Probabilistic non-linear finite element analysis of concrete buttress dams

Assessment of concrete buttress dams today consists of using analytical methods to investigate the stability in different failure modes. However, they simplify the problem at hand and neglect certain important features linked to geometry and thus not truthfully appraise the structures real stability. This could be economically costly in assessments of both existing dam structures as this would eventually lead to unwarranted rehabilitation. The same can be said for dams under design. Kalhovd dam is a concrete buttress dam that has recently undergone assessment where certain sections proved unstable in either failure mode. This study focused specifically on a buttress, numbered 49, of said dam which was deemed in the assessment to be unstable in the overturning and sliding failure mode. The methodology used was to assess and compare stability for different methods of calculation, meaning analytical versus numerical analysis. The numerical analysis was performed with two-dimensional finite element analyses in a commercial FEM-software called ATENA which focuses on reinforced concrete structures. The numerical models made, were then probabilistically analysed by randomizing various material parameters to see their effect on stability. Some FEA models discarded the conventional way of modelling loads, as required by standards, to more realistically portray load actions on dams based on historical measurement. Results from this study showed dissimilarity of stability for buttress 49 depending on which method of analysis was performed. Analytical methods proved to yield the most conservative results and concluded the structure unsafe, while including an accurate representation of the geometry in FEA models improved the stability such that it can be considered stable against load actions stated in guidelines. Furthermore, material parameters randomized in the probabilistic analysis substantiated that various concrete variables had relatively little effect on structural overall strength in the most common failure modes