back to Geological models and case histories
(Habib, 1987; Londe, 1987a; Wittke and Leonards, 1987)
The 66.5 m high, double curvature concrete arch dam at Malpasset was completed about 1954. Its first filling took place very slowly. In December, 1959, when the reservoir level was 0.3 m below spillway crest level, the dam suddenly failed. The resulting flood wave killed 424 people in the town of Frejus downstream.
The dam is sited in the crystalline Tanneron Massif and consists of banded gneiss, striking generally N-S. The foliation dips between 30o and 50o downstream and towards the right bank. Tectonic disturbances from the Hercynian orogeny onwards have led to close jointing of the rocks, with associated low moduli, and numerous shears and faults. The site investigation was very restricted, consisting of just a few boreholes and visual inspection of the excavation bottom (Londe, 1987b).
Subsequent investigations, conveniently summarised in Leonards (ed. 1987), found no fault with the design or construction of the dam itself. Its collapse was brought about by excessive deformations of the rocks forming the left abutment, culminating in the development of high groundwater pressures which caused the uplift and removal of a large wedge beneath the dam foundation. This wedge was defined upstream by a foliation surface, partially opened by cracking, and downstream by the "downstream fault", missed by the site investigation. The high sensitivity of the permeability of the broken rock mass to applied stress contributed to the development of the high groundwater pressures. Monitoring was very limited, depending entirely on geodetic measurements of targets on the downstream face of the dam, made approximately annually. It is probable that piezometric and displacement measurements in the foundation combined with strain meters in the arch would have given a warning of at least several months before failure (Londe, 1987a).
The failure of the Malpasset arch dam was due entirely to geological and hydrogeological conditions which, because of inadequate site investigation, were not foreseen. In addition, instrumentation that would have permitted the observational method to be applied was not provided. Subsequent events demonstrated that it would have been wise to discard the arch design (Londe, 1987a; Terzaghi, in Goodman, 1999).
Associated models: Mobile belt (Fig. 2.5); Gneisses and migmatites (Fig. 3.17)