of measurement (line)
• Offside measurements contributed to 33% of agreement with user opinion.
• Multiple measurement lines around the wheelpath did not improve agreement
Portorož, Slovenia
Measuring “Bumps” on local roads
o User surveys recorded bumps using button presseso Wavelet analysis suggested wavelengths of interest lie
between 1 and 3m.o Existing measurements (variance, IRI etc) did not reliably
report the locations of the features causing this bump-like discomfort.
o Considered many approaches, e.g.• 1.25m enhanced variance, change of vehicle acceleration,
derivative of longitudinal profile (features too small to impact on a car’s tyre)
o The Central Difference Method • Calculates a “derivative” for each point along the road (profile
measurements {yi}, taken at distances {xi} along the road):
• Similarly for F’’.• The maximum of these values is calculated over 1m lengths.• If max(F’) and max(F’’) both exceed set thresholds, then the
length contains a bump and a value of “1” is reported for that length. Otherwise “0” is reported.
11
11)('
ii
ii
xx
yyixF
Portorož, Slovenia
Measuring “Bumps” with the CDM – local roads
o Tests to review locations where the bump measure responded• Reported 84% of user button presses.• Potential high number of false positives.• Inspection of 3D profile and video showed features of note
where CDM responds, but users had not always pressed the button.
o Concluded • This is an appropriate method for identifying “bumps”.• We should use a combination of this and 3m enhanced
variance for assessing general ride and bump density on local roads