|dc.description.abstract||Traditional stability analysis of concrete dams is based on the Mohr-Coulomb failure criteria with a peak friction angle to find the shear capacity. The peak friction angle is often in the range between 40 and 50 degrees, and is highly dependent on the foundation roughness. It is either assumed or set by expert judgement from literature. This method is highly inaccurate, and gives a high error margin. More advanced methods exist, but there is to the authors knowledge no analytical models that takes the deformation in the dam into account.
The aim of this thesis is to find out how well use of the finite element method can describe the shear capacity of a concrete dam, by applying the real geometry of the foundation together with a basic friction angle. The biggest challenge with failure of concrete dams, is that there are so many possible combinations of failure modes. The idea is to let the model calculate all possible failure modes, to find the weakest block in the chain. The simulations were run in the finite element software Atena.
When doing this investigation, it has been important to benchmark the models against a real case. It was therefore decided to model the shear box tests done by Dipen Bista at LTU at the same time as this thesis was made. There were done 22 shear tests of samples of concrete casted on rock, and done material tests. This provided valuable material parameters and test results for the benchmarking of the numerical models made in this thesis.
It was soon made clear that the modelling of the shear test was not straight forward, and it proved hard to get results that matched the tests. Due to lack of time and good test results, only 4 of the 22 tests were modelled. Of these, only two represent the shear tests in a good way, although the two others also give valuable insight, and have many similarities to the shear tests modelled. A parametric study shows that a probable explanation for the differences are the rotational stiffness of the test setup.
At the end, a case study is carried out on one pillar of Dam Kalhovd. The pillar is assessed with finite element analysis, and the real geometry of the foundation with a resolution of about 20 cm. For this section, this increases the factor of safety against sliding from 1.14 with the traditional approach to 1.98 with the new approach. Use of the FEM seems to be a better description of reality than the traditional method. This seems to be a promising way of assessing old dams that are deemed unsafe by new regulations.||