Risk Analysis for Dam Overtopping—Feitsui Reservoir as a Case Study

• Published in: Journal of hydraulic engineering (New York, N.Y.) Vol. 133; no. 8; pp. 955 - 963
• Main Authors: Kuo, Jan-Tai, Yen, Ben-Chie, Hsu, Yung-Chia, Lin, Huei-Fen
• Format: Journal Article
• Language: English
• Published: Reston, VA American Society of Civil Engineers 01.08.2007
• Subjects: Applied sciences; Buildings. Public works; Computation methods. Tables. Charts; Dams and subsidiary installations; Exact sciences and technology; Hydraulic constructions; Structural analysis. Stresses; TECHNICAL PAPERS; Asia; China; Taiwan; Monte Carlo method; Uncertainty; Uncertainty principles; Case reports; Spill; Risk analysis; Statistics; Recommendation; Case study; Statistical method; Dam; Water tank; Reservoirs; Risk management; Mathematical model; Performance
• ISSN: 0733-9429; 1943-7900
• DOI: 10.1061/(ASCE)0733-9429(2007)133:8(955)

Risk and uncertainty analysis by mathematical and statistical methods is often used to assess systematic risks and uncertainties. This research presents the procedure and application of risk and reliability analysis to dam overtopping. Annual maximum series of peak discharges of Feitsui Reservoir in northern Taiwan are used to analyze five extreme flood events with different frequencies. The highest water levels of the five extreme flood events were computed by using reservoir routing and considering seven factors subject to uncertainty. Afterward, the overtopping risk of Feitsui Dam was assessed by five uncertainty analysis methods: Rosenblueth’s point estimation method (RPEM), Harr’s point estimation method (HPEM), Monte Carlo simulation, Latin hypercube sampling, and the mean-value first-order second-moment (MFOSM) method. The results show that values of overtopping risk computed by different methods are similar. One may apply some approximated methods (MFOSM, HPEM and RPEM) to avoid the computational burden by applying sampling methods. Furthermore, the accuracy of results by approximated methods compared with that by sampling methods may differ from case to case. The selection and application of the uncertainty methods depend upon the information availability of the model parameters and model complexity. One may need to examine the model parameters and model complexity before determining appropriate methods to be used in a study.