Artificial intelligence and visco-elastic analyses for modelling of pavement structures - Archive ouverte HAL Accéder directement au contenu
Communication Dans Un Congrès Année : 2009

Artificial intelligence and visco-elastic analyses for modelling of pavement structures

Résumé

As shown by strains measured on full scale experimental aircraft structures, traffic of slow-moving loads leads to asymmetric transverse strains that can be higher than longitudinal strains at the bottom of asphalt pavement layers. This effect is amplified when the considered structure is solicited by multiple-load axles. Conventional computations that rely on the elasticity theory cannot fully predict such behaviour. To handle this effect, a model, ViscoRoute, accounting for the thermo-viscoelastic behaviour of asphalt layers has been developed both at ENPC and LCPC (Duhamel et al., 2005, http://www.lcpc.fr/en/sources/blpc/index.php). In this, the structure is represented by a multilayered half-space and the Huet-Sayegh rheological model for asphalt materials is employed. To the contrary of Veroad software (Hopman, 1996) and similarly to the 3DMOVE software (Siddharthan et al., 1998), inertial forces are considered in the modelling which is governed by the elastodynamic equations. The developed model has been implemented in a software that computes a semi-analytical solution of the quasi-static problem derived in a coordinate system attached to the moving load. The solution is first calculated in the frequency domain. The solution in the spatial domain is then obtained by using the double Fast Fourier Transform. The Huet-Sayegh model coefficients are determined from complex modulus test obtained for asphalt mixes at different temperatures and frequencies. The free software Viscoanalyse (Chailleux et al., 2006, www.lcpc.fr) fits the Huet-Sayegh coefficients on the lab data. In the aim to offer a list of Huet-Sayegh coefficients that depends on properties of asphalt mixes, an internet viscoelastic material database, ViscoMatdata, is under project as well. The second version of the software Viscoroute, that has its kernel and its GUI programmed in C++ and Python respectively, enables to consider multiple moving loads and elliptical-shaped loads. Comparisons with other viscoelastic simulation have been done and contribute to validate ViscoRoute modelling assumptions (Chabot et al., 2009). The analysis of several configurations of moving loads confirms that it is necessary to incorporate viscoelastic effects in the modelling to well predict the pavement behaviour and to anticipate possible damages in the structure. If it is confirmed, these informations might be taken into account in the future load coefficient update of the French pavement design guide to better predict fatigue life of asphalt pavement with damage modelling. Finally, the last new version of ViscoRoute that enables to consider perfect slip interlayer relations will be soon available (Chupin et al., 2009). As expected, the introduction of a slip interface within a multilayered pavement led to higher values of deflection and designing fields commonly used in France. Due to the thermo-viscoelastic behavior of asphalt layers, these fields also increased with temperature. These effects are quite noticeable on the deflection. Modelling the response of pavement structures that incorporate slip interfaces could help the development of methods for the detection of interface flaws.
Fichier non déposé

Dates et versions

hal-00421062 , version 1 (30-09-2009)

Identifiants

  • HAL Id : hal-00421062 , version 1

Citer

Armelle Chabot. Artificial intelligence and visco-elastic analyses for modelling of pavement structures. Artificial Intelligence and Visco-Elastic Analyses for Modelling of Pavement Structures, TU Delft, WOENSDAG, Apr 2009, Delft, Netherlands. ⟨hal-00421062⟩
169 Consultations
0 Téléchargements

Partager

Gmail Facebook X LinkedIn More