- Reservoir drawdown is a management technique increasingly used in maintenance of aging infrastructure, decommissioning dams, and to promote the flushing of fish and sediment to downstream reaches. Erosional processes in the reservoir may result in excessive delivery of sediment to downstream habitats and infrastructure, a critical long-term consideration. Typically, numerical models of the reservoir’s geomorphic response have been limited to 1D incisional erosion without the incorporation of lateral widening. The effective design of drawdown operations via predictive modeling provides a missed opportunity to address concerns around sedimentation of reservoirs, controlling the outcomes of dam maintenance and removal, and mitigating the effects of sediment starvation downstream of dams. A numerical model was developed and simulated to examine how drawdown rate and grain size affect the rate, magnitude, and timing of lateral erosion in a reservoir. The incorporation of retrogressive bank erosion and groundwater drawdown in this model are two new contributions to address sequential slumping, which is expected to improve the accuracy of modeling lateral erosion.
Field measurements from the drawdown for Elwha Dam removal were coupled with five drawdown scenarios: Staged, Staged with Short drawdown increments, Staged with Long drawdown increments, Slow, and Rapid drawdown. These scenarios were then re-evaluated with coarse material to understand the integrated impact of grain size and drawdown process. Three overarching effects of drawdown emerged from the numerical experiments: 1) Staged drawdowns can produce a range of erosion volumes depending on how long the drawdown phases are relative to the hold periods, dictating the generation of undrained or drained conditions; 2) the timing of erosion was directly related to the drawdown rate, creating a tradeoff between the amount of impact created and when the impact is produced; and 3) coarse material reduced the magnitude of erosion compared to fine material when the internal friction angle was greater than the bank slope angle. The inclusion of retrogression as an avenue for sequential block failures proved to be an important contribution as all fine-grained scenarios exhibited at least one time step with more than one failure. The duration of hold periods was found to be less important than the duration of drawdown, as the drawdown increment drives slope stability (e.g. undrained/drained conditions). As a majority of failures occurred under undrained conditions where cohesion is present, the shear strength provided under drained conditions highlights the ability to achieve slope stability.
Several aspects of lateral erosion were simplified or not included, such as the immediate transport of failed materials, no incorporation of vegetation, homogeneous bank sediment, de-coupled incision and widening, assumed drained and undrained conditions, and a limit of model simulation to pre-high flow events. In addition, no validation data were available for model verifications. The use of high-frequency bathymetry surveys or estimations of bank retreat through UAV surveys could be used to better understand the processes behind lateral erosion with quantitative pre- and post-drawdown data. Further work with this model could focus on its integration into an existing hydrodynamic model for more accurate predictions of the magnitude, rate, and timing of lateral erosion. Due to the complexity of sediment transport both spatially and temporally, comprehensive models of both fluvial and geotechnical processes are crucial in understanding the interconnected impact of drawdown operations.