Numerical Methods for Modeling Water Quality in Distribution Systems: A Comparison

Abstract

A comparison is made between the formulation and computational performance of four numerical methods for modeling the transient behavior of water quality in drinking-water-distribution systems. Two are Eulerian-based (the finite-difference and discrete-volume methods) and two are Lagrangian-based (the time-driven and event-driven methods). The Eulerian approaches move water between fixed grid points or volume segments in pipes as time is advanced in uniform increments. The Lagrangian methods update conditions in variable-sized segments of water at either uniform time increments or only at times when a new segment reaches a downstream pipe junction. Each method is encoded into an existing distribution-system simulation model and run on several pipe networks of varying size under equal accuracy tolerances. Results show that the accuracies of the methods are comparable. The Lagrangian methods are more efficient for simulating chemical transport. For modeling water age, the time-driven Lagrangian method is the most-efficient while the Eulerian methods are more memory-efficient.