Introduction and main objectives

It is widely recognized that the life-time performance of current model-based operation support systems, like Model Predictive Control (MPC), Real-Time Optimization (RTO) and soft-sensors for large-scale complex dynamic processes is rather limited, particularly due to the fact that the underlying dynamic models need to be adapted/calibrated regularly, requiring dedicated measurement campaigns executed by highly specialized engineers. Therefore in this project a model-based operation support technology is developed that enables control and model calibration at a considerable higher level of autonomy than currently possible. This operation support system should be able to optimize plant performance under varying operational conditions and adapting to changing circumstances. The academic partners cover the wide theoretical scope of the project and the industry is specifically involved to guide research in an industrial relevant direction and to ensure industrial validation of the technology to be developed.

Technical approach

In the proposed procedure, the performance of any model-based application is continuously monitored. If a deteriorating performance is detected, a performance diagnosis application is started that has to indicate whether the reason for the deteriorated performance is a model deterioration or requires a retuning of the application on the basis of the same model. If a plant test is desired to calibrate the model, this should be done in an automated way, where an economically least-costly identification experiment is designed. Based on the calibrated model the model-based application is retuned and activated.

Expected Impact

The expected impact of the proposed technology is threefold:

1) an increase of the life time performance and the economic benefits of model-based operation support systems against a reduced investment

2) much tighter control and more accurate model-based measurements of industrial processes resulting in higher energy efficiency and a significant reduction of the environmental load

3) a reduction of the need for on-site presence of model-based experts for the implementation and the maintenance of model-based systems. This technology will therefore inevitably strengthen the competiveness of the European process industry and of the European suppliers of model-based operation support systems.

Project team

The project is a collaboration between four academic partners, TU Delft, Eindhoven University of Technology, RWTH Aachen University and KTH, and three industrial partners ABB, Boliden and SASOL.

Division of Decision and Control Systems KTH

Project funding and duration

The project is funded by the EU FP7 framework program with duration 2010-2013.

Project website

Håkan Hjalmarsson
Professor of Signal Processing

My research interests cover system identification, process modeling and control, and communication network