Project topics and materials
Get Complete Project Material File(s) Now! »
DETECTION OF PAVEMENT AGEING BASED ON EMBEDDED SENSORS – ACCELERATED PAVEMENT TESTING
Introduction
We have seen in the introduction that the establishment of strategies for sensor data interpretation is crucial for pavement monitoring and damage detection. So here we consider the accelerated pavement testing (APT) of four sections having four different bituminous layers and we use embedded asphalt strain gauges and temperature probes to detect the occurrence of fatigue of the bituminous layers.
APT is a method used to assess full scale pavement performances in few weeks or months instead of the lifetime of a road (20-30 years) (Transportation Research Board, 2004, 2009). The APT facilities allow a better control of test conditions, namely the structure composition (materials; layer thickness, construction process), as well as axle load (applied value and point of application), loading speed and sensor locations. Different methods are used to track the evolution of pavement performances and evaluate how different loading configurations may influence the fatigue behaviour of road pavement. Notably, the sections are often instrumented with sensing devices that allow for measurement of pavement response to loading (see section 1).
APT is a perfect tool to test methodology to access pavement ageing based on embedded sensors. We worked with a dataset provided by the full scale fatigue experiment conducted within the European project BioRepavation (J Blanc, Chailleux, et al., 2019) (J Blanc, Hornych, et al., 2019) at the University Gustave Eiffel APT facility (also called fatigue carousel). We first compared strain measurements with measurements from traditional monitoring procedures, such as visual crack inspections, Benkelman’s beam deflections, and FWD deflections. Deflections measurements at the end of the fatigue test were used to inverse calculate the layer moduli of pavement. Based on this model, numerical strains are calculated and compared with sensor measurements. The analyses conducted proved that embedded asphalt strain gauges and temperature probes can be used valuably to assess pavement fatigue. Compared to previous studies carried out in (Timm et al., 2013), the analyses carried out in this thesis apply to greater level of damage and cracking. Finally, a procedure for temperature correction of strain measured within the pavement is presented as a tool improving sensor data interpretation.
Description of the full scale experiment and monitoring methods
The fatigue carousel facility
The fatigue carousel (Figure 2.1), located in Nantes (France), is an outdoor road traffic simulator designed for the assessment of real scale pavement behaviour under accelerated heavy traffic. The carousel has a perimeter of 120 m and is equipped of a central electro-hydraulic motor unit and 4 loading arms that can reproduce various loading configurations (single wheel, dual wheels, tandem, tridem). The arms can turn at the maximum speed of 100 km/h, their loads can vary between 40 and 150 kN and they feature the possibility of move transversely. Being outdoors, the fatigue carousel is subjected to climate variations, thus the different tests are carried out during the most suitable seasons. Usually fatigue tests are carried out during months with expected temperatures below 20 °C.
Tested pavement sections
Four sections, having the structures presented in Figure 2.2-Right were tested simultaneously. The sections differ from each other by the asphalt mix used as surface layer:
A reference material that is a high modulus asphalt mix (called EME2 in France, see Section 1.4.3), with 15/25 pen grade bitumen and 20 % of reclaimed asphalt.
Three sections with the GB5® type mixes (F Olard, 2012; F Olard & Pouget, 2015a; Pouget et al., 2016) used in BioRepavation containing respectively the performance additive SYLVAROADTM RP1000 provided by Kraton Chemical (in the following called Mix1), the bio-binder Biophalt® (Pouget & Loup, 2013) from Eiffage Route (in the following called Mix2), and the bio-based additive EMS (Epoxidized Soybean Soyate) from Iowa State University (in the following called Mix3).
The 9 cm-thick bituminous surface layer lays on a 76 cm-thick subbase layer of 0/31.5 mm unbound granular material, a 1.6 m-thick stone bed 50/120 mm (subgrade), a clay soil of low baring capacity.
The reference section is 32 m long while the other three are 22 m long (Figure 2.2-Left). All the sections are 4.5 m wide. Another section of 22 m was tested but it is not included in this study (J Blanc, Chailleux, et al., 2019).
Outline of the fatigue evaluation
The fatigue evaluation was carried out between November 2017 and March 2018. One million 65 kN dual wheel loads (corresponding to the French standard axle load) were applied at radius 19 m at 72 km/h. After this first phase no significant surface damage was observed on the four sections, therefore it was decided to continue with additional 400,000 75 kN dual wheel loads at 43 km/h. The loading conditions are summarised in Table 2.1 (J Blanc, Hornych, et al., 2019). No tests were performed outside the wheel path to separate e fatigue damage (within the wheel path) from ageing (outside the wheel path). However binder evolution in the four sections was analysed using the non-destructive pavement micro-sampling technic as described in (J Blanc, Hornych, et al., 2019).
Monitoring methods
The non-destructive monitoring procedures adopted for the assessment of pavement performances over the duration of the test are described in the following sections.
Visual crack inspections
A detailed visual inspection of pavement surface was conducted for all the sections in order to record crack appearance over the duration of the fatigue test. For a better identification cracks were painted with different colours according to different inspection times. Conventionally on the APT the extent of cracking is defined as the percentage of the pavement affected by cracks. For longitudinal cracks, the “cracked length” corresponds to the measured length of the cracks. For transversal cracks, a length of 50 cm is conventionally attributed to each crack (J Blanc, Hornych, et al., 2019).
Benkelman beam deflections
Pavement deflections were measured with the Benkelman beam (Figure 2.3) every 100,000 loads starting with the initial (pre load) configuration. The Benkelman beam provides punctual measurements of the pavement maximum deflection (AFNOR, 1992). The tip of the beam is placed exactly in the middle of the two wheels of the carousel arm (Figure 2.3). It measures the rebound of the pavement surface as the arm is moved away (at about 3 km/h). On each section measurements were spaced about every 4 m, which results in a total of 4 measurements on sections Mix1, Mix2 and Mix3 and 5 on section EME2 (J Blanc, Hornych, et al., 2019).
Falling Weight Deflectometer (FWD) deflections
Pavement deflections were also measured with the Dynatest FWD apparatus in order to obtain the deflection semi-basin of the pavement (Henia & Braber, 2008). The 65 kN-load was applied through a load plate (diameter of 30 cm) and the pavement response was measured by 9 geophones as shown in Figure 2.4. The load application is such as to simulate the charging time of a truck at 70 km/h (33 ms-pulse loading corresponding to about 30 Hz). As shown in Figure 2.4 the maximum deflection is detected by the geophone D1 (under the point of application of load) and the minimum deflection is detected by the geophone D9 (the most distant from the point of application of load). It is demonstrated that the deflection measured by a geophone located at a distance D from the load is affected by pavement layers at a depths H=D and higher (P Ullidtz, 1987) (Horak, 2008) (Horak & Emery, 2009). Measurements were performed every 500,000 loads from 0 to 1 million loads, and then every 100,000 loads until the end of the fatigue tests. On each section measurements were spaced about every 5 m at 0 and 500,000 loads and every meter from 1 million to 1.4 million loads (J Blanc, Hornych, et al., 2019).
Figure 2.4 Outline of the FWD loading and measurement conditions. The load is applied with a load plate and the pavement response is measured by 9 geophones.
Pavement instrumentation
The response of the asphalt layer in terms of strain under moving loads was monitored through horizontal ASGs and temperature sensors (see Section 1.7.1).
The ASGs are located at the centre of the APT wheel path, namely at a radius of 19 m with respect to the centre of the carousel. The desire position is spray-painted on the bottom layer before placing the ASGs. Then the sensors and their cables are covered with some manually compacted asphalt mix in order to ensure protection before regular construction operation started. Each section is instrumented with four ASGs (KM-l00HAS provided by Tokyo Sokki Kenkyujo (Duong, 2017)) at the bottom of the bituminous layer, three in the longitudinal direction and one in the transverse direction. The layout of the instrumented areas in the fatigue carousel is reported in Figure 2.5. Sensors location for Mix1 and Mix3 sections is detailed in Figure 2.6. The Mix2 section is equipped with three additional sensors that are not part of this study as they are still research devices and their outputs were precisely being validated during the trial against ASGs (Figure 2.7) (Alavi et al., 2016) (Manosalvas-Paredes et al., 2020). For the same reasons the EME2 section is equipped with 13 additional sensors; among them only the sensors CTL1 and CTL2 (from CTL Group) and Dyn1 and Dyn2 (from Dynatest) are part of this study (Figure 2.8). Measurements were performed approximately every 20,000 loads (1,200 Hz acquisition frequency).
Temperature evolution within the pavement was measured with three temperature probes PT100 placed at the top, middle (- 5 cm) and bottom (- 10 cm) of the bituminous layer in Mix2 section. Temperatures were measured at 10 min intervals. The average of the three measurements is considered as the temperature of the bituminous layers.
Comparison between traditional monitoring methods and embedded sensors
In section 2.3.1 we summarize results from (J Blanc, Hornych, et al., 2019) (J Blanc, Chailleux, et al., 2019) focusing more on a descriptive analysis of experimental results. Then in sections 2.3.2 and 2.3.3, we provide new contents deepening the analyses of results from embedded sensors as well as comparison with other monitoring methods.
Visual crack inspections
By the end of the test two sections exhibited cracks, Mix 3 and EME2 (Figure 2.9). The first cracks were observed on section EME2 at about 900,000 loads. On this section cracking increased consistently and at 1.4 million loads the extent of cracking reached 28 % (see section 2.2.4.1 for the calculation of the extent of cracking). On section Mix 3 the first cracks were observed after 1 million loads and at 1.4 million loads the extent of cracking reached 10 % (J Blanc, Hornych, et al., 2019). No cracks were observed on section Mix 1 and Mix 2. In all cases cracks were transversal with respect to the traffic direction. Please note that a deeper analysis of the crack pattern is carried out in section 2.3.3.
Table of contents :
1 Context: Road pavement monitoring and nanotechnologies
1.1 Introduction
1.2 Road pavement
1.2.1 Pavement structure composition
1.2.2 Pavement structures in France
1.2.3 Mechanical behaviour of pavement structure
1.3 Bituminous materials
1.3.1 Aggregates
1.3.2 Bituminous binders
1.3.3 Overview of bituminous mixtures in the road industry
1.4 Thermo-mechanical behaviour of bituminous materials
1.4.1 Linear viscoelasticity (LVE) of bituminous materials
1.4.1.1 Creep and relaxation for a LVE material
1.4.1.2 Definition of complex modulus and Poisson ratio
1.4.1.3 Time Temperature Superposition Principle (TTSP)
1.4.2 Modelling LVE behaviour of bituminous materials
1.5 Modelling and pavement design
1.5.1 Modelling of pavement mechanical behaviour
1.5.1.1 Alizé-LCPC
1.5.1.2 Viscoroute©
1.5.2 Notions of pavement design
1.5.2.1 The ASSHTO pavement design guide
1.5.2.2 The French pavement design guide
1.6 In situ pavement monitoring with traditional technologies
1.6.1 Sensing solutions
1.6.1.1 Strain measurements
1.6.1.2 Deflection measurements
1.6.1.3 Temperature measurements
1.6.1.4 Moisture measurements
1.6.1.5 Pressure measurements
1.6.2 Instrumented sections
1.6.2.1 The Virginia Smart Road (Al-Qadi et al., 2004)
1.6.2.2 Test section in the State of Maine (Swett et al., 2008)
1.6.2.3 State of Virginia: instrumentation of Route 114 in Christiansburg (L. Wang et al., 2012)
1.6.2.4 The instrumentation of the A41N highway in France (Gaborit et al., 2014)
1.6.2.5 SMARTVIA®, the smart road (Pouteau et al., 2016)
1.6.2.6 Instrumented test section in China (Ai et al., 2017)
1.6.2.7 Monitoring of the motorway A10 in France (Juliette Blanc et al., 2017)
1.6.2.8 Continuous monitoring by using geophones and ASGs (French motorway) (Duong, Blanc, Hornych, Bouveret, et al., 2018)
1.7 Potential of carbon nanomaterials (CNMs) for road pavement monitoring
1.7.1 Carbon nanomaterials (CNMs): definition, structure and main properties
1.7.2 Applications in civil engineering
1.7.2.1 CNM-based composite materials
1.7.2.2 CNM-based strain sensors
1.8 Conclusions
2 Detection of pavement ageing based on instrumentation – Accelerated Pavement Testing
2.1 Introduction
2.2 Description of the full scale experiment and monitoring methods
2.2.1 The fatigue carousel facility
2.2.2 Tested pavement sections
2.2.3 Outline of the fatigue evaluation
2.2.4 Monitoring methods
2.2.4.1 Visual crack inspections
2.2.4.2 Benkelman beam deflections
2.2.4.3 Falling Weight Deflectometer (FWD) deflections
2.2.4.4 Pavement instrumentation
2.3 Comparison between traditional monitoring methods and instrumentation
2.3.1 Traditional monitoring methods
2.3.1.1 Visual crack inspections
2.3.1.2 Benkelman beam deflections
2.3.1.3 Falling Weight Deflectometer (FWD) deflections
2.3.2 Monitoring based on instrumentation
2.3.2.1 Survival rate to installation
2.3.2.2 Definition of signal shape parameters
2.3.2.3 Evolution of the maximum longitudinal strain
2.3.2.4 Signal shape evolution
2.3.3 Discussion
2.4 Comparison between field measured strain and inverse calculated strain
2.4.1 Assessment of pavement conditions via inverse calculation of FWD measurements: methodology
2.4.1.1 Inverse calculation from representative basin
2.4.1.2 Inverse calculation from basin in non-damaged zone
2.4.1.3 Inverse calculation from basin in damaged zone
2.4.2 Comparison between numerical and experimental strains at the bottom of the bituminous layers
2.5 Development of a procedure for temperature correction of measured strains
2.6 Conclusions
3 Inverse calculation of pavement response based on instrumentation – Corbas trial section
3.1 Introduction
3.2 Description of the trial section
3.2.1 Experimental site
3.2.2 Sensors installation
3.2.3 Data acquisition system
3.2.4 Measurement campaign
3.2.4.1 Geometry of the truck used for the measurement campaign
3.2.4.2 Definition of truck trajectories
3.2.4.3 Truck speed
3.2.5 Modelling in Viscoroute©
3.3 Feedback from the InTRACK solution
3.3.1 Resilience to construction phase
3.3.2 Quality of interface and impact on the road pavement
3.3.3 Discussion
3.4 Laboratory characterizations of the pavement structure
3.4.1 Coring campaign
3.4.2 Characterization of the bitumen of the tack coat emulsion
3.4.2.1 Complex modulus test: procedure description
3.4.2.2 Complex modulus test: analysis of results
3.4.3 Asphalt mixes
3.4.3.1 Cyclic Indirect Tensile Test (CITT): procedure and results
3.4.3.2 Modal Test (MT): procedure and results
3.4.3.3 Direct Tension-Compression Test (DTCT): procedure and results
3.4.3.4 Comparison of results and comments
3.5 Measurement campaign: results
3.6 Modelling in Viscoroute©: results
3.7 Conclusions
4 Development and deployment of an embeddable nano-carbon based strain technology
4.1 Introduction
4.2 Device fabrication process
4.2.1 Materials: Carbon-clay nanocomposite
4.2.2 Water based ink formulation
4.2.3 Device design, ink deposition and encapsulation
4.2.3.1 For layer morphology analysis
4.2.3.2 For electrical, thermal and electromechanical characterizations in the lab
4.2.3.3 For deployment in the pavement
4.2.4 Deployment in the pavement
4.3 Device characterizations
4.3.1 Lab characterization
4.3.2 Field characterization
4.3.2.1 Accelerated pavement test
4.3.2.2 Acquisition system
4.3.2.3 Data post-processing
4.4 Performances of C/Sep-MWCNT-bases strain sensors
4.4.1 Morphology and electrical properties of the C/Sep-MWCNT material
4.4.2 Device variability on E-glass
4.4.3 Strain sensors on E-glass
4.4.4 Strain sensors on PI PCBs
4.5 Performances of C/Sep-MWCNT-bases strain sensors under accelerated pavement testing
4.5.1 Resilience to the construction phase
4.5.2 Resilience to the accelerated pavement testing
4.5.3 Sensitivity to wheel passage
4.5.4 Quality of interface and impact on the road
4.6 Conclusions
5 Conclusions and Perspectives
5.1 Conclusions
5.2 Perspectives
6 Appendixes
6.1 Appendix 1
6.2 Appendix 2
6.3 Appendix 3
6.4 Appendix 4
6.5 Appendix 5
6.5.1 Bitumen of the tack coat emulsion
6.5.2 Asphalt mixes
6.6 Appendix 6
6.7 Appendix 7
6.7.1 Four-point bending test
6.7.2 Gauge factor calculation and regression coefficients
6.7.3 Definition of the densification ratio
6.7.4 Resistance of devices on E-glass as a function of the number of drops for batch B34
6.7.5 Thermal sensitivity of device NA30 on E-glass
6.7.6 Loading of the sensors with strain cycles
6.7.7 Geometry and resistance values for the devices used for pavement embedding
6.7.8 Sensitivity to wheel passage and on-site GF
6.7.9 Regression statistics and coefficients
7 Bibliography of the author
7.1 List of publications
7.1.1 Published papers
7.1.2 Papers under review
7.1.3 Papers under submission
7.2 Presentations in conferences and workshops
7.3 Awards
8 Bibliography
