Résumé Cet exemple du barrage d’Avène dans le département de l’Hérault (France) concerne la caractérisation de l’aléa sismique pour un ouvrage de type voûte mince de 61 m de hauteur achevé en 1961 et reposant sur des fondations en quartzites très rigides en bancs verticaux. Après définition des zones sismo-tectoniques, on s’appuie sur celles du séisme maximal historiquement vraisemblable (SMHV), puis du séisme majoré de sécurité (SMS). Le calcul des spectres de réponse a été fait à partir de 8 lois d’atténuation spectrales moyennées à partir de leur importance (la loi d’Ambraseys et al, 1996, est prépondérante). Cette étude réalisée en 1999 se base sur l’utilisation d’un spectre de dimensionnement pour contrôler sommairement le barrage et doit être considérée comme telle. Tout contrôle détaillé nécessiterait de procéder à des calculs à partir de signaux temporels et prenant en compte la durée de l’événement.

Abstract The Avène Dam in the department of France Hérault illustrates how seismic hazards were characterized for a thin arch dam, 61m high, completed in 1961, and built on a very rigid quartzitic sandstone foundation forming vertical layers. After defining the seismotectonic zoning, the assessment is based on the Maximum Historically Probable Earthquake (SMHV), then on the Safe Maximum Earthquake (SMS). Traditional methods have been updated thanks to state of the art systems. Using seismotectonic information and available historical and instrumental earthquake data, earthquake magnitude and depth values can be allocated to the retained SMHV. The SMS is then obtained by adding half a degree to the SMHV, which can be considered to represent an additional degree in intensity. Earth movements on the site studied are then directly characterized in response spectrums based on spectral attenuation laws and the retained SMS values. The local SMS is often decisive at high frequencies, the distant earthquake at low frequencies, if either one or the other is not predominant throughout the whole frequency range of interest. The earthquake hazard for Avène Dam is therefore as follows. For the nearby earthquake (SMS), the reference is the event of 16 May 1939, with an intensity of VI at the epicentre, a depth of 15 km, and a magnitude of 4.9. The SMS of the same earthquake on the Avène Dam site (1 km away) is considered to be 5.4. The distant earthquake taken into account (SMS intensity = VII – VIII) must be one that would affect the furthest part of the zone nearest to Avène, i.e. approximately 50 km away. Considering the (extremely blurred) relation between epicentral intensity $ I_{0 } $and magnitude M, (h = depth in km), M=0.55 $ I_{0} $ + log h + 0.35, the approximate magnitude of this earthquake is 5.6. The response spectrums were calculated using 8 spectral attenuation laws. These laws, except for those of Dahle et al. and Petrovski, applicable to all sites in general, enable the ground type (rock in this case) to be defined. In order not to increase the safety margin unduly by cumulating margins, the mean values obtained for each law were retained. Lastly, the spectral accelerations obtained with the different laws were averaged. Considering that the law of Ambraseys et al. is the most recent and is based on the greatest number of data from Europe and the Middle East, it was weighted by 3 in these calculations. All the other laws were weighted by 1. The nearby earthquake magnitude is thus 5.4 at a distance of 1 km and a depth of 15 km. Figure 5 shows the spectral accelerations obtained with the above-mentioned attenuation laws. They are elastic spectrums at 5% damping. It can be seen that there is a high amount of variation in the results, which is typical, especially for the near field. There are two reasons for this variation. First, some laws take the distance from the epicentre into account (here 1 km) while others are based on the hypocentre (here, 15 km). Second, the variability of the accelerations in the near field is high (for the same magnitudes and distances), and the attenuation laws are based on a limited number of data for the near field, hence greater uncertainty concerning short epicentral distances. For the distant earthquake (SMS = VII-VIII, distance 50 km), the magnitude retained was 5.6 at a distance of 50 km from Avène Dam and to a depth of 13 km. This study was carried out in 1999 using a calibration spectrum to briefly check the dam and it is important to remember this. For more precise inspection, it would be necessary to perform calculations based on temporal signals taking into account the duration of the event.