Dam Risk Assessment Using the Event Tree Analysis and Bayesian Networks

This paper shows the application of a methodology according to the Norm "Dams Safety Operation" NMX-AA-175-SCFI-2015 to identify, estimate and evaluate the risk of earthen dam failure due by the overtopping, before and after the mitigation measures. Two methods were used, which are: 1) the Event Tree analysis (ETA); and 2) the Bayesian Network (BN) that was programmed in a model called @Dam_Risk. A homogeneous earth dam was analyzed that failed in its first fill (Feb 2010) due to a poor spillway design which had a failure probability of 7.23×10^-2 for ETA and 1.16×10^-2 for BN, additionally, it placed 7.56×10^-2 lives/year and $36,638.00 USD in material damage in the event of a failure. A modification alternative was proposed to mitigate the risk and it was newly calculated until reducing the failure probability to 5.67×10^-5 and 1.20×10^-5, respectively. The event trees presented serve to make a preliminary classification along with the help of experts` judgment, afterwards, however, the @Dam_Risk model is recommended to carry out a formal risk analysis. The proposed methodology will permit the use of the Bayesian inference to obtain a fragility curve given the evidence of conditional probabilities for real cases of failure, so that the fragility curves according to type of dam can be categorized and used for creating a list of dams with like characteristics. The proposed methodology will allow decision-makers to equitably compare the risk between dams, prioritize specialized studies, make investments or emergency action plans and propose alternatives until arriving at a risk that society is willing to tolerate or to put a dam out of service in case of persistent, unacceptable risks. It is important that the managers and users of the dams be aware of the consequences in case of failure and the responsibilities they have according to the law, so that they are involved in the risk assessment process and if necessary in the discussions for solution proposals in a formal study, otherwise they will suffer the economic or social brunt for the reconstruction of the dam, due to flood damage. This methodology opens up the possibility of developing indices that allow the evaluation of the benefit of mitigation measures for risk control, the cost that these imply along with the prioritization of financial resources, the security of the dam and the population.