Integrated framework for flood risk assessment across hydroelectric reservoirs by coupling multi-scale models: from hydrological simulation to spillway gate management

Effective assessment and management of flood risks is of great importance for the safeguarding and sustainability of the man-made environment. The aim of this thesis is to develop an innovative research framework for flood risk assessment of hydroelectric reservoirs, by combining models of different time scales. The proposed methodological procedure includes a coupling of modelling approaches, from the hydrological simulation of the river basin to the final flood management, through appropriate manipulations of gates and turbines. For its pilot implementation, the Pournari I dam at Arachthos river was chosen. The study begins with the development of a daily hydrological simulation model of the river basin, which describes the sequence of water balances across the different soil zones, in order to convert rainfall data into runoff. Starting from these inflows, a model of "continuous" simulation of reservoir operation in a daily step is constructed, aiming at representing the storage changes and identifying overflow episodes, resulting from a strategic management policy. In the disaggregation model of average daily flows, the inflow data of the selected flood events are converted to a finer time scale, i.e., hourly, through empirical rules that ensure both mathematical and hydrological consistency. Then, an appropriate and effective policy for handling the flood control works (turbines, gates) is investigated for the optimal utilization of the available storage, in order to maximize the potential energy production and at the same time protect the project from flooding events. This policy is implemented by means of a hourly operation model of the reservoir, driven by the disaggregated hourly hydrographs. Finally, the need to extract reliable statistical results leads to the production of synthetic time series of daily precipitation of 1000 years length, for the stochastic application of the developed modelling framework. The present research demonstrates that such an integrated approach, through the above model coupling, can contribute to the field of flood risk assessment of hydropower reservoirs. Moreover, it provides the possibility of applying it to spillway sizing through a statistically consistent manner, thus overcoming the rough and rather ultraconservative assumptions of the usual practice of hydrological design of safety works.