Predictive control with dynamic constraints for closure and opening operations of irrigation canals

  1. Galvis Restrepo, Eduard
Dirigida por:
  1. Manuel Gómez Valentín Director/a
  2. José Julián Rodellar Benedé Director/a

Universidad de defensa: Universitat Politècnica de Catalunya (UPC)

Fecha de defensa: 15 de enero de 2016

Tribunal:
  1. C. Ocampo-Martínez Presidente/a
  2. José María Maestre Torreblanca Secretario
  3. Pierre-Olivier Malaterre Vocal

Tipo: Tesis

Teseo: 414904 DIALNET lock_openTDX editor

Resumen

Water delivery systems usually work in continuous way based on some prescribed flow conditions and user's needs . However there are situations in which abrupt changes in the operating conditions must be carried on. Typical examples are the alternative closing of a canal system during the non-demand periods to save water for other purposes as energy production, and the closure of a canal due the danger of water pollution in the supplier river. Closure of a canal means setting zero flow conditions by closing the gates along the canal, while maintaining specific water levels under the maximum allowed value. The closure operation requires a progressive and well planned set of actions to avoid overtopping and cracking in the canal lining, which can involve both economic and environmental issues. The opening operation involves restarting the canal to its normal operating condition from zero flow condition. This thesis is devoted to develop a supervised decentralized predictive control strategy for solving the problems related to the closure and opening operations of canal systems. The evaluation is fulfilled by means of numerical simulation on two cases of study in a variety of operating scenarios. The strategy is also experimentally validated through real-time implementation in a laboratory canal available in the Technical University of Catalonia (canal PAC-UPC). The control strategy has been developed in a two-level architecture: (i) a set of individual decentralized downstream water level predictive controllers, which are formulated via an optimal control problem under dynamic constraints and implemented by upstream local gates; and (ii) a supervising level to achieve the compromise of fast execution with smooth gate trajectories and water level regulation, even in the presence of disturbances. The simulation and real-time implementation scenarios have demonstrated that the proposed strategy is convenient for closure and opening of irrigation canals. Problems presented when the canal closure operations are not managed properly, such as overtopping, have been avoided in all the scenarios.