Scenario based model predictive control with application to water resources systems
Project Leader: Erik Weyer
Staff: Chris Manzie, Wenyan Wu, Ye Wang
Collaborators: Angus Simpson (University of Adelaide)
Primary Contact: Erik Weyer (email@example.com)
Keywords: control and signal processing; optimisation; water resources
Disciplines: Electrical & Electronic Engineering
Urban water distribution networks are critical infrastructure, and it is taken for granted that water is available on demand. The operations of an urban water distribution network is becoming increasingly complex and costly. In particular the energy cost can be high due to pumping and the operations of desalination plants. Water authorities all over Australia and elsewhere are facing new operational challenges in the near future due to climate variability and increased awareness of the environmental footprint of urban water supply.
In this project we will investigate an Economic Model Predictive Control (MPC) approach to optimise the operations of water distribution networks. In MPC an optimisation problem is repeatedly solved over a moving time horizon in order to determine pumping schedules, flow through turbines, transfers to and between reservoirs, etc. Every time the optimisation problem is to be solved, the problem is updated to reflect the most recent forecasts of demands and electricity prices and measurements of levels in storages and balancing tanks, etc. The word ‘economic’ indicates the optimisation criterion is of an economic nature, but it can be a more general performance measure which in particular takes the environmental footprint of the operations into account.
There is significant uncertainty, in particular with respect to the electricity prices, but also with respect to seasonal rainfall. How to take this uncertainty into account is an open problem, but a scenario based approach to MPC utilising ensemble forecast of rainfall and possibly also electricity prices is a promising approach which will be investigated in this project.