Built and operated by EDF (Electricité de France), the St-Egrève project, commissioned in 1991, is a run-of-river power station located in the Grenoble urban area in the French Alps (9270 km² catchment area), downstream of the confluence of the Isère and Drac rivers (Figure 1).
The St-Egrève power plant is composed of two 23 MW bulb units operating with 9 m gross head and a total design discharge of 540 m3/s.
The dam comprises five identical outlets with overflow flaps, and a 25 m wide tainter gate with 6 m of lifting height and a weir at elevation 196.50 m NGF (Nivellement Général de la France). The normal reservoir Free Surface Level (FSL) during operation is 205.50 m NGF.
The length of the reservoir is about 4.5 km including 2.5 km downstream of the Drac-Isère confluence. The width of the reservoir downstream of the confluence is 250 m. The embankment levees, which constitute lateral boundaries of the reservoir, are developed over a length of approximately 20 km. These levees pre-existed before the construction of the dam. They were reinforced and locally raised as part of the construction of the hydro scheme, and their permanent hydraulic loading is linked to the creation of the reservoir. Only a stretch of approximately 4 km (downstream of the Drac-Isère confluence) is under permanent load. Upstream of the confluence, the level of the natural ground in the downstream face of the retaining levees, is in fact higher than the level of the normal reservoir.
Hydrology and sediment
The initial capacity of the reservoir in 1992 was 5.9 Mm3. The initial capacity downstream of the confluence was 3.9 Mm3.
The hydropower plant is designed with a maximum turbined discharge of 540 m3/s.
The Isère and the Drac are rivers with a mixed snow-rainfall driven regime (Figure 2); they feature average flows (“modules”) of 182 and 102 m3/s for catchment areas of 5791 and 3656 km2, respectively. The flow is greater than 600 m3/s six days a year on average (Figure 3).
The Isère and the Drac catchments have a very significant natural sediment production. However, due to their location downstream of a plain formed by an old lake, upstream facilities (dams, deposition beaches, etc.) and past extractions, the coarse sediment supply into the reservoir remains limited. The input of fine sediments, despite the upstream facilities, remains very significant. The suspended sediment transport (SST) was estimated at the construction of the dam at 1.9 Mt/year which is significant in comparison to the initial reservoir capacity. Intra- and inter-annual variations in sediment flows are very significant (Figure 4) with hourly concentrations that can exceed 30 or even 40 g/l on the Isère River during some events.
Sediment challenges
As the reservoir is in an urban area, special vigilance must be taken to avoid the risk of levees overtopping during floods. A freeboard of 1 m along the dikes must be guaranteed for a flood of 3,000 m³/s downstream of the Isère-Drac confluence. It is therefore necessary to guarantee that reservoir fine-sediment siltation stays compatible with this objective.
Management measures
Initial management measures
The risk of filling the reservoir with fine sediments in a few years was considered in the design of the project and the definition of its management. Thus, frequent periodic flushing operations were considered as soon as the dam construction enabled control of the reservoir aggradation.
The flushing flow threshold was set at 600 m3/s on daily average to ensure an annual flushing event (this value was lowered to 550 m3/s if no flushing was carried out the previous year).
To monitor the evolution of the reservoir, the following measurements have been carried out:
- frequent bathymetric surveys (with a minimum area including the sector downstream of the confluence),
- flow discharge measurements,
- suspended sediment concentration measurements (permanent turbidity station upstream and downstream the reservoir),
- sediment samplings in the reservoir.
As anticipated, in the first 20 years, the Saint Egrève reservoir showed a sediment accumulation which was contained thanks to flushing operations (Figure 5). Observations showed that a channel is deepened during flushing operations while a bank on the left side continues to silt up (Figure 6).
This bank sedimentation has caused two major issues:
- Flooding: If a hydraulic model without bottom evolution is used, the freeboard of 1 m is not guaranteed for some of the bathymetric surveys. The global erosion of the channel occurs but a sufficient erosion at the flood peak is not guaranteed.
- The flow capacity of the two gates on the left side of the dam have decreased.
New set of management measures
The basis of sediment management measures was synthetized, at EDF-Hydro, with the concept of “target-state” strategy. “Target-state” refers to a durable sedimentation situation where all key functions remain fully satisfied with reasonable margins and maintained in the long-term with a set of actions defined by tools (expertise, modeling, data analysis) and monitoring (Figure 7).
1/ Define the maximal level of the channel bottom
Due to the monitoring carried out in the reservoir, it was possible to calibrate and validate a 1D hydro-sedimentological model (sediment module COURLIS of the open source TELEMAC-MASCARET system) and to determine the kinetics of the evolution of the riverbed during a flushing event. This model includes several sedimentary layers whose erosion characteristics are based on the age of the sediments in the reservoir deduced from historical bathymetries (Figure 8).
The results of the calibration and validation of this model are given on Figure 9 and Figure 10 and indicate that it is possible to correctly reproduce the evolution of the bed and suspended sediment concentrations (downstream of the dam) during flushing events.
These numerical model results highlight the importance of having a sufficient data set to build a model that is as predictable as possible. Figure 11 highlights different settings (different combinations of critical bed shear stress for erosion and Partheniades parameters) that give the correct cumulated erosion in 2008 but not the right erosion kinetic and/or a wrong cumulated erosion in 2010 (ref. [1] and [2] for more detail).
Finally, due to this model, it is possible to demonstrate that the freeboard of 1 meter is guaranteed for a flood event with a discharge peak of 3000 m3/s (reference flood event) and the bathymetric bottom of 2008 (historical highest bottom).
2/ Manage the bank geometry
The banks continue to silt up despite the flushing events.
It was decided to dredge the bank on the left side of the reservoir to limit the loss of gate discharge capacity, to prevent vegetation from growing and to continue to be able to manage it. The operations were designed to have a bed elevation maintained between Free Surface Level (FSL) minus1 m(to prevent vegetation development) and FSL minus two meters (to optimize the dredging period, forecast every ten years and to limit deposition due to a too large water column).
The dredging was a dilution-pumping method with transfer of the sediment downstream passing through the power plant units (Figure 12). The dredging volume was about 100 000 m3 per campaign (one meter of sediment removal on 100 000 m2 of bank) from April to June. The first campaign took place in 2011, the second in 2019.
However, EDF allows siltation of some small areas on the right bank to promote wet area.
A 2D hydraulic numerical model was then developed (hydrodynamic module TELEMAC-2D of the open source TELEMAC-MASCARET system) to optimize the dredging area of the left bank immediately upstream the left gate of the dam in 2011 (Figure 13). This optimization of the geometry was intended to
- improve the flow rate of the left gate,
- define the most sustainable geometry,
- improve the management of the gates during flushing.
The results of the model show that it was possible to highly increase the bed shear stress in the dredged area with an optimal geometry combined with the closing of the rightmost gate.
3/ Other management actions
EDF considered doing a few targeted sluicing releases based on the largest annual supplies. SST flux is still monitored, and alarm thresholds are determined. Sluicing is experimental to allow some significant sediment flux to transit (Figure 14) to:
- limit the filling of the reservoir and thus the sluicing concentration,
- limit the sedimentation of the bank and thereby increase the time between dredging operations.
Bathymetric surveys and SST flux are analyzed every year during the spring period to estimate the volume of sediment accumulated in the reservoir, to optimize the reservoir lowering at the beginning of the flushing event and to limit the concentration of flow downstream of the dam.
Finally, even if coarse sediments have not been an issue to date (no contribution from the Isère River; low contributions from the Drac River), the evolution of the upstream bottom of the reservoir on the Drac River has been monitored for several years and an analysis on how to manage these coarse sediments (dredging with or without depositing a part of the excavated material downstream the dam, depending on the environmental needs of the downstream reach) is in progress.
Conclusion
St Egrève reservoir has large fine sediment loads. Even if flushing has been planned since the commissioning of the reservoir, this management approach is not sufficient.
A study has been carried out, based on the “target state approach” scheme shown in Figure 7, to integrate all the issues in relation to sediment management.
An adaptive management strategy has been set up, including different management tools (flushing, dredging, sluicing) depending on the issue and the area to be treated. The optimization of this management is based on expertise analysis and modelling made possible due to an important set of data.
References
[1] Valette, E. Jodeau, M. (2012). How to predict the sedimentological impacts of reservoir operations? Submitted to ICSE conference Paris France
[2] Valette E., Tassi P., Jodeau M., Villaret C. (2014). St-Egrève reservoir – Modelling of flushing and evolution of the channel bed. Submitted to River Flow 2014 conference, Lausanne, Switzerland.