Third International Workshop on Modern Trends in Geomechanics IW-MTG3, 4-5 September 2012 64 New design approach for tailing dams based on recent study on instability of sand J. Chu 1 and D. Wanatowski 2 1 Department Civil, Construction & Environmental Engineering, Iowa State University, USA 2 Nottingham Centre for Geomechanics, Faculty of Engineering, University of Nottingham, UK E-mail: jchu@iastate.edu Summary: Static liquefaction or flowslide is considered as one of the most common failure mechanisms for granular slopes or tailings dams. One of design approaches adopted is to use the residual strength or the so-called post-liquefaction undrained shear strength. However, there are a number of problems associated with this approach. One of them is that the so-called post-liquefaction strength cannot be determined properly experimentally. The assumption of an undrained condition is also questionable for sand or tailings with relatively high permeability under static loading conditions. Based on the findings of a recent study on instability of sand, a new design approach to use the stress ratio of instability line or the peak strength ratio are suggested in the paper. 1. INTRODUCTION Slope failure or landslide is still one of the common geotechnical hazards. This includes failures of tailing dams of mine waste, mineral sands, or municipal solid waste. As explained by Davies et al. (2002), the failure of loose granular soil slopes or cohesionless tailing dams is often considered to be triggered by static liquefaction occurring under undrained conditions, as shown in Fig. 1. Two approaches, the effective stress analysis (ESA) and the undrained strength analysis (USA), have been suggested (Martin and McRoberts, 1998). (a) (b) Figure 1 Response of loose, saturated cohesionless tailings under monotonic and cyclic loading (after Davis et al. 2002) According to Martin and McRoberts (1998), in the ESA method, effective stresses during shear are assumed unchanged from those that existed immediately prior to the onset of shear. In other words, failure is calculated as the failure shear stress corresponds to the in- situ effective stresses using the effective failure envelope, at point F, as marked in Fig. 1(a). This method may be applicable to dense, dilative soil where the excess pore pressure generated during shear is very small or negative. However, for loose, contractive soil where positive excess pore pressure is generated, this method is unconservative, as failure occurs at point P, not at point F, as shown in Fig. 1(a). In the USA method, the undrained shear strength is defined as the residual strength, or the steady state strength, or the post- liquefaction strength, S us , as shown in Fig. 1a. This approach is also called the steady state method (Poulos et al. 1985). As elaborated by Martin and McRoberts (1998), for contractive materials, design analyses must include both undrained strength analysis (USA) and effective stress analysis (ESA), with design controlled by the analysis type giving the lowest factor of safety. For dilative or fully drained materials, only ESA is required. However, the use of post-liquefaction strength as the residual strength is problematic. In this paper, the deficiencies related to the use of post-liquefaction strength are discussed. Alternative approaches using the stress ratio of instability line that defines the instability condition for loose sand under both drained and undrained conditions or the peak strength ratio are suggested. 2. PROBLEMS WITH THE USE OF POST-LIQUEFACTION STRENGTH There are several problems associated with the use of post- liquefaction strength. Firstly, granular soil or tailings have to be very loose to exhibit a contractive behaviour as shown in Fig. 1. Many granular soils at its in-situ density do not liquefy under static, undrained conditions as shown by Vaid and Thomas (1995) and Chu et al. (2003b). Secondly, the value S us cannot be determined properly. The S us should be measured from a test where the soil liquefies. However, once a soil liquefies, the specimen collapses suddenly from the point where liquefaction is initiated, e.g., from the peak point P in Fig. 1(a). Thus the stresses and strains in the post-peak region cannot be measured properly. The so-called post-liquefaction strength is, in fact, measured as the post-peak strain softening behaviour, where the specimen does not collapse, but the shear stress is reduced gradually. Although the strength measured at the end of the test is taken as the residual or the ‘post-liquefaction strength’, liquefaction does not occur in this test. It has been demonstrated experimentally by Chu and Leong (2001) and Wanatowski and Chu (2007, 2012) that the post-peak behaviour during strain softening and instability (or liquefaction) is quite different. Therefore, the S us obtained from tests with strain softening may not be relevant to soil or tailings that liquefy. After liquefaction, a soil has completely changed in terms of void ratio, stress history, soil structure and fabrics. Thus the post- liquefaction strength can only be measured by testing liquefied soil, not on soil that has not liquefied before test. The differences between the post-peak liquefaction and strain softening behaviour are further explained in Fig. 2 where a comparison of two undrained plane-strain tests on two nearly identical loose sand specimens, but one conducted under load- controlled and another deformation-controlled loading mode is shown.