Project Description

Fellow: ESR4 - Mahmood Mahmoodian Host institution: LIST (formerly CRP Henri Tudor) Duration: 36 months on QUICS (48 months PhD) Planned start date: November 2014
Actual start date: May 2015
Project title: Model reduction analysis under uncertainty for a Multiple Model Based Predictive Control approach to dynamically manage water quality (WP4, WP6)
Supervisor name: Dr Ulrich Leopold, Dr Georges Schutz, Prof. Francois Clemens & Prof. Peter Vanrolleghem PhD enrolment: Y (Awarded by TU Delft)

Objectives:

  • Develop methods to reduce a full scale model considering the sensitivity of model states, input/output variables and potential uncertainty of online measurement, especially in the context of a given operational task.
  • Develop methods to link the overall sensitivity analysis of a model to the targeted operational objective.
  • Considering the uncertainty propagation from the inputs through to the final control action.
Tasks and methodology: The methodology is based on connecting input uncertainties with model complexity and calibration uncertainties by considering the context of the MPC or diagnostic approach needs. Develop a MPC approach based on a reduced MM which can subsequently be used in software products and technologies to reduce the impact of unavoidable overflows into sensitive water bodies. ESR4 will be seconded to TUDelft in order to exchange knowledge about the Eindhoven RTC, and seconded to ULaval to discuss the MPC approach for urban drainage systems while focussing on uncertainty and controller performance. ESR4 will be seconded to USFD, to benefit from the knowledge on uncertainty in urban drainage water quality processes in urban drainage network models.
Results: Methodology, and associated software tools to guide the model reduction approach in the context of a given operational task under uncertainty (D4.6, D6.1, D6.6). MPC approach based on a reduced but richer model that can consider water quality during the underlying optimization phase of the MPC and also accounting for uncertainties.
Dissemination: Journal papers. Two presentation/year as MC Ambassador. Protocols and manuals for end-users. Annual conference presentation.
Planned secondments: To TUDelft (5months) to compare modelling approaches and associated uncertainties for TUD and TUDOR case study catchments. To ULaval (2 months) to discuss the MPC approach for urban drainage systems while focussing on uncertainty and controller performance. To USFD, (2 months) to study uncertainty in sediment and pollutant modelling in sewer systems.