Empirical travel time estimation in a distribution network

Abstract

In the calibration procedure adopted for most hydraulic network models, parameters are adjusted to ensure that measured and computed pressures are approximately equal. Little emphasis is placed on confirmation of a correct flowrate distribution. Such procedures can therefore promote erroneous flowrate solutions. However, it is becoming increasingly common for network models to have water quality modelling capabilities. The age of water, and many chemical reactions associated with changes in water quality are largely dependent on residence time, which is a function of the flowrate distribution. The need for greater accuracy in flowrate prediction has highlighted the need to enhance the standard calibration and verification procedure. An automated and easily repeated procedure for the empirical estimation of travel-time is described. The method involved increasing the conductivity of a pocket of water entering the network via the addition of a tracer solution, and tracking its progress using conductivity loggers. Results of the application of this method in a distribution network are compared with age of water predictions from network water quality modelling software. Discrepancies were attributable to the inaccuracy of a model calibrated using pressure alone. Large discrepancies between modelled and measured age of water existed where the flow regime was complex.