1. Development and Use of a New Pavement Strain Coil Measuring System at CAPTIF
David Alabaster
Principal Pavements Engineer
New Zealand Transport Agency

2. Outline Background on NZ interest in Strain measurement New System
Installation Procedure
Installation Logistics
Data Collection Procedures
Reliability of Data
Data Analysis Approach
Changes to Agency Practice
Conclusions

3. Background The New Zealand Transport Agency is part of AUSTROADS
Our design models relate elastic strain imposed by a design load at critical points in a pavement to allowable loading
Critical Elastic Strains
Vertical Compressive strain top of Subgrade
Horizontal Tensile Strain bottom of Bound layers
Horizontal Tensile Strain bottom of AC layers
Understanding Elastic Strain Behaviour is thus important!

4. Previous Systems Used at CAPTIF
1970’s Bison System
1990’s Emu System
2000’s National Instruments LVDT

5. Pavement Strain Coil Measuring System
Greenslade, F Development of a New Pavement Strain Coil Measuring System at CAPTIF. The Roles of Accelerated Pavement Testing in Pavement Sustainability

6. Planning

7. APT Installation Procedure

8. Field Installation Procedure

9. Installation Logistics
Coils are made locally. $NZ50 - $100. (0.5 hr/coil)
Calibrate Coils (10 mins / coil pair).
Locate Coils (2 min - 2 hr /stack).
Install Coils (15 mins – 1 hr / layer)
Testing.
At CAPTIF:
– 40 coil pairs (10 mins total)
In Field:
- Beam truck – Slow test (cycle time)
- Beam truck – Fast test (cycle time)
- Network traffic (trigger time)

10. CAPTIF Collection Procedure
The dynamic strain software operates as follows:
Setup test parameters on front panel.
- vehicle load / speed / offset / laps
2. The software is ‘RUN’.
- loads coil layout file.
- switches to first pair of the array.
- waits for a trigger and data acquired.
- software plots strain measured.
- Next coil pair in layout file selected.
Set and forget for 40 coil pairs across 4 test sections.
Static Strains – similar – but monitored at set time intervals.

11. Reliability of Data Accuracy
- Integrate elastic strain to get deflection
- Integrate plastic strain to get deformation
2. Sensor failure rate < 5 pairs in 17 years
3. Precision / Repeatability
1% (25 High Strains (2500 Microstrain)
- 8% (3 Low Strain (50 Microstrain)
4. Reproducibility ?

12. Hilab Data Reliability

13. Hilab Data Reliability

14. Hilab Data Reliability - FWD

15. Data Analysis Approach
Comparisons between sections
Comparison with other sensors
Trend Monitoring
Comparison with Linear Models
Comparison with non-linear models

16. Comparisons between Sections

17. Sensors Comparison – Moisture and Suction

18. Sensors - B Permanent Strains

19. Trend Monitoring

20. Simple vs Complex Models
Arnold G (2004) Rutting of Granular Pavements
Stevens B (2005) The Development and verification of a pavement response and performance model for unbound granular pavements
Hayward B (2006) Investigation of Road Base Shear Strains Using In-Situ Instrumentation
Gonzalez A (2009) An Experimental Study of the Deformational and Performance Characteristics of Foamed Bitumen Stabilised Pavements
Werkmeister S (2009) Prediction of Pavement Response using Accelerated Test Results of New Zealand’s CAPTIF Facility.
Hussain (2012) Performance of unbound granular basecourse materials under varying moisture conditions

21. Changes to Agency Practice
Lab to Field modelling
- need realistic elastic model (Austroads FEM)
Residual Life Determinations
- behaviour more complex than current models
Foamed Bitumen Stabilisation
- much better than unbound
Precracking 4% cement bound base
- elastic strains quickly reverted to unbound levels
Dacite – meets spec – performs poorly
- elastic strains crazy! Confirms ban.

22. Conclusions Strain Coil Systems are:
Easy to build system (and cheap sensors)
Easy to install
Easy to use
Produce reliable data
But most important
generate data that changes your thinking
Use with deflection testing to extend coverage.
Elastic and Plastic Strains are key parameters in our design systems.
Faster, easier and cheaper to measure strains than infer them.

23. Thanks for listening

24. Coil Monitoring Circuit Diagram
The inductive coils function electrically in a very similar way to a displacement sensor known as the Linear Variable Displacement Transformer (LVDT). LVDT’s are commonly used in scientific and industrial applications. The electronic technology to measure LVDT’s has become very accurate and reliable. Because of this similarity it is possible to use the electronic circuits that have been devised to work with LVDT’s with the strain coils. Analog Devices Incorporated produce an integrated circuit (IC) the AD698. The AD698 is described as ‘Universal LVDT Signal Conditioner’. This IC provides the essential functions to create an LVDT measurement system and hence can be adapted to drive and receive the strain coil technology. The AD698 is the heart of the new CAPTIF system. An electronic circuit (fig 2.) was designed with the AD698 IC being supported by some peripheral components. External components were required to set the transmitter coil frequency and excitation. A nominal frequency of 10 kHz was chosen with an excitation voltage of 9VAC (pk-pk). Driver amplifiers (LT1010) were used to buffer the output from the AD698 to the low impedance of the transmitter coil. An instrumentation amplifier (OPA2277) was used to amplify the very small signal returned from the receiver coil.

25. Multiplexing
At CAPTIF strain coils a placed in the pavement as an array (fig 3). Any of the coils can be defined as a transmitter or a receiver. Each coil pair is measured synchronously therefore a multiplexor must be used to route the correct transmitter and receiver coils to the signal conditioning circuit. A simple ‘reed relay’ component is used with a driver IC (fig 4). A digital control signal is sent to turn on each relay as required.

26. Dynamic Strain Measurements
Transient loading events causing a dynamic resilient strain can be captured with the strain coils. The data acquisition hardware is set up to acquire data at the rate of 5000 samples per second for a period of 2 seconds. The data is post processed with a 16th order low-pass Butterworth Filter at a cut-off frequency of 10Hz. Typically, displacements of less than 20µε are discernible (fig  7).

27. Typical Install at Post-mortem

28. Simple and Complex Models

29. Future Needs
Even thin asphalt layers influence the stress and strain distribution within basecourses.
In situ measurements of stress and strain in the subgrade showed a gradual increase with increasing load cycles.
Strains measured in the basecourse can either decrease or increase with loading indicating a stiffening or softening is possible.
Loading rate and recovery effects on subgrades need to be investigated.

30. Coil Sensors
The inductive coils consist of a circular plastic disk with a groove cut into the perimeter in which copper wire is wound to form an inductive circuit (fig1.) The coils can be customised to any diameter with any number of turns of wire. CAPTIF has settled on a configuration that uses an HDPE 300 plastic disk 60mm in diameter and 7mm thick. A groove is cut 3.5mm wide and 12.5mm deep. Then 350 turns of 0.25mm diameter enameled copper wire is wound into the groove. A flying lead cable of a suitable length is attached and the completed coil is sealed in marine grade polyurethane.

31. Coil Pairs/Arrays
The inductive coils are used as a pair. One of the pair is used as a transmitter in which an alternating current is driven from an electronic source. This has the effect of generating an alternating magnetic flux field around the coil. The receiver coil is placed within this magnetic flux field and this causes an alternating current to be generated within the receiver coil. This phenomenon is known as mutual induction. The magnitude of the generated signal is proportional to the distance between the coils. This relationship is not linear as the magnetic flux density gradient is not linear. The strength of the receiver coil output is measured by electronic detection circuits. Thus a pair of coils will provide a non-contact strain measuring device. The magnetic flux occurring in the coil pair can be influenced by metallic objects and external magnetic radiation. It is advisable to keep metallic objects at least 300mm away.

32. Calibration Jig
Both the voltage output from the coil pair and the displacement transducer are recorded at a rate that gives approximately 2000 samples over the cam range. The data can then be processed to give an XY plot (fig 6) and then a unique calibration equation is calculated.

33. Calibration curve
The relationship between the coil voltage and displacement is non-linear because the magnetic flux density gradient is not linear. At CAPTIF the data is processed using a Natural Logarithm Best Fit function which returns an equation of the form:-
y = ln(x) + b
where y = mm, and x = volts.

34. DAQ
NI CDAQ9174 chassis which communicates to a control PC via the USB bus. Three modules are used:-
The NI9239 is a 24bit analog to digital converter (ADC) module. This is used to measure the voltage signal coming back from the receiver coil via the signal conditioning circuit board.
The NI9403 is a digital output module. This is used to select the correct coils via the relay circuit.
The NI9215 is an 8 channel 12bit ADC. This is used for peripheral associated devices such as a displacement sensor when calibrating, temperature sensors and triggers.
National Instruments Labview software controls the data acquisition system.

35. Installation Logistics

37. Installation Logistics
Coils. (0.5 hr/coil)
Coils are made locally. Approx $NZ50 - $100.
Coil bobbins machined from plastic
Coil bobbins wound with wire
Tails joined to bobbin.
2. Calibrate Coils (10 mins / coil).
Coils named and mounted on jig
Calibration jig turned
Coils demounted
3. Locate Coils (2 min - 2 hr /stack).
Establish control points for stack
Line up coil locations
Could use coils to locate themselves (using signal strength)
4. Install Coils (15 mins – 1 hr / layer)
Build layer
Dig hole / Wire trench
Place coils
Sieve material (if required)
Replace material
Compact
Testing.
At CAPTIF:
– 40 coil pairs (10 mins total)
In Field:
- Beam truck – Slow test (cycle time)
- Beam truck – Fast test (cycle time)
- Network traffic (trigger time)

38. CAPTIF Collection Procedure
The dynamic strain software operates as follows: 1. The setup parameters for the test are entered on the front panel. 2. The vehicle loaded as required and run up to the required speed. 3. The software is ‘RUN’. 4. The software loads the coil layout file. 5. The software switches in the first pair of the array. 6. The software waits for a trigger. 7. Data is acquired for that pair.
8. The software plots a graph of the coil strain measured.
9. The software then switches to the next coil in the layout file and waits for a new vehicle trigger.
10. The process is repeated until all the coil pairs have been tested.
11. The full set of coils will be tested again a number of times according to the number of LAPS set.
12. A data file is stored on the hard disk.
Basically set and forget for 40 coil pairs across 4 test sections in current project