Federico Castiglione

Dam management optimization is a fundamental way to improve the efficiency of artificial reservoirs, especially in the light of future water management challenges such as, for example, the increase in water demand, the decrease in water quality, and climate change. However, the optimization problem is extremely complex, mostly because of its nonconvex, multiobjective, nonlinear, and discontinuous nature. Furthermore, it is affected by uncertainties and by a large number of constraints, and it involves stakeholders with a great political and social relevance. Indeed, despite the great interest of countries and scientific research towards the creation of methods for dam management optimization, a large gap between theory and practice still exists.

The research project tries to fill the gap between theory and practice in dam management, which is mostly due to: i) the scepticism of dam managers and operators about using models that replace their judgement; ii) the mathematical complexity of optimization models, that makes them difficult to understand; and iii) the need to adapt models and methods to new case studies and to upgrade them to evolving systems.

A dam management optimization methodology is proposed that directly involves one or more decision-makers, as for example a committee of stakeholders and experts on the case study. The decision-makers can provide their preferences concerning a set of dam management strategies. Such information is used to create a mathematical model of their preference, which is then used to guide the optimization towards the strategy that is most preferred by the decision-makers. The methodology can be applied to multi-purpose dams with objective functions concerning flood mitigation, sustainable use of water resources, and hydroelectric power.

The methodology was applied to the case study of a dam in Sicily with 5 objective functions, 3 for flood mitigation, 1 for irrigation supply, and 1 for hydroelectric energy supply, and it was able to: i) obtain management strategies superior to the currently used one; ii) greatly simplify the problem formulation by using the preference model to carry out the normally very complex estimate of the objectives’ trade-offs; and to iii) ease the stakeholders scepticism by directly involving them in the optimization process.