Feasibility and effectiveness of metaheuristic algorithms in optimizing valve placement for Water Distribution Networks optimization

In Italy, the majority of municipalities are small towns, with 70% of them having a population of less than 5,000 inhabitants. These small towns often face more complex challenges compared to larger cities, both due to limited economic resources and less developed infrastructure. This disparity is also evident in the water services sector, where management companies tend to focus more on the networks of larger cities. The latter are generally prioritized in the planning and execution of interventions, whether for renovation or maintenance, due to their higher population density and greater economic returns. Conversely, small towns are often overlooked, resulting in limited funding and modernization of their water distribution networks. This insufficient funding barely allows small towns to keep their water distribution networks operational in day-to-day management, making it almost impossible to invest in essential renovation and modernization efforts. The absence of monitoring, detection, and metering devices within the network prevents managers from gaining a comprehensive understanding of its current conditions, often leading to inefficient management. Such inefficiency is evident in the lack of awareness regarding the location of leaks, the unsystematic approach to pipes maintenance and replacement, the increased management and maintenance costs, and delays in the allocation of financial resources for these purposes. It is also reflected in the reliance on a reactive, post-intervention policy rather than a proactive, preventive approach, which could considerably reduce the financial burden on the system. During drought periods, small towns often rely on emergency measures, such as using water tankers to deliver water to residents. Such reliance stresses their vulnerability and weakens their resilience compared to larger cities. All these factors exacerbate the deteriorating condition of the WDN in small towns, rendering it less resilient, reliable, and effective. Basilicata region is no exception with 75% of its municipalities are small towns and leakage rate accounts for 60%, the highest in Italy. The mountainous terrain of the region makes small towns more isolated and dispersed, complicating the renovation and maintenance of the WDN and increasing associated costs. Considering these challenges, small towns with limited financial resources must face significant issues such as reducing water and energy losses ensuring networks are resilient to climate change impacts. The introduction of technologies for pressure regulation and monitoring can significantly improve the management of water distribution networks, such as those analysed in this study. Pressure monitoring and control provide several immediate advantages: on the one hand, optimal pressure management helps reduce water losses; on the other, a substantial improvement in informational capabilities enables the implementation of alert systems, allowing for timely interventions in case of malfunctions. This approach enhances infrastructure resilience, avoiding costly structural modifications. Despite its clear potential, small-scale entities often face challenges in adopting such solutions due to management complexities and high initial investment requirements. To address these challenges, this thesis focuses on the application of the Green Valve System (GVS), an integrated device that combines monitoring, pressure control, communication systems into a single solution that is fully energy self-sufficient. this study proposes a methodology to optimize the placement of devices within the network. Although the problem may appear theoretically straightforward, optimizing valve placement is a complex, nonlinear, and non-convex problem. For example, considering a network of 150 pipes, placing five PRVs would require evaluating over 591,600,030 possible combinations, not including the subsequent optimization of valve settings, which would further increase the number of combinations. This large search space highlights the importance of developing and applying advanced optimization methodologies capable of identifying practical and efficient solutions to enhance the sustainability and efficiency of water networks in small towns. In this regard, there are two main approaches to optimization: deterministic and metaheuristic methods. The deterministic approach, such as linear programming (LP) and nonlinear programming (NLP), guarantees finding the optimal solution. However, it requires prior knowledge of the problem and the relationships between its parameters. Additionally, it is characterized by slow convergence and is computationally expensive. These disadvantages have led to the emergence of metaheuristics, which are problem-independent, much faster, and computationally less costly. In the contrary to blind Pure Random Search, the metaheuristic algorithms use the randomness in an intelligent and biased way (Stützle 1999). They do not guarantee the optimal solution, though, they offer a practical and near-optimal solution. Metaheuristics expands the exploration over new regions in the search space and find a balance between the exploitation and exploration capability over the search space (Blum and Roli 2003). These algorithms have been widely used in many engineering fields to solve different optimization problems, thanks to their flexibility and capability of finding a near optimal solution for complex problem, where the full enumeration of feasible solutions is beyond the practical objectives. The objective of this thesis is to reduce water losses and promote the sustainability of water distribution networks in small towns through the use of metaheuristic optimization algorithms and the implementation of green valves. The proposed approach has been validated on benchmark networks and applied to real water networks in two municipalities in the Basilicata region, southern Italy. The thesis is structured as follows: • Chapter 1: examines the challenges facing water distribution networks and describes innovative solutions to improve their efficiency, sustainability, and resilience. • Chapter 2: provides a literature review of metaheuristic algorithms, highlighting their advantages in addressing complex optimization problems. • Chapter 3: details the adopted methodology and the characteristics of the study area. • Chapter 4: discusses the results obtained and their implications. • Conclusions: summarizes the main findings of the research and offers perspectives for future developments.