AI for Smart Water Solutions in Developing Areas: Case Study in Khelvachauri (Georgia)

Small and mid-sized water utilities face persistent challenges due to limited technical expertise and financial resources, impeding effective management and decision making. This study presents an enhanced version of the MACS Water Smart application, which integrates artificial intelligence and EPANET-based hydraulic modelling with GIS (geographical information system) functionalities to optimize water supply networks. The methodology was applied to the potable water system of Khelvachauri, Georgia, which experiences significant pressure deficits, particularly in its southern area during peak consumption time. By employing machine learning algorithms, the WS tool automates tasks such as pipe diameter optimization and pressure recovery, gradually eliminating the total need for expert intervention. The AI-powered optimization achieved pressure increases above 25 m, reduced flow velocities below 1.5 m/s, improved pumping efficiency by 15%, and lowered leakage rates by 8%. Additionally, computational time was reduced by 35% compared with traditional methods. These findings validate the performance of AI-based hydraulic simulation and its ability to replicate engineering decisions. Furthermore, the tool provides a scalable solution for planning future network expansions. This work highlights the practicality of combining AI and hydraulic modelling for sustainable water management in resource-constrained settings, emphasizing its cost-effectiveness and potential for widespread adoption in small and mid-sized utilities.