Risk Assessment for a Large Cascade Dam under Flood Hazards

In recent years, cascade reservoirs have been planned and constructed in many river basins of China. These large dams of cascade reservoirs are frequently located in mountain areas affected by extreme climates, and facing a huge challenge of increasing flood hazards caused by extreme climates, especially intensive rainfall events. If a dam fails during a flood disaster, downstream communities are expected to serious impacts. Therefore, risk assessment of dams threatened by flood hazards is urgently required and essential in basin risk management. However, in the light of growing climate change concerns and uncertainties in the complex risk system, evaluating the dam risk under floods is arduous. In the past, few studies considered simultaneously all of the following factors: the major flood hazard sources, the cascade dam effects, and the probabilistic relations between the influencing indicators involved in risk analysis. In addition, previous studies have illustrated BN’s effectiveness and superiority to traditional risk analysis methods, and rare application to risk analysis for dams exposed to flood hazards. Therefore, in the study, a risk assessment framework for dam overtopping under floods is proposed by combing the Bayesian network (BN) theory, risk analysis theory, stochastic hydrology and Monte Carlo (MC) simulation. Based on a cascade dam affected by flood hazards, located in the Dadu River Basin, Southwestern China, and the collected and processed data of climate, hydrology, engineering and topography, a detailed application of the approach on risk evaluation is provided. Wherein, an instructive procedure for determining the prior and conditional probabilities associated with these risk sources and causal relationships is provided, which quantifies the occurrence likelihood for flood risk event and the strength of the flood effect on dam overtopping risk. The results reveal that the selected dam has a very low annual dam overtopping probability over its life cycle and satisfies the corresponding risk control standard. Compared with conventional approaches, the developed framework addresses uncertainties in risk analysis, provides an effective and feasible methodology of identifying, analyzing and estimating dam overtopping risk affected by flood hazards for making adaptation strategies and implementing mitigation measures in the risk management procedure, and can be developed, modified and extended to a more complex system to include various dam failure modes and more hazard sources. It contributes to the improvement of current knowledge and best practices on such an important dam safety concern, and can also be extended to other likelihood-related hazard studies.