The objective of this communication is to highlight the limitations in the normal criteria for dimensioning dams with respect to floods and to measure to what extent other criteria might be useful. This seems to be absolutely coherent with recent regulatory developments (stricter requirements regarding directives in the case of floods and especially hazard studies). It is within this framework that we have used recent feedback to produce an analysis of cases of recorded ruptures and simple quantitative analyses. This has led to some practical propositions:

• To better take into account the uncertainty in hydrological studies, with an appropriate mathematical analysis or even more simply by “testing” the occurrence of higher floods (for example twice as high as the project flood)
• Dimension the crest and the freeboard depending on the flood risk and accept, that for certain dams, highest water levels (PHE) may be exceeded
• More generally, use the probability of dam break in a flood (or moreover, the probability of it leading to human victims) as a dimensioning tool; specific tools are described to this effect, which give results coherent with published statistics and are sometimes more convincing than traditional criteria
• Finally, take into consideration the inherently uncertain nature of dams as flood limiters by strengthening the dimensioning criteria for these structures and organizing well-controlled impoundments