The paper develops a new method of optimized calibration of large-scale non-linear water networks. Calibration requirements are first discussed with relevance to the limited monitored operating data, the ill-defined network parameters and the potential benefits of more accurate network simulations for analysis and control purposes.
A formulation is then made which incorporates simultaneous adjustment of network physical parameters and related operational data in order to force agreement between measured and simulated variables. The system non- linear response is represented by a high-dimensional set of linear- equations relating controllable adjustment parameters to the observed discrepancies in operating conditions. Quadratic performace indices, with optional user-defined weighting factors, are shown to be most appropriate for the control variables. Use of a hierarchical optimization technique permits a solution to be obtained to the problem of high dimensionality with constrained controls and variable performance factors. Application of the scheme to a test distribution system leads to optimal adjustment of all the permitted variables to give the best overall solution for the available measurements.