Reliability-based Assessment of Concrete Dam Stability

Sammanfattning: Risk management is increasingly used in dam safety and includes risk analysis, risk evaluation and risk reduction. Structural Reliability Analysis (SRA) is a probabilistic methodology that may be used in the risk assessment process. SRA has been frequently used for calibration of partial factors in limit state design codes for structures (not dams). In a reliability analysis a mathematical description of the failure mode, a limit state function, is defined. All parameters describing the limit state function should be random variables and are described by stochastic distributions (or, where appropriate, a deterministic value). The safety index (or probability of failure) may be determined by e.g. First Order Reliability Method and the result is compared to a target safety index to determine if the structure is safe enough. Several difficulties exists in the use of SRA for concrete dams, mainly due to the fact that only a few examples of such analysis for dams exist. One difficulty is how to define the failure modes. In this thesis a complete system of failure modes is identified, where failure is considered as a series system of “failure in the concrete part”, “failure in the concrete-rock interface” and “failure in the rock mass”. Failure in the concrete-rock interface may occur due to sliding or overturning. Sliding is the joint occurrence of sliding with a partially bonded contact (fails at very small displacement) and sliding with broken contact (fails at larger displacement) and both have to occur for sliding to occur, hence they are treated as a parallel system. Adjusted overturning is a combination of overturning and crushing of the concrete or crushing of the rock. A substantial part of the work has been to define the necessary input data. - Cohesion in the interface is very important. Due to the expected brittle failure in a partly intact interface, treatment of the shear resistance as a brittle parallel system is proposed. - Description of the headwater results in a series system; either failure occurs for water levels at retention water level (rwl) or for water levels above rwl, the latter described by an exponential distribution. - Uplift is one of the most important loads. A geostatistical simulation procedure is presented, where the hydraulic conductivity field of the foundation is described by a variogram and uplift is simulated by a FE-analysis. This methodology is demonstrated to be very useful and gives estimates of the statistical distribution of uplift. Three papers on this subject are included; the first is a description of the methodology, the second presents a sensitivity analysis performed for a large number of different combinations of input data and the last is an application to a Brazilian dam, where water pressure tests and monitoring results are available. In two papers SRA is applied to concrete dams and the system reliability is determined. In the first paper a spillway section where information of e.g. cohesion, friction angles etc. were available is analysed. In the last paper an idealized dam and a power intake structure are analysed. The conclusions are that SRA may be used for assessment of concrete dam stability and that it is well fitted for the dam safety risk management process. Every dam is a unique prototype and SRA enables specific behaviour and properties of a certain structure to be taken to consideration. The system reliability analysis is a very valuable tool in understanding the relationship between failure modes and enables the safety for the whole structure to be determined. In a reliability analysis the most important parameters may be identified and thus safety measures can be focused where it gives the best possible output. A general safety consideration is that development of the safety concept for concrete dams, from deterministic to probabilistic or semi-probabilistic, will give a known and more uniform level of safety.