In the latest years there has been a trend towards an implementation of structural reliability analysis, SRA, in the stability assessment of concrete dams. In these analyses, it is in general assumed that the basic variables are independent of each other. However, this may be incorrect. For example, when the shear strength of the bonded concrete-rock interface is approximated with the Mohr-Coulomb failure criterion, the linear approximation of the curved failure envelope implies that a negative correlation between cohesion and friction could be expected. The aim of this paper is to study how correlation between friction angle and cohesion for a bonded interface may affect the sliding stability of concrete gravity dams founded on rock. For this purpose, experimental data from uniaxial and triaxial compressive tests performed on intact rock were used in order to study the existence and magnitude of a possible negative correlation between friction angle and cohesion. After that, the reliability index for a concrete gravity dam was calculated using negatively correlated and uncorrelated friction angle and cohesion. The results showed that the safety index increased significantly as the correlation coefficient decreased. Therefore, in order to calculate the reliability of dams correctly in the future, it is important to study and determine the magnitude of this correlation.

The limit equilibrium method (LEM) is widely used for sliding stability evaluation of concrete gravity dams. Failure is then commonly assumed to occur along the entire sliding surface simultaneously. However, the brittle behaviour of bonded concrete-rock contacts, in combination with the varying stress over the interface, implies that the failure of bonded dam-foundation interfaces occurs progressively. In addition, the spatial variation in cohesion may introduce weak spots where failure can be initiated. Nonetheless, the combined effect of brittle failure and spatial variation in cohesion on the overall shear strength of the interface has not been studied previously. In this paper, numerical analyses are used to investigate the effect of brittle failure in combination with spatial variation in cohesion that is taken into account by random fields with different correlation lengths. The study concludes that a possible existence of weak spots along the interface has to be considered since it significantly reduces the overall shear strength of the interface, and implications for doing so are discussed.

The shear strength of concrete-rock interfaces, used in reliability-based analyses of sliding stability of concrete dams, is commonly quantified using a simplified version of the Mohr-Coulomb shear strength criterion. This criterion is based on the assumption of a mean value driven process occurring according to the requirements imposed by perfectly-plastic failure theory. However, tests on cores with bonded interfaces exhibit semi-brittle post-peak behavior, which in combination with a possible spatial variation in cohesion implies that a progressive failure mechanism is a more correct description of the interface behavior. This has to be considered in sliding stability analyses in order to avoid overestimation of dam safety. Uncertainties between model and real behavior like the one described above can be taken into account by introducing a random variable representative of the specific uncertainty into the limit state function. A drawback with this technique is that the real shear strength of the interface is difficult to define since physical observations are not easy to acquire. Krounis and Johansson (2014), therefore, used a technique where numerical analyses were used in order to estimate the real behavior. The technique requires several numerical calculations and is extremely time-consuming. Another, less time-consuming, technique was used by Westberg Wilde and Johansson (2013). In their work, the progressive failure of the bonded interface was taken into account by simulating new distribution parameters for cohesion. This was done by means of an analytical procedure where the cohesive strength of the interface was considered to behave as a brittle parallel system. The stress state of the interface, however, was not considered, which may have significantly affected the results of the analyses. In this paper, a comparison between these two techniques for a hypothetical dam monolith is made in order to study the magnitude of the possible discrepancy between the two techniques. The benefits and limitations of each method are also discussed based on the results of the analyses.

For concrete dams, cohesive strength often constitutes a significant share of the overall shear strength of partially bonded concrete-rock interfaces. However, cohesive strength is also associated with great uncertainties that may have a significant impact on the assessed stability of the analyzed structure. In this paper, the merits of including cohesion are evaluated using a probability-based approach to analyze the sliding stability of an existing concrete gravity dam. The shear strength properties of the interface are inferred from a limited number of site-specific tests and previous knowledge from similar structures using Bayesian updating. The study shows that the potential gain from cohesive strength is strongly related to the involved uncertainties and identifies the bonding percentage and basic friction angle as the most influential parameters. The importance of testing, both with regard to the specific project and for future projects, is also highlighted.

The shear strength of the concrete–rock interface has a substantial influence on the sliding stability of concrete gravity dams founded on rock. While several studies have been done on concrete–rock contacts, there remains uncertainty regarding the peak shear strength of partially bonded interfaces. There exists, in particular, an uncertainty regarding the contribution from surface roughness of the unbonded parts to the peak shear strength of the interface due to the dependency of mobilized strength on shear displacement. In this study, a series of 24 direct shear tests are performed under CNL conditions on concrete–rock samples with different bonding conditions. Tests on samples with fully bonded and unbonded interfaces are conducted to study the strain compatibility of the different contacts, while the results of samples with partially bonded interfaces are evaluated in the context of linking the joint roughness of the unbonded parts to the peak shear strength of the interface. The results indicate that a significant part of the surface roughness of the unbonded parts is mobilized prior to degradation of bond strength, in particular for interfaces with low bonding percentages. It is recommended that further research should be conducted to understand how the contribution from roughness change with an increase in scale and degree of matedness.