Optimal calibration and sampling design for hydraulic network models (1997-2000)

Funding body: Teaching Company Directorate, DTI and Ewan Optimal Solutions Ltd.

Calibration of computer models for network analysis is a regular component of the model building process. The process generally first involves a series of field tests during which pressures and flows are recorded at strategic locations in the system. This is followed by a desk exercise during which adjustments are made to the roughness values used in modelling the system until a satisfactory match is obtained between modelled and observed values. The selection of a satisfactory set of roughness values can be a tedious business when undertaken by the traditional trial and error approach. In this work, several methodologies were proposed to assist the modeller in the following two tasks: firstly in the selection of sensitive sampling locations in a water distribution system and secondly in the derivation of a good calibrated hydraulic network model. A new calibration approach which consists of adjusting the pipe roughness, the pipe diameter and the nodal demand which certain limits, is proposed by using a genetic algorithm search method.

Three new sampling design approaches were also proposed. The first two approaches rank potential sampling locations based on the shortest path algorithm logic, while the third approach searches for the optimal set of monitoring points by maximising the entropy function using a genetic algorithm search method. The calibration and sampling design approaches are demonstrated by using six hydraulic network models, including three real-life networks. The calibration results show that the genetic algorithm approach consistently achieves more accurate fits than manually worked solutions while the sampling design results demonstrate the potential for financial savings through more efficient equipment deployment.


  • De Schaetzen, W.B.F. (2000) Optimal Calibration and Sampling Design for Hydraulic Network Models, PhD thesis, University of Exeter.
  • De Schaetzen, W.B.F., G.A. Walters and D.A. Savic, (2000), Optimal Sampling Design for Model Calibration Using Shortest Path, Genetic and Entropy Algorithms, Urban Water, Vol. 2, No. 2, pp. 141-152.
  • De Schaetzen, W., V.J. Ewan, D.A. Savic and G.A. Walters (1998), A Genetic Algorithm Approach for Rehabilitation in Water Supply Systems, International Conference on Rehabilitation Technology for the Water Industry, Lille, France, 23-25 March.
  • De Schaetzen, W., D.A. Savic and G.A. Walters (1998), Genetic Algorithms for Pump Scheduling and Cost Optimization in Water Supply Systems, Hydroinformatics 98, Babovic, V. Larsen, L.C. (eds.), Balkema, Rotterdam, pp. 897-899.
  • Walters, G.A., D.A. Savic, M.S. Morley, W. de Schaetzen and R.M. Atkinson (1998) Calibration of Water Distribution Network Models Using Genetic Algorithms, in Hydraulic Engineering Software VII, Blain, W.R. (ed.), Computational Mechanics Publications, pp. 131-140.
  • De Schaetzen W., D.A. Savic, G.A. Walters and M. Randall-Smith (1999) Optimal Loger Density in Water Distribution Network Calibration, in Water Industry Systems: Modelling and Optimisation Applications, Vol. 1, Savic, D.A. and G.A. Walters (eds.), Re-search Studies Press, Baldock, Hertfordshire, England, pp. 301-308.

Back to Water distribution system management