The paper gives an overview of a ground penetrating radar (GPR) experiment to survey debonding areas within pavement structure during accelerated pavement tests (APT) conducted on the university Gustave Eiffel’s fatigue carrousel. Thirteen artificial defect sections composed of three types of defects (Tack-free, Geotextile, and Sand-based) were embedded during the construction phase between the top and the base layers. The data were collected in two stages covering the entire life cycle of the pavement structure using four GPR systems: An air-coupled ultra-wideband GPR (SF-GPR), two wideband 2D ground coupled GPRs (a SIR-4000 with a 1.5 GHz antenna and a 2.6 GHz-StructureScan from GSSI manufacturer), and a wideband 3D GPR (from 3D-radar manufacturer). The first stage of the experiments took place in 2012–2013 and lasted up to 300 K loadings. During this stage, the pavement structure presented no clear degradation. The second stage of experiments was conducted in 2019 and continued until the pavement surface demonstrated a strong degradation, which was observed at 800 K loadings. At the end of the GPR experiments, several trenches were cut at various sections to get the ground truth of the pavement structure. Finally, the GPR data are processed using the conventional amplitude ratio test to study the evolution of the echoes coming from the debonded areas.

The monitoring of pavements along roads is generally based on the use of indicators directly derived from measurements. More specifically, the bearing capacity of pavements is often simply deduced from either the maximum deflection value measured or the difference between two measured values along the deflection basin. This paper proposes a methodology to define a set of orthogonal indicators adapted to the structure being evaluated. This methodology is presented for deflection measurements recorded on a flexible pavement simulated by the Burmister model and consists of searching the weighting functions to calculate different indicators as linear forms of the deflection bowl. Weighting functions are defined for each indicator in order to maximize its sensitivity to a given structural parameter without being sensitive to the other structural parameters. The paper presents the various steps involved in constructing the indicators. A numerical example of an application shows that variations of each indicator follow the Young’s modulus variations specific to this indicator. Several extensions of this method are also introduced for other mechanical models or instrumented pavements.

Pavement instrumentation sensors have become a common practice for structural monitoring. Data processing often consists of extracting different values generally representative of variations in the whole pavement, but this makes it more difficult to follow a specific characteristic of the structure. To overcome this limitation, this paper aimed to transpose an original method, labeled the Optimized Indicators Method (OIM), which consists of finding some weighting functions of a signal to evaluate some indicators linked to a physical characteristic of the structure. The main advantage of this method is that the inferred indicators are particularly sensitive to a specific characteristic of the structure without being sensitive to the others. This research work consisted of analyzing, conventionally and with the OIM, the signals of strain gauges, which were recorded during an experimental campaign carried out on a circular instrumented pavement and submitted to an accelerated fatigue testing. The OIM, calibrated through an experimental reference signal, makes it possible to evaluate independently, through weighting functions, some specific physical characteristics of the pavement. It showed a two-stepped degradation, starting with a damaging of the bituminous layer, followed by an alteration of the base layer, which could not be easily deduced from a conventional processing method.

Among the commonly used non-destructive techniques, the Ground Penetrating Radar (GPR) is one of the most widely adopted today for assessing pavement conditions in France. However, conventional radar systems and their forward processing methods have shown their limitations for the physical and geometrical characterization of very thin layers such as tack coats. However, the use of Machine Learning methods applied to GPR with an inverse approach showed that it was numerically possible to identify the tack coat characteristics despite masking effects due to low timefrequency resolution noted in the raw B-scans. Thus, we propose in this paper to apply the inverse approach based on Machine Learning, already validated in previous works on numerical data, on two experimental cases with different pavement structures. The first case corresponds to a validation on known pavement structures on the Gustave Eiffel University (Nantes, France) with its pavement fatigue carousel and the second case focuses on a new real road in Vend{é}e department (France). In both case studies, the performances of SVM/SVR methods showed the efficiency of supervised learning methods to classify and estimate the emulsion proportioning in the tack coats.