Correlation Analysis for the Causes of Deteriorations of Earth Dams in an Inverse Bayesian Belief Network Model

Worldwide, there are over 80,000 high dams(higher than 10 m) in existence, and 50% of them have been more than 50 years in service since construction, and 10% are aged over 80 years. Although some researches in the dam arena have focused on safety issues, most of them only contemplated the hydraulic functions of these structures. Unfortunately, due to the accelerated global warming climate condition, the risk of catastrophic structural failure increases as time goes by, mainly because extraordinary rainfalls can surprise dam owners and operators anytime. Hence, risk assessment and management for old dams are strongly needed, which could provide information for maintenance planning. In general, the failure of a dam leads to significant economic loss and usually to catastrophic impacts. So, this paper aims to comprehensively examine the variables that influence earth dam breaks and describe their interactions. Consequently, a probabilistic model has been developed that enables to assess risks and to prioritize the distribution of resources with the choice of Bayesian belief networks(BBN) as a tool for assessing risks. The model may be of interest for dam owners, practitioners and academics on their efforts to manage small dams risks.