Water demand forecasting for the optimal operation of large-scale drinking water networks: The Barcelona Case Study.

Ajay Kumar Sampathirao, Juan Manuel Grosso, Pantelis Sopasakis, Carlos Ocampo-Martinez, Alberto Bemporad, Vicenç Puig

Research output: Contribution to journalArticlepeer-review

32 Citations (Scopus)


Drinking Water Networks (DWN) are large-scale multiple-input multiple-output systems with uncertain disturbances (such as the water demand from the consumers) and involve components of linear, non-linear and switching nature. Operating, safety and quality constraints deem it important for the state and the input of such systems to be constrained into a given domain. Moreover, DWNs' operation is driven by time-varying demands and involves an considerable consumption of electric energy and the exploitation of limited water resources. Hence, the management of these networks must be carried out optimally with respect to the use of available resources and infrastructure, whilst satisfying high service levels for the drinking water supply. To accomplish this task, this paper explores various methods for demand forecasting, such as Seasonal ARIMA, BATS and Support Vector Machine, and presents a set of statistically validated time series models. These models, integrated with a Model Predictive Control (MPC) strategy addressed in this paper, allow to account for an accurate on-line forecasting and flow management of a DWN.
Original languageUndefined/Unknown
Pages (from-to)10457 - 10462
JournalIFAC Proceedings Volumes
Issue number3
Publication statusPublished - 2014

Bibliographical note

19th IFAC World Congress


  • Demand Forecasting, Model Predictive Control, Drinking Water Networks, Control of Large-Scale Systems

Cite this