One month after a reservoir was impounded, there was a significant drop in the water level retained by an embankment dam. The water is used for forest firefighting operations and tourism.
Water was channelled in twice via the toe of the downstream shoulder, in October 2006 and November 2011. The structure then remained inactive for nearly 10 years. In 2016, final rehabilitation work was performed on the bottom pipe crossing the dam. A tower was built at the dam’s upstream toe and connected to the bottom pipe. Two knife gate valves were installed at the tower. One was an upstream gate valve kept open during normal operation. The other was a downstream regulating valve gate kept closed during normal operation to keep water out of the bottom pipe running under the dam. The tower and bottom pipe were connected by a stainless steel pipe mounted between two sleeves. The dam’s impoundment reached its normal level on 15 March 2018.
On 27 April 2018, a significant drop in the reservoir’s level, accompanied by a high flow downstream of the dam, was observed. The reservoir completely emptied itself in two days. The dam’s downstream shoulder was checked for leaks, but none were found. An ecologist visiting the site considered that the impact of this emptying on the waterway and biodiversity was low.
On 29 April 2018 it was found that the 3 mm DN600 stainless steel pipe between the two sleeves had suffered very significant buckling (crushing). During works, the initial plan of the pipe was to fit flanges at both ends (flange-to-flange connection with the bottom drain control valve on the upstream side and flange-to-flange connection with an expansion joint on the downstream side). This would allow the pipe (without any water inside) to withstand 25 m of external water column pressure (i.e., 2.5 bar). However, it was ultimately decided to install this pipe without any flanges between two sleeves and without any expansion joint on the downstream side. Although removing the flanges resulted in greatly reducing the rigidity and buckling strength, the pipe’s design note was not updated. In addition, according to the sleeve manufacturer, the sleeves should not have been used without pipe supports to take up the vertical forces. With a calculated external pressure at the time of the rupture of approximately 16 m, whereas the pipe in this new configuration could only bear a few metres, parasitic movements at the sleeves may have triggered the instability that caused the pipe to buckle between the two sleeves.
The flanges and expansion joint initially planned were fitted five months later. Six months after the reservoir was impounded, a thorough technical inspection confirmed that the hydraulic and mechanical behaviours were satisfactory.
