AN INTEGRATED PROBABILISTIC APPROACH FOR DETERMINING THE EFFECTS OF EXTREME HYDROLOGICAL EVENTS ON A FLOOD EVACUATION SYSTEM

The well-known PMP-PMF method has been developed for estimating the most extreme hydrological events likely to occur. It is however subject to many uncertainties, which require as many simplifying assumptions. The imprecision and biases, and consequently the inaccuracy of the estimate, are summing up all along the hydrological chain. The resulting hydrograph however is considered representative of the most critical situation for the dam and its flood evacuation system. To alleviate these drawbacks and damp the consequences of oversimplifications, an integrated probabilistic approach has been developed. All the uncertainties (precipitation, snow melt, infiltration, etc.) are considered as random events leading to a – random – response of the evacuation system. A series of simulations is carried out, which include not only the generation of possibly critical floods, but route these events through the reservoir and its evacuation system. All the independent variables are randomly generated within their respective estimated variation range. The statistical analysis of the response distribution focuses on an indicator of interest (max. water level, max. outflow discharge, e.g.). This allows to determine – among others – the extent of its probable variation range. On the basis of this response, the design engineer can decide whether the occurrence probability of the indicator is compatible with the safety rules in force. The method has been successfully applied to check the preliminary design of the spillway at Ostour dam (Iran). This case presents strong uncertainties shrouding the PMF determination. The hydrologists relied on the integrated approach through a Monte-Carlo analysis to statistically estimate the Probable Maximum Response of the reservoir for a given flood evacuation system.