Modifier-adaptation-based RTO scheme for water distribution networks under demand and parametric uncertainties

The goal of pressure management in a water distribution network (WDN) is to avoid losses due to excessive pressure while meeting the minimum target pressure at nodes. Since nodal demands can fluctuate, real-time control of nodal pressures is critical for normal network operation. Optimising the operation of WDN using a model with uncertain parameters and unaccounted nodal demands generates solutions that are not truly optimal and may even be infeasible. This work aims to achieve real-time optimal operation of a WDN in the presence of various uncertainties. A modifier-adaptation (MA)-based real-time optimisation (RTO) strategy is used to drive the WDN to its optimal point. However, the MA-based RTO scheme assumes knowledge of key variables which may not be available in practice. Therefore, a Bayesian matrix completion approach for robust state estimation is used to impute unknown model parameters with limited measurements. Simulation results demonstrate the ability of this approach.