Efficient Reservoir Modelling for Flood Regulation in the Ebro River (Spain)

Abstract

The vast majority of reservoirs, although built for irrigation and water supply purposes, are also used as regulation tools during floods in river basins. Thus, the selection of the most suitable model when facing the simulation of a flood wave in a combination of river reach and reservoir is not direct and frequently some analysis of the proper system of equations and the number of solved flow velocity components is needed. In this work, a stretch of the Ebro River (Spain), which is the biggest river in Spain, is simulated solving the Shallow Water Equations (SWE). The simulation model covers the area of river between the city of Zaragoza and the Mequinenza dam. The domain encompasses 721.92 km2 with 221 km of river bed, of which the last 75 km belong to the Mequinenza reservoir. The results obtained from a one-dimensional (1D) model are validated comparing with those provided by a two-dimensional (2D) model based on the same numerical scheme and with measurements. The 1D modelling loses the detail of the floodplain, but nevertheless the computational consumption is much lower compared to the 2D model with a permissible loss of accuracy. Additionally, the particular nature of this reservoir might turn the 1D model into a more suitable option. An alternative technique is applied in order to model the reservoir globally by means of a volume balance (0D) model, coupled to the 1D model of the river (1D-0D model). The results obtained are similar to those provided by the full 1D model with an improvement on computational time. Finally, an automatic regulation is implemented by means of a Proportional-Integral-Derivative (PID) algorithm and tested in both the full 1D model and the 1D-0D model. The results show that the coupled model behaves correctly even when controlled by the automatic algorithm.

Impounding Reservoirs, Benefits and Risks: A Review of Environmental and Technical Aspects of Construction and Operation

“…This direction can be described as leading the way in terms of innovative research into flood-control reservoirs, for which the latest developments in the field of artificial intelligence are also being used [34,35]. Analyses based on the case studies confirm the feasibility of implementing models with sufficiently fast computation times [27,30]. They show a significant increase in the effectiveness of minimising the negative effects of flooding, assuming rational management of available resources with the support of appropriately designed IT tools [28][29][30][31][32][33].…”

“…[ [24][25][26][27][28][29][30][31][32][33][34][35] The analyses concern 11 very large and 11 large reservoirs in Asia, 1 very large reservoir in Europe, 2 very large reservoirs in Africa and 1 large reservoir in the USA. In addition, in publication [28], reservoir capacity was a variable value, and in [35], a review based on a large dataset was performed.…”

“…There are also studies in the scientific literature on optimising the management of reservoir operations according to the hydrological year [24,25], as well as publications proposing tools and models to improve real-time decision making, depending on weather forecasts or prevailing weather conditions [26][27][28][29][30][31][32][33][34][35]. This direction can be described as leading the way in terms of innovative research into flood-control reservoirs, for which the latest developments in the field of artificial intelligence are also being used [34,35].…”

“…Development of guidelines for suggested damming levels at a given time [24,25]. Implementation of models to support real-time operations [26][27][28][29][30][31][32][33][34][35].…”

Shallow Water Equations in Hydraulics: Modeling, Numerics and Applications

“…In [10], the performances of several modeling approaches were compared in order to evaluate their results and computational requirements in a transient river flow event in a reach of the Ebro river (Spain) that includes a reservoir covering a large area. A 2D distributed shallow water model solved over a triangular grid and a 1D shallow water model was used to discretize the full domain.…”