How to Model an Intermittent Water Supply: Comparing Modeling Choices and Their Impact on Inequality

Intermittent water supply (IWS) networks have distinct and complicated hydraulics. During periods without water supply, IWS networks drain, and consumers rely on stored water; when supply resumes, pipes and consumer storage are refilled. Draining, storage, and filling are not easily represented in standard modeling software. We reviewed 30 ways modelers have represented the hydraulics of IWS in open-source modeling tools and synthesized them into eight distinct methods for quantitative comparison. When selecting methods, modelers face two critical choices: (1) whether to ignore the filling phase, and (2) how to represent consumers as attempting to withdraw their demand: as fast as possible (unrestricted), as fast as possible until a desired volume is received (volume-restricted), or just fast enough to receive a desired volume by the end of supply (flow-restricted). We quantify these choices’ impact on consumer demand satisfaction (volume received/volume desired) and inequality using three test networks under two supply durations, implemented in two different hydraulic solvers (EPANET and EPA-SWMM). Predicted inequality and demand satisfaction were substantially affected by the choice to represent consumer withdrawals as unrestricted, volume-restricted, or flow-restricted, but not by the specific implementation (e.g., three different flow-restricted methods agreed within 0.01%). Volume-restricted methods predict wider inequalities than flow-restricted methods and unrestricted methods predict excessive withdrawal. Modeling filling delayed water provision unequally, reducing the volume received by some consumers (by ∼20%), especially where water supply is brief. All else being equal, we recommend using volume-restricted methods, especially when modeling system improvements, and including the filling process when studying inequalities.