Development of the hydroelectric production optimization framework within Hydronomeas software - Investigation in the Acheloos-Thessaly hydrosystem

Hydropower systems optimization presents a strong challenge for the modeling community, since it is characterized by a number of difficulties such as the stochasticity of inflows and the non-linear nature of energy production. It is also an issue of increasing importance, in a world of increasing demands and rapid changes in energy production, where the imminent fossil fuel depletion leads to the rise of renewable energy resources. Older optimization attempts were restricted on simple hydrosystem topologies and heuristic rules, while newer attempts combine non-linear optimization with many supplementary techniques, such as fuzzy logic and neural networks, often with limited results on more complex issues. Hybrid simulation-optimisation approaches are also gaining ground, as they allow stochastic elements in the analysis. In the case of the Hydronomeas Decision Support System (DSS), an innovative parameterisation-simulation-optimisation scheme is employed, based on the work of Nalbantis and Koutsoyiannis (1997), allowing a more elaborate approach on hydrosystems management. In the following thesis, this scheme is expanded to include the management of hydropower systems under a variety of different management scenarios and policies. The DSS is then employed in the case of Acheloos-Thessalia hydrosystem, which presents a challenging case, having seven hydropower plants, both serial and in parallel, with many conflicting uses and stakeholders. Firm energy potential is explored, as well as potential benefit from the combined use of hydropower generation and irrigation. A cost/benefit approach on energy production and irrigation is also employed and analyzed. The optimal scenario under various prerequisites is suggested, taking into account both up-to-date environmental flow policies and irrigation demands. The stochastic nature of inflows is also taken into account; stochastic time-series are generated based on historical data, having either HK or short-term autocorrelation dynamics. The impact of the assumed stochastic dynamics on hydrosystems management is then evaluated. Finally the sensitivity of solutions against the assumptions of the stochastic simulation model, as well as the uncertainty of inputs, is examined. The results show that Hydronomeas is a solid Decision Support System, able to find the optimal management policy and explore a hydropower system's potential, either with firm energy policies or with cost/benefit strategies.