Numerical simulation of triaxial testing of rockfill for the Romaine-2 rockfill dam and dikes

Alain Plaisant

In the proceedings of: GeoVancouver 2016: 69th Canadian Geotechnical Conference

Session: FUNDAMENTALS - X Physical & Numerical Modelling

ABSTRACT: Triaxial testing of rockfill specimens (by others) from the Romaine-2 Hydroelectric project was simulated analytically and numerically for verification of the numerical models of constitutive behavior, software and modeling procedures that could be used to simulate the behavior of rockfill dams during construction and reservoir filling. The simulations were based on the Duncan & Chang model of rockfill behavior that considers a hyperbolic stress-strain response, a stress-dependant internal friction angle, and an appreciable loss of strength and rigidity upon wetting. The volumetric behavior as a function of confining stress and axial strain was also considered. Analytically, the testing was simulated by direct application of the equations of Duncan & Chang. The numerical simulations were conducted using PLAXIS 2D with the built-in Hardening Soil model (hyperbolic model) and FLAC 2D with a customized (FISH) routine coupled to the built-in Mohr-Coulomb model that simulated the Duncan and Chang model. The analytical and numerical simulations correspond quite well to the results of laboratory testing.

RÉSUMÉ: Des essais triaxiaux (réalisés par d'autres) enrochement utilisé lors du projet hydroélectrique de la Romaine-2 ont été simulés analytiquement et numériquement aux fins modèles numériques de comportement constitutif, des logiciels et des procédures de modélisation pouvant être utilisés pour simuler le comportement des barrages en enrochement pendant leur construction et le remplissage du réservoir. Les simulations réalisées se sont basées sur le modèle de comportement Duncan et Chang enrochement, qui considère une réponse hyperbolique en contrainte-déformation, un angle de frottement interne dépendant du nivune perte appréciable de résistance et de rigidité lors du mouillage. Le comportement volumique en fonction de la contrainte de confinement et de la déformation axiale a également été considéré. Les essais ont été simulés analytiquement par l'application directe des équations de Duncan et Chang. Les simulations numériques ont été réalisées à l'aide de PLAXIS 2D avec le modèle prédéfini « Hardening Soil » (modèle hyperbolique) et FLAC 2D avec modèle personnalisé (FISH) couplé au modèle Mohr-Coulomb prédéfini afin de reproduire le modèle Duncan et Chang. Une bonne concordance des simulations numériques et analytiques avec les résultats des tests de laboratoire a été obtenue. 1 INTRODUCTION Rockfill has been used as a construction material for thousands of years as it is relatively easy to produce, transport and place. Additionally, under most engineering applications of low to moderate stress, rockfill is relatively strong and rigid. Typically in engineering practice, rockfill structures such as dams and embankments are designed and evaluated assuming that the rockfill has a linear elastic-plastic stress-strain response, a constant angle of internal friction and is not affected by wetting. In the 1970s researchers including Leps (1970), Duncan & Chang (1970) and Nobari & Duncan (1972) conducted extensive studies on rockfill and found that it is characterized by a hyperbolic stress-strain response, a stress-dependant angle of internal friction, and an appreciable loss of strength and rigidity upon wetting. The differences between assumptions made in typical engineering practice and the actual behavior of rockfill have resulted in a number of incidents when large rockfill embankments and dams have undergone excessive differential deformations due to wetting (Kulhawy et al., 1969; Justo, 1991). The Romaine-2 project consists of a rockfill dam and five rockfill dikes varying from 30 to 110 m in height with asphalt concrete cores less than 1-m-thick. During the design process a comprehensive program of laboratory testing, test fills and numerical analysis was realized to guide the design, aiming to assure good behavior of these dam and dikes during construction, reservoir filling and operation (Longtin et al., 2012; Longtin et al., 2013). Numerical modelling can be a very important element in the geotechnical design process, particularly for large earthfill and rockfill structures where deformations induced by relatively high shear stresses or wetting can produce excessive local and global deformations. Modelling can provide engineers with insights as to the behavior of a dam during construction, reservoir filling and operation with respect to the internal zoning, external geometry, materials properties, degree of compaction, etc. However, to be an effective design tool, modelling must be capable of capturing the behaviors of interest in a reasonable manner and any important discrepancies between the actual material behavior and the model must be quantified and considered. This is done by calibration and validation of the modelling system which consists of the numerical model of constitutive behavior, the software program used

Please include this code when submitting a data update: GEO2016_3646

Access this article:
Canadian Geotechnical Society members can access to this article, along with all other Canadian Geotechnical Conference proceedings, in the Member Area. Conference proceedings are also available in many libraries.

Cite this article: