Determination of the critical state of tailings is essential to the stability analysis of tailings dams, especially to assess the susceptibility to flow liquefaction failures. In contrast to the peak resistance, which can be determined in triaxial tests at relatively small strains, the critical state is often reached at higher strains. Therefore, determining the critical state in triaxial extension tests is more difficult than in compression tests because the concentration of stresses and strains in extension tests is more severe and usually inevitable, resulting in necking that may greatly affect the computed deviatoric stress and void ratio (if the test is drained). This paper presents a simple and cost-effective method of necking correction using photography. The method was used in the determination of the critical state in drained triaxial extension tests. A sample of silty sand tailings obtained from the reservoir of the Fundão dam two years prior to its collapse was used. The critical state friction angle was approximately the same in compression and extension, showing that the proposed method is promising as many authors consider that the critical state friction angle is the same in these two conditions.

Sponsored by Geotechnical Engineering

Tailings dam accidents emphasize the importance of an adequate understanding of the strength parameters of tailings to improve the efficiency and effectiveness of the design, construction, and operation of such structures. Usually, the tailings strength is addressed in a deterministic manner. However, a statistical approach would better represent their behavior due to its inherent heterogeneity. The literature about tailings strength distribution is relatively rare or superficial, which impairs the probabilistic analyses which are essential for risk management. Therefore, this article focuses on the probability density function (PDF) of the effective friction angle (ϕ′) of iron ore tailings from the reservoir of Germano dam, Mariana, Brazil, based on data from publicly available CPTu tests. The influence of the relative density (D_r), and the presence of plastic layers amidst the sand tailings on the strength of the sand are also discussed herein. Several correlations were employed to estimate ϕ’ and D_r. According to the results, the presence of plastic layers influences the estimated properties, and the relative density has a log-normal distribution. The effective friction angle, on the other hand, presents a normal distribution.

Mining tailings may be divided in non-plastic (sand, silty sand, or sandy silt tailings) and plastic (fine tailings, usually mixtures of silt and clay-sized particles). The understanding of the behavior of these materials is a great geotechnical challenge because of their wide variations in mineralogy, physical–chemical, and geotechnical characteristics. Sand tailings have been widely studied in the last decades because they are often susceptible to liquefaction, and they usually constitute the beach of the tailings dam which is the most important zone for stability analysis in upstream dams. However, the study of the characteristics and the strength of the plastic tailings is also important since layers of plastic tailings are sometimes found in the midst of the sand tailings. The aim of this article is to study the normalized undrained strength ratio (s_u/σ′_v) of plastic tailings of the Germano dam in Mariana, Brazil, and to present a criterion developed for their identification. The method consisted of analyzing and scrutinizing Piezocone (CPTu), Field Vane (FV) tests, and the water levels. An extensive review of the iron ore plastic tailings properties was also performed. A total of approximately 900 occurrences of plastic layers were analyzed. The results included the discovery of perched water tables and significant vertical downward flow in some locations. The conclusion is that s_u/σ′_v histograms are well fitted by lognormal distributions and the layer thickness influences the undrained strength. Moreover, the strengths are relatively low (for example, s_u/σ′_v = 0.11 on average for plastic layers with thickness of 1.0 m or more).

The direct simple shear (DSS) test is often employed to assess the undrained strength of fine-grained soils or the liquefaction behavior of coarse soils. Effective strength parameters are not derived from it because the state of stress within the specimen is not considered sufficiently well defined or uniform enough to allow a rigorous interpretation of the results. In the DSS, the failure plane is not imposed, and the principal stresses are not measured, unlike in the direct shear or triaxial tests, respectively. In the past, the DSS was often dismissed because comparisons with other types of tests indicated that it produced lower effective strength parameters. However, this paper presents results of DSS with stacked rings and direct shear tests that are in good agreement with each other. A finite element analysis and Mohr’s circle representations are used to assess the internal stresses and strains in the DSS specimen and to locate the failure plane. Based on these results, a reinterpretation of other works is conducted. It is shown that the stress distribution becomes approximately uniform after the early phases of the shearing, and the inclination of the failure plane can be predicted. It is also discussed how the use of low height-to-diameter ratios may influence the results. Thus, the DSS with stacked rings may be considered suitable to determine effective strength and deformability parameters, provided an appropriate height-to-diameter ratio is employed, and the results can be expected to agree with those from other types of tests.