1. Traffic speed deflection data applied to network asset management
The Kaikoura Bypass Reality Check

2. Introduction Traffic Speed Deflection Data:
10 m centres
Accurate and repeatable
TSD “dynamic” bowls -> equivalent FWD “static” bowls
2015 TSD data has been used to produce two FWPs:
Empirical Drainage Study (William Gray / Elke Beca + NZTA)
Regional Calibrated Mechanistic Study (Martin Gribble, NZTA)
NZTA’s Kaikoura Bypass project is a prime opportunity for a “reality check” to compare both methods on a 800 lane km “test track”.

3. Traditional Empirical Procedure
RAMM
High Speed Data
Visual Condition
Traffic Data
Geometry
ONRC
Maintenance Costs
SNP
Data QA,
Historic Evaluation
Functional Evaluation (surface)
dTIMS Surface Performance
Coordinated Outputs:
FWP for Resurfacings
Preliminary Specified FWP from Visual Surveys
Functional Levels of Service
Rut, IRI (Skid)
RCA Requirements

4. Integration of Mechanistic Procedure
RAMM
High Speed Data
Visual Condition
FWD & TSD
Traffic Data
Geometry
ONRC
Maintenance Costs
Mechanistic Models
Data QA,
Historic Evaluation
Functional Evaluation (surface)
Structural Evaluation (sub-surface)
dTIMS Surface Performance + Subsurface added from Mechanistic- FWP
Data QA, Validation of Terminal Sites (RAPT), Regional Precedent Performance for Fatigue Criteria
Loading (ESA)
Mechanistic Analyses
(Moduli, Stresses & Strains)
Remaining Structural Life
Additional Outputs:
Sub-surface Drainage Priorities
Quantified HPMV Route Impacts
Coordinated Outputs:
FWP for Resurfacings &
Validated Specified M-FWP for Renewals
Feedback loops to optimise Structural FWP
Preliminary Specified FWP from Visual Surveys
Functional Levels of Service
Rut, IRI (Skid)
Structural Levels of Service
% Terminal Fatigue, % Excessive cost
RCA Requirements
Supplementing surface condition with subsurface structural condition.
Mechanistic details (moduli, stresses and strains thoughout the pavement) are readily incorporated into traditional procedures.
Outcome: State-of-the-Art asset management principles (as being implemented in Europe, USA and South Africa)

5. Traditional Analysis – Distress Modes
Empirical or traditional mechanistic approaches: Consideration of only 1-2 criteria for pavement life prediction (e.g. SNP or subgrade strain)
Through extensive data collection at high speed (incl. TSD): Mechanistic approach enables more criteria and multiple distress modes to be considered and calibrated to region or sub-region (using methodology of the Regional Precedent Performance (RPP) Study recently undertaken for NZTA on 5 of their regional networks).

6. Multiple Distress Criteria
Rutting
Roughness
Degradation
Flexure Shear
Andrew Dawson
International workshop for the development of Mechanistic Design Methods for Unbound Pavements
Very substantial progress in the last 3 years for New Zealand Unbound Pavements
Andrew Dawson, 2002

7. Kaikoura Bypass Case History
Background
Kaikoura earthquake
Inundation of parts of SH1
Need for bypass route
Original 25-Year Traffic will be experienced by mid next year
The ultimate “Reality Check” of life prediction models: Real traffic on real roads with a range of real environments.
TSD data collected in 2015
Impacts and Distress modes?
November 2016
Baseline testing TSD carried out in 2015, yet because of the increased traffic resulting from the Kaikoura bypass the originally intended 25 year traffic will be experienced by mid next year presenting a prime opportunity to observe accelerated pavement trafficking in a situation where the pavements have full exposure to rainfall and temperature changes throughout all seasons.

8. Mechanistic Analysis – Distress Modes
Observed Distress Modes from Kaikoura Bypass Maintenance
Shallow Shear Dominates
Contractors have had to maintain serviceability with digouts of certain parts of the road since the TSD testing was done in Because we can see exactly which part of the road was in trouble when we see patches (opposed to full rehab sites which may have been done on both sides of the road even though only one side was in trouble). So with patches evident we can unequivocally establish where terminal conditions have developed and correlate with the findings of the TSD analysis.
Large sector needs further subdivision.

9. Mechanistic Analysis – Distress Modes
Distress Modes from NZ Data Mining
Surfacing distress modes
19 Seal Deformation (more likely as multiple seal layers accumulate) Flexure (top down cracking in seal or thin AC) Reflection Cracking 22* Seal Flushing Scabbing/Ravelling
Economic Triggers
24 Excessive Maintenance costs for the surfacing (seal, thin AC) Excessive Maintenance costs for structural layer(s)
Other characteristics or causes that affect timing of triggers include:
26* Loading frequency effects on inter-particle bonds 27* Cement curing 28* Bitumen embrittlement (environmental ageing) 29* Subgrade –subbase intrusion 30* Frost heave 31* Particle breakdown in freeze-thaw cycles Foundation subsidence (vertical depression) Foundation slumping (lateral deformation) *7, 11, 12, 22 & Not explicitly included in current modelling Vehicle speed and temperature included for individual modes
Structural distress modes
1 Shallow Shear – Low Strength (shoving) 2 Shallow Shear – Spreading (strong but inadequate support) 3 Shallow Shear – Heave (in loose or low broken faces BC) 4 Shallow Shear – Hybrid (from above) 5 Aggregate Instability (pumping >75%S, potholing, heave) 6 Aggregate Rutting (VSD in basecourse or subbase) 7* Aggregate Weathering (mineralogical changes in fines) 8 Aggregate Degradation (physical generation of fines) 9 Cracking (conventional, bottom up) of bound layers 10 Flexure (top down cracking) of bound layers 11* Binder Curing/Hardening (aging) 12* Bond loss (cement bound reverting to unbound) 13 Subgrade Rutting (vertical deformation) 14 Subgrade Shear (lateral and vertical deformation) 15 Accumulated Deformation (multiple layers contributing) 16 Slumping/Edge Break (lack of shoulder support) 17 Roughness Progression Shrinkage Cracking (viz FBS with curing/ thermal)
Distress modes that have been observed during the RPP study when inspecting candidate treatment length for rehabilitation. Models for these are being progressively implemented. As you can see already here Shallow Shear has been observed ubiquitously and can be subdivided. For example shear can develop in a weak basecourse or in a strong basecourse that has high horizontal strains due to spreading of underlaying layers. These are the 2 predominant types of shallow shear found.

10. Mechanistic Analysis – Distress Modes
Distress Modes from Mechanistic Analysis
Distress Modes from Kaikoura Bypass Maintenance
Shallow shear dominates

11. Mechanistic Analysis – Distress Modes
Kaikoura Bypass Maintenance – Reality Check (Preliminary Calibration)
Comparison of predictions with actual digout data

12. Mechanistic Analysis – Distress Modes
Kaikoura Bypass Maintenance – Reality Check (Preliminary Calibration)
Reasons for not predicting distress:
Only network calibration, not site-specific
TSD test data averaged over 10m spacings; shallow shear often initiates for only 1-2m length
 “Dilution” of signal of distressed portion
Finer subdivision than 10m is now being explored.
Reasonable expectations: 90% reliability of predictions
80% pretty good – could be better if test spacing would be smaller. Since TSD test spacing is 10m and digouts are usually of 1-2m length, the distress might be averaged over 10m and therefore the characteristic indications would have been diluted
But now you’ll be wondering how even get to these numbers and how does this help you.
Of the total length of LWP digouts, the current mechanistic model using the TSD data does not indicate a terminal distress mechanism. Most likely reason is that incipient failure would have been indicated often over a 1-2m length but the process of averaging all intervals to 10m has diluted the characteristic indications.

13. Structural Treatment Length
PE-Grapher Output
Our Typical Evaluation Software Output
Column charts
We analyse every road individually and it will be put out in a programme called PE-Grapher. It will give us all the different distress modes, required overlay thicknesses, etc. per test point.

14. Structural Treatment Length
PE-Grapher Output
Overlay (0-300 mm)
Our Typical Evaluation Software Output
Column charts
We analyse every road individually and it will be put out in a programme called PE-Grapher. It will give us all the different distress modes, required overlay thicknesses, etc. per test point.
Remaining Life (0-25 Years)
Test Location (km)

15. Structural Treatment Length
PE-Grapher Output
Overlay (0-300 mm)
Our Typical Evaluation Software Output
Column charts
We analyse every road individually and it will be put out in a programme called PE-Grapher. It will give us all the different distress modes, required overlay thicknesses, etc. per test point.
Remaining Life (0-25 Years)
Test Location (km)

16. Structural Treatment Length Table
STL file
Summary of all structural treatment lengths
File name, start and end of section and lots of structural parameters like different lives, critical distress mode and rehabilitation options etc.
STLs ranked by rehab priority. Lots of different parameters go into the priority ranking, so there is a different tab letting the client to customize his priority ranking.
Which looks like this. It has load damage exponents, remaining lives, critical distress mode and treatment options.
It might be a bit overwhelming, but basically it’s just a summary spreadsheet of all the sections ranked by rehab priority. The one on the top is the one recommended to be rehabbed first than the one below and so forth.
Obviously there are different factors influencing this ranking, which is why there is another tab that will let you play with those.

17. Structural Treatment Length Table
STL file
Different sliders to weight influence parameters which will then rank the STL table to the requirements.
And it looks like this. It also gives you pretty overview graphs. By playing with the sliders, the ranking on the other tab is customisable.
After finding the settings that you want, the STL file will be updated and the user remaining life adjusted.

18. Google Earth File 500 m .kmz file for illustration (snip legend)
Mechanistic Life (RPP)
RPP Distress Mode = Shallow Shear - Spreading
STL Remaining Life (years) = 1
Subgrade CBR = 10
Drainage Priority = 6.2
Subgrade LDE = 6.4
Google Earth File
R – (503)
Empirical Life (dTIMS)
FWP Remaining Life (years) = 6
.kmz file for illustration (snip legend)
500 m
Navigation to the required location
Information layered, so different information can be switched on and off
Start with dTims remaining life – legend in left top corner shows meaning of colours; two lines for increasing and decreasing lanes; bright pink sections are digouts that have been undertaken since 2015 testing. Possible to click lines and get more information as shown.
Next lines show the mechanistic life prediction according to our model; same colour coding; columns show distress observed during drive-over;
Last lines show the distress mode that is most likely to trigger first for the section
Clearly obvious that by subsectioning roads further, a lot of money can be saved because the money will only be spent on pavements that actually need rehab. See green section
You can navigate yourself through the file and zoom into the section that you’re interested in. Then there are multiple things you can look at.
These lines here will show the remaining empirical life according to dtims. The legend in the top left corner tells you which colour represents how many years life. In this case, dark purple basically means no life left or “0-1”, while orange means 5-10 years. Sections that have been maintained with digouts, are shown next to the line in bright pink.
Outer lines – empirical FWP
Dual inner lines – mechanistic FWP

19. Outer lines – empirical FWP Dual inner lines – mechanistic FWP

20. Outer lines – empirical FWP Dual inner lines – mechanistic FWP
Triangular columns represent GeoSolve’s drive-over visual survey of Shallow Shear (at approx km/h).
Outer lines – empirical FWP
Dual inner lines – mechanistic FWP

21. Mechanistic-Empirical Analysis
Subsurface moduli, stresses and strains
Based on network precedent performance
Calibrated to region or sub-region evaluating pavements on the same basis throughout the country
Empirical
Surface analysis and simplified subsurface parameter
Visual evaluation based on staff familiar with the network

22. Interested? Please come and see us.
For you to take away
Mechanistic approach now allows us to analyse pavements better than in the past:
12+ distress modes
More meaningful sub-sectioning with lanes differentiated
Optimised Forward Work Programming
All of this incorporated into the Empirical FWP leads to more informed decision-making.
Interested? Please come and see us.
Thank you.

23. Where to from here?
800 lane km test track – the ultimate reality check
Highest real benefit/cost research ever for NZTA?
Significant finding after only 3 months
Ideal database leading to betterment of existing models
Additional inputs:
Accurate dates of each digout or AWT
Definitive terminal distress mechanism
Joint research with University of Queensland & TMR
Strong commitment within NPTG

24. The KMZ file used for this presentation can be downloaded from here:
The End