A trunk-preserving, demand-weighted edge betweenness centrality framework for partitioning water distribution networks

ABSTRACT The figure shows the trunk preserving demand weighted edge betweenness centrality procedure. Starting with DWEBC computation, then identifying the trunk network, growing DMAs on the residual network, and concluding with evaluating the resulting DMAs using suggested pressure and head loss metrics to select a preferred configuration. District metered areas (DMAs) play a crucial role in managing leakages and pressure, and monitoring operations within water distribution networks (WDNs). However, purely topological design of DMAs overlooks hydraulics and the operational role of highly critical pipes, undermining redundancy and resilience. This study proposes a trunk-preserving hydraulically informed partitioning framework that (i) utilises demand weighted edge betweenness centrality (DWEBC) to delineate a trunk network (TN) and (ii) grows capacitated, size-balanced DMAs on the residual network using an inverse of the DWEBC as edge weight. The approach employs farthest-first k-centre seeding, graph-Voronoi growth, Kernighan–Lin/Fiduccia–Mattheyses connectivity repair, uniformity balancing, and minimum DMA size enforcement. The DMAs are then evaluated using a pressure uniformity index (PUI), head-loss gradient (HLG), demand satisfaction ratio (DSR), and boundary control complexity (BCC) metrics, after which collector edges interfacing TN and DMAs are flagged. The approach is deterministic, bounded in runtime, and implemented in Python using EPANET/WNTR. Results on the benchmark networks demonstrate easy identification of TN, formation of balanced DMA in terms of size and pressure, and interpretable interfaces for metering and control. The framework generalises to real-world WDNs and can seed multi-objective refinement if Pareto exploration of PUI–HLG–DSR–BCC is desired.