Measurements and modeling for aqueduct systems

Generic placeholder image

Providing measures and models for aqueducts helps improve water cycle management. A good model can make it possible to improve the management of the water network and allow a general reduction in energy costs related to the pumping of water in the network during the day.
Optimizing this model requires knowledge of the complex mathematical-physical constraints that regulate the water distribution process in a water network; a wide set of historical data (regarding pressure, tank level, etc.) is also necessary to learn an optimal control strategy. All this information can contribute to the construction of a model and to the related software tools for the optimal management of the water network.

Planned activities within PerFORM WATER 2030

Activities regarding measurements and modeling for aqueduct systems are part of the Water research line included in PerFORM WATER 2030.
They include the analysis of the quality of data on water networks supplied by CAP and the subsequent definition of two optimization approaches aimed at decreasing energy costs for pumping systems and at obtaining an optimal mixing of sources with different water quality values.

Objectives and expected results

  • Identification of the key parameters that determine the optimal energy costs of pumping systems.

  • Development of a model of optimization of pumping processes in water networks and related software.

  • Development of applications aimed at optimizing the management and production processes for a reduction in energy costs.

Description of the activities

  • Identification of test sites and water data available.

  • Selection of software tools for application implementation.

  • Implementation of the proposed approaches for a preliminary evaluation on benchmark examples to identify their potential and limitations.

  • Optimization of pumping processes in idylic networks is carried out both offline, using Bayesian optimization techniques, and online through "reinforcement learning" approaches (reinforcement learning).

Partners involved

DISCo Department of Università degli Studi di Milano - Bicocca carries out the analysis of water data provided by CAP Group.

Generic placeholder image
Generic placeholder image
Generic placeholder image

Back to Water research activities

Generic placeholder image

Contact us

Send Email



Reserved Area

Log in

FESR image missing
FESR image missing