Shankar Narasimhan S

Optimal operation of water distribution networks (WDNs) is concerned with meeting consumer demands at desired pressures in an efficient and equitable manner while conserving resources. This can be achieved by implementing advanced control schemes such as model predictive control (MPC). If sufficient water is available, the control objective is to meet consumer demands while preventing wastage. On the other hand, if the available water is insufficient or inadequate to meet consumer demands at the required pressures, equitable distribution of the available resource is of primary importance. In this contribution, a nonlinear model predictive controller is proposed for optimal operation of WDNs that can deal with both the above situations. The proposed approach takes into account availability of storage facilities at the source and demand points. In addition, the control algorithm can account for plant-model mismatch. Performance of the proposed model based control strategy is illustrated through numerical simulations of an illustrative WDN operating under various water availability scenarios. In the water sufficient scenario, the proposed MPC strategy is able to meet the consumer requirements while minimizing the excess amount of water supplied. In the water deficient scenario, the MPC algorithm is able to exploit the available storage facilities at consumer end to reduce the daily supply deficit by about 20%. Using a longer prediction horizon in MPC results in a further reduction of about 40% in the daily supply deficit.

Estimation of pipe roughness coefficients is an important task to be carried out before any water distribution network model is used for online applications such as monitoring and control. In this study, a combined state and parameter estimation model for water distribution networks is presented. Typically, estimation of roughness coefficient for each individual pipe is not possible due to non-availability of sufficient number of measurements. In order to address this problem, a formal procedure based on K-means clustering algorithm is proposed for grouping the pipes which are likely to have the same roughness characteristics. Also, graph-theoretic concepts are used to reduce the dimensionality of the problem and thereby achieve significant computational efficiency. The performance of the proposed model is demonstrated on a realistic urban water distribution network. © 2009 Springer Science+Business Media B.V.