Community Detection in Water Distribution Networks: An Analysis of Graph-Based Algorithms

Community detection helps in finding groups in networks where connections inside the group are stronger than outside. This study compares three popular algorithms—Louvain, Leiden, and Girvan-Newman—to understand how well they identify such groups. Louvain and Leiden focus on improving modularity (a measure of good grouping), with Leiden performing slightly better. Girvan-Newman works differently by removing important connections to form groups, but it takes more time and is not suitable for very large networks. These methods are applied to the Balerma water distribution network to divide it into District Metered Areas (DMAs), which help in managing water supply efficiently. The study uses factors like pipe length, diameter, and pressure to form these groups. Leiden gives the best-defined DMAs, making leak detection and pressure control easier. Louvain is a faster alternative, while Girvan-Newman provides more detailed divisions but is slow. The results can help in better planning of water supply, reducing water wastage, and improving infrastructure management in cities.