Unified parameter optimisation approach for leakage detection and extended-period simulation model calibration

Abstract

Leakage detection is of great interest and significance for water utilities. There are different field testing apparatuses developed for leakage detection, but no systematic modelling approach was available for predicting the most likely leakage hotspots that might guide leakage detection crew to quickly locate exact leakage locations. In this paper, leakage is represented as pressure-dependent emitter flow at a node in a water distribution model. Leakage detection is formulated as a nonlinear parameter identification problem to search for the possible emitter node locations and the emitter coefficients while minimising the same objective function as model calibration, namely the distance between the field-observed and the model-simulated flows and hydraulic grades. Thus the leakage detection optimisation is developed as an integral component of the unified model parameter identification framework. The comprehensive methodology is applied to the water system of a district meter area (DMA) in the UK. It illustrates the effectiveness of the unified approach for both leakage hotspot detection and extended-period simulation model calibration using the real field data.

Keywords:

• water distribution system
• hydraulic model
• leakage detection
• water loss
• model calibration
• genetic algorithm and optimisation

Acknowledgements

United Utilities Water and Atkins Water and Environment Group in the UK are gratefully acknowledged for providing the hydraulic network model and field data used in this paper. The results presented, however, do not represent or imply the opinions of the organisations.