A SANISAND-STRUCTURE INTERFACE MODEL

• Published in: Iranian journal of science and technology. Transactions of civil engineering Vol. 35; no. C1; p. 15
• Main Author: Lashkari, A
• Format: Journal Article
• Language: English
• Published: Shiraz Springer Nature B.V 01.02.2011
• Subjects: Anisotropy; Interfaces; Liquefaction; Plasticity; Plastics; Sand; Sand & gravel; Shear stress; Soil mechanics; Strain; Stress state; Structural engineering; Studies
• ISSN: 2228-6160; 2364-1843

Constitutive modeling of interface behavior is a relatively a new matter. In the literature, on the basis of various assumptions, a number of interface models have been suggested. Clough and Duncan [8] introduced an interface model based on the non-linear elasticity theory. Brandt [9] proposed a rigid-plastic constitutive model for soil-concrete interface. Within the elastoplasticity framework, Ghaboussi et al. [10] suggested an interface model using a cap yield surface. Among the recent works of this type, Ghionna and Mortara [6] proposed an elastoplastic interface model using a non-associated flow rule together with two Cam Clay type constitutive surfaces. Employing the disturbed state concept, several soil-structure interface models have been introduced [7, 11, and 12]. In the frameworks of generalized plasticity and critical state soil mechanics, recently, Liu et al. [13] and Liu and Ling [14] introduced interface models which are capable of predicting the behavior under various stress levels and densities. The latter model considers the effect of particle breakage on the interface behavior. More recently, Mortara et al. [15] suggested an advanced interface model in the bounding surface plasticity context. [Manzari] and Dafalias [16] introduced a stress ratio based bounding surface plasticity model, the socalled SANISAND (Simple ANIsotropic SAND) model, for the state dependent aspects of sand behavior such as dilatancy and peak shear stress. Using the Been and Jefferies [17] state parameter, the critical state compatible model of Manzari and Dafalias [16] is capable of distinguishinig dense samples from loose ones and providing realistic predictions for stress-strain-strength behavior of sands. Since its first proposal, this family of sand models has been widely developed and applied to various subjects. Among them, Papadimitriou and Boukovalas [18] and Dafalias and Manzari [19] considered the effect of stress induced anisotropy and significantly improved the capability of the basic model under cyclic loading. Li and Dafalias [20], Dafalias et al. [21], and Loukidis and Salgado [22] considered the effect of inherent anisotropy on sand stress-strain-strength behavior. Lashkari and Latifi[23] suggested a modified SANISAND model for non-coaxial flow of sand subjected to rotation of principal stress axes. Sadrnejad [24] introduced a SANISAND multi-plane model. [B. Lashkari] [25] proposed a modified SANISAND model for sand liquefaction under rotational shear. Taiebat and Dafalias [26] modified the basic SANISAND model in order to account for particle crushing under high confining pressures. Finally, Chiu and Ng [27] proposed a SANISAND model for unsaturated sands. Similar to the other members of the SANISAND family, in addition to the yield surface, the model has three other constitutive surfaces: bounding surface, critical state surface, and dilatancy surface (Fig. 2). As implied by its name, the bounding surface defines a boundary on the evolution of stress due to kinematic hardening and as a result, corresponds to a domain for all permissible stress states. Critical state surface (defined by x = M o,, in x-o,, plane) is the terminal loci of all samples reaching critical state. Finally, the dilatancy surface is a reference state boundary which dictates the type and magnitude of the plastic volume change. An interface is in loose state and contracts due to shearing when the current stress state is inside the dilatancy surface. On the other hand, the mentioned interface dilates when its corresponding stress state is located beyond the dilatancy surface. The critical state surface size remains unchanged during shearing; however, the sizes of bounding and dilatancy surfaces are direct functions of the interface state. This issue is further discussed in the following section. A constitutive model for simulation of sand-structure interfaces behavior was introduced within the SANISAND framework. As a mutual point between all SANISAND-based models, the presented model consists of a bounding surface, a narrow yield surface, dilatancy, and critical state surfaces. Among them, the yield surface may be subjected to kinematic hardening associated with shearing. Similar to the other members of the SANISAND family, the sizes of bounding and dilatancy surfaces are direct functions of the interface state. To this aim, the concept of state parameter [17] was used as an index of the current state. Due to the particular definitions of the bounding and dilatancy surfaces, both dilatancy and plastic hardening modulus are state dependent. The mentioned definitions significantly enhance the model capability to capture the state dependent aspects of interface behavior such as phase transformation and peak shear strength. The model formulation, calibration method, and the model evaluation were presented in detail. Having a simple formulation together with a reasonable number of parameters, it was shown that the model is capable of simulating general sand-structure interface behavior. The model predictions are compared with 26 tests reported by four independent research teams. As an important advantage, the model can predict the mechanical response of interfaces in wide ranges of states (wide ranges of densities, and applied normal stresses) using a unique set of parameters. Technically, direct incorporation of Been and Jefferies [17] state parameter in the model formulation has given the mentioned ability to the model to predict the state dependent behavior of interfaces without the need to change parameters. It must be mentioned that those models which do not include the state parameter in their formulations (such as [5- 12]) may not be used for prediction of the mechanical behavior of interfaces over wide ranges of normal stresses and densities.