Safety assessment of existing dams becomes more and more important with their increasing age. In addition, new regulations, due to climate changes and changes in knowledge, demand a re-evaluation of the existing dams’ safety. One of the failure modes considered in safety assessments of concrete gravity dams is sliding at the dam-foundation contact. Sliding failure is assumed to occur when the applied load exceeds the shear strength of the interface. Thus, the shear strength of the concrete-rock interface will ultimately determine if sliding of the dam will occur or not and it is, therefore, of utmost importance to be able to quantify it.
The shear strength of concrete-rock interfaces is in general defined based on the Mohr-Coulomb shear strength criterion, which is governed by the cohesion, friction angle and normal stress along the interface. This simplified model of the shear strength is subject to several uncertainties, due for example to the natural variability of the involved parameters, statistical errors etc. In addition, analyses are commonly performed using the averaged values of cohesion, friction angle, normal and shear stresses, based on the assumption of ductile failure. This may be incorrect for bonded or partly bonded interface since test results show that bonded concrete-rock cores exhibit brittle behaviour.
The uncertainties related to bonded or partly bonded interfaces are taken into account in the Swedish guidelines, RIDAS, for by treating all concrete-rock interfaces as unbonded, i.e. the effect of cohesion is not included when evaluating the shear strength of the interface. This is a conservative method, but it may lead to expensive and unnecessary strengthening of existing dams. Other deterministic guidelines/regulations, e.g. FERC, allow the use of cohesion but apply higher target safety factors when both the cohesive and the frictional strength are taken into account. To evaluate the adequacy of using cohesion in sliding stability analyses of concrete dams, the effect of the uncertainties on the calculated sliding stability of the dam has to be assessed.
This thesis highlights several uncertainties related to bonded concrete-rock interfaces. However, the thesis focuses mainly on increasing the knowledge regarding the model uncertainty due to the brittle failure mechanism in combination with a possible spatial variability of cohesion. The magnitude of the model uncertainty is studied using numerical analyses. Its influence on the assessed behaviour of a hypothetical dam monolith is then evaluated using probability based methods. A conclusion drawn from this particular case, but which is likely to be generalized to other dam-foundation systems, is that the ductile sliding failure is too coarse an approximation of the failure behaviour of dams with bonded interfaces and could lead to an overestimation of dam safety. In addition, the potential spatial variability of cohesion along the interface further diminishes the validity of the ductile failure model.
Areas that require further research in order to take into account the most significant uncertainties related to bonded interfaces include the definition of the ratio of bonded area to total area and its influence on sliding stability. The statistical uncertainty due to the limited number of tests also needs to be studied.
Stockohlm: KTH Royal Institute of Technology, 2013. , xvi, 34 p.