From Real to Laboratory Scale: Simulation and Scaling Methodology for a New Benchmark Water Distribution Network Model

Optimisation of water distribution networks (WDNs) for leakage reduction and energy recovery has been extensively investigated through numerical simulations. However, a significant gap remains in the physical validation of these methods because of the lack of scalable laboratory benchmarks. This study addresses this limitation by developing a systematic scaling methodology to translate the Jowitt and Xu benchmark network into a controllable numerical implementation laboratory-scale model, E-NET. The methodology uses a flow-scale factor of 4.33, derived through recursive pipe-diameter adjustment to preserve hydraulic similarity, specifically similarity in head loss and Reynolds number, using commercially available HDPE pipes. To manage leakage, pressure-reducing valves (PRVs) were implemented using a genetic algorithm (GA) to optimise both valve locations and pressure settings under different load conditions. The optimisation followed a pressure-dependent leakage model with a minimum service pressure constraint. Quantitative results showed close agreement between the Jowitt and Xu and the E-NET model, with RMSE values of 0.38 m for head loss. The GA successfully identified broadly consistent optimal PRV locations in both networks, particularly under low-demand conditions. The leakage discrepancies were generally small across the tested valve configurations. This work provides a numerically validated basis for a laboratory benchmark platform and represents a first step towards bridging the gap between theoretical optimisation and physical implementation in WDN pressure management.