Energy-cost efficiency of water supply systems through optimization methodologies

Water Supply Systems (WSS) are critical infrastructures that ensure access to safe drinking water, support economic development, and safeguard public health. However, delivering these services requires substantial energy re sources, with pumping operations typically accounting for 70-80% of total system energy use. As energy prices rise and sustainability goals become increasingly demanding, water utilities face the challenge of operating these systems in a cost-efficient, reliable, and environmentally responsible manner. Despite decades of research on optimization-based pump scheduling and control, the practical implementation of such methodologies remains lim ited, constrained by problems of scalability, interpretability, and integration within existing operational workflows. This thesis addresses these challenges through the development of an opti mization and decision-support framework for pump scheduling in WSS. The work begins with a systematic analysis and quantitative comparison of the existing mathematical formulations, highlighting their modelling trade-offs and limitations. From this analysis, the duty-cycle optimization model is highlighted, providing a continuous-time representation of pump operations that enhances operational flexibility while maintaining scalability to complex network configurations. Building upon this formulation, a novel hybrid metaheuristic designated Smart Dynamic Local Search (Smart-DLS) is developed. The algorithm combines deterministic local search with rule-based perturbation and multi start strategies, achieving high-quality solutions across diverse operational scenarios. The Smart-DLS is subsequently integrated into a modular Deci sion Support System (DSS) that enables operator-in-the-loop optimization through the generation and evaluation of multiple scheduling alternatives according to the water utility specific key performance indicators (KPI). The proposed DSS has been successfully validated using two literature bench marks, the Van Zyl network and the AnyTown modified network. Addition ally, it was tested in two real Portuguese water supply systems: the Ronqueira WSS, which operates with fixed-speed pumps, and the Portuguese Refer ence WSS (PRWSS), which uses variable-speed pumps. For the Ronqueira network, the results demonstrated consistent energy-cost reductions of ap proximately 20% while preserving operational reliability and water tank levels compliance. In the PRWSS, the DSS exhibited both scalability and robust ness under more complex hydraulic and operational conditions, consistently producing feasible and energy-efficient pumping schedules that resulted in a cost reduction of 22%. To ensure robustness under demand uncertainty, an adaptive control mech anism is also introduced, allowing real-time adjustment of pump operation to maintain a feasible and reliable system performance. This feature further increases operator confidence in the adoption of optimization-based method ologies. This thesis contributes significantly to the field of pump scheduling opti mization by providing a systematic review and quantitative comparison of existing formulations, validating the duty-cycle model and developing the hybrid Smart-DLS algorithm. These advancements are incorporated into a modular DSS that has been tested and validated in real operational conditions. The DSS demonstrates its impact on water utilities by improving the cost-efficiency, transparency, and adaptability of WSS management, while also reducing the gap between optimization research and practical operational applications.