The integration of digital twin technology with optimization tools has the potential to improve operational efficiency and decision-making processes in complex engineering systems. Accordingly, this work presents a smart predictive digital twin developed for Water Supply Systems (WSS), with the goal of reducing costs in real-time operations. This digital twin uses high-fidelity hydraulic models of the physical system combined with real-time data and advanced nonlinear optimization techniques to enable a continuous monitoring, predictive analysis and optimization of the system. Optimization allows automatic adjustment of operating strategies based on energy tariffs and water demand, with the aim of minimizing energy costs, improving resource management and increasing reliability. The developed framework provides a comprehensive visualization of the WSS's performance by combining real-time data from sensors, historical trends, and predictive water demands. This digital twin has been successfully implemented in a real WSS in Portugal, demonstrating significant improvements in operational efficiency and decision-making capabilities. This work provides a valuable framework for stakeholders to examine system behavior, test scenarios, and implement optimized control strategies without interfering with real-world operations. Ongoing developments aim to improve the adaptability, scalability, and integration of the digital twin with emerging technologies, such as machine learning, to model the WSS hydraulic behavior.