Comparative analysis of approaches to modelling water distribution networks

Abstract

Models for water distribution networks have become indispensable tools for understanding system behaviour by simulating pressures and flows at different locations and times in the networks. This helps inform actions that should be taken in order to improve network performance. The rationale behind traditional modelling of water networks uses a simplifying assumption that pressures in the network are a function of demand, so that it is demand that drives network performance. This demand-driven approach yields good results when pressures in the network are generally sufficient, but exhibits profound weaknesses as it produces unreliable results when network pressures are low. It therefore appears that if the network to be modelled has low pressures, then rather than employing the aforementioned demand-driven approach, the pressure-/head-driven approach should be used. In demand-driven analysis, the higher the outflow the lower the pressure while in head-driven analysis, the higher the pressure, the higher the outflow. The key difference and advantage of the head-driven approach is observed when a more realistic network model is analysed. This paper uses a real-world network to clarify the difference between the demand-driven and head-driven approach, and highlights the superiority of the latter during low-pressure conditions.