It is well known that road safety issues are closely dependent on both pavement structural damages and surface unevenness, whose occurrence is often related to ineffective pavement asset management. The evaluation of road pavement operability is traditionally carried out through distress identification manuals on the basis of standardized comprehensive indexes, as a result of visual inspections or measurements, wherein the failure causes can be partially detected. In this regard, ground-penetrating radar (GPR) has proven to be over the past decades an effective and efficient technique to enable better management of pavement assets and better diagnosis of the causes of pavement failures. In this study, one of the main causes (i.e. subgrade failures) of surface damage is analyzed through finite-difference time-domain (FDTD) simulation of the GPR signal. The GprMax 2D numerical simulator for GPR is used on three different types of flexible pavement to retrieve the numerical solution of Maxwell's equations in the time domain. Results show the high potential of GPR in detecting the causes of such damage.
TopIntroduction
Over the past decades, ground-penetrating radar (GPR) has been increasingly used in road network and runway surveys both for maintenance and for construction processes (Benedetto et al., 2012a; Saarenketo and Scullion, 2000; Tosti and Benedetto, 2012). Many information on both physical and geometrical properties of pavement can be inferred through such non-destructive technique, thereby enabling more comprehensive detection of the main causes of road structural damages. GPR is widely used for the evaluation of layer thicknesses (Al-Qadi and Lahouar, 2004), voids detection underneath pavements (Lau et al., 1992; Benedetto, 2013), reinforcing bars monitoring (Huston et al., 1999), pipes location (Ayala-Cabrera et al., 2011), asphalt stripping (Scullion et al., 1994), and bridge inspections (Benedetto et al., 2012b). Further applications in the area of pavement engineering have been addressed to the evaluation of the main causes of structural damages, such as the loss of bearing ratio of load-bearing layers (Benedetto and Tosti, 2013a; Diefenderfer et al., 2005), even by other integrated non-destructive techniques, e.g. light falling weight deflectometer (Benedetto et al., 2012c; Benedetto et al., 2014). Several GPR-based methods have been developed for evaluating subsurface moisture in different soil types (Topp et al., 1980; Huisman et al., 2003; Grote et al., 2010; Benedetto, 2010; Tosti, 2013). Water infiltration in clayey soils throughout the pavement section that causes the transport of fine-grained materials towards the outer surface, can be considered as one of the main issues affecting the loss of bearing capacity (Benedetto and Pensa, 2007). In this regard, many efforts have been recently devoted to the evaluation of clay content in soils using different radar instruments and techniques (Tosti et al., 2013; Benedetto and Tosti, 2013b; Patriarca et al., 2013).
At the same time, an effective strategy in pavement rehabilitation processes is represented by GPR-based decision support systems, that are increasingly developing also for improving asset management through the widening of pavement structures inventories for pavement management systems.
In addition, many efforts have been recently focused to improve air-coupled GPR systems for providing time-efficient inspections (Tosti et al., 2014; Lambot et al., 2006) along with the possibility to collect large-scale high resolution data (Benedetto et al., 2013; Minet et al., 2012).
Another major issue concerns the possibility to help decision makers in understanding the deeper observed mismatches in case of early-stage or unrevealed failures at the pavement surface level by using GPR-based numerical modeling (Diamanti and Redman 2012). Many researches have dealt with successful modeling attempts of ground-penetrating radar (Bergmann et al., 1998, Giannopoulos, 1997; Bourgeois, 1996), and many approaches based on the finite-difference time-domain (FDTD) method have been proposed. It is well known that the use of such method is favored by the ease of implementation in a computer programme and its better scalability than other electromagnetic (EM) modeling methods (e.g. finite-element, integral) (Millard et al., 1998). On the other hand, high computational requirements and stair-like stepped interfaces, due to the discretization of the volume of the problem (Giannopoulos, 2005b; Taflove, 1995), may be encountered.