Note: This content is accessible to all versions of every browser. However, this browser does not seem to support current Web standards, preventing the display of our site's design details.


Model Predictive Control for Large Scale Systems: Two Industrial Applications


A. Sahin

no. ETH - Diss.-Nr. 18792

This thesis is concerned with the application of Model Predictive Control (MPC) to large scale systems. Two industrial applications are studied here: reduction of the energy consumption in hot blast stoves and water level regulation in a cascade of river power plants. The contents of this thesis is summarized in Chapter 1. In Chapter 2, a brief introduction to MPC is given, where the general concept of MPC is explained and the components of the optimal control problem are described. In Chapter 3, the application of MPC to hot blast stove operation is presented. The hot blast stove operation requires a continuous heat supply, which is obtained by burning combustion air enriched with fuel in the combustion chamber of the stoves. The amount of fuel used in this operation corresponds to 30% of the total fuel consumption in the whole steel making process. Therefore, it is of primary importance to minimize the heat supply and hence the energy consumption in the hot blast stove operations in order to reduce the costs of the steel making process. While minimizing the heat supply, several constraints on the system variables need to be taken into account. By an appropriate selection of the controller structure, the minimization of energy consumption in the presence of constraints is formulated as a constrained optimal control problem. A simple linear control model is derived by performing step response experiments on a detailed simulation model and a linear MPC scheme is developed based on this control model. The performance of the MPC scheme is demonstrated by performing simulations on the detailed simulation model. In Chapter 4, the application of MPC to a cascade of river power plants is presented. The river power plants interrupt the natural flow of a river and they create artificial fluctuations in the water level and discharge within the river. As these fluctuations are undesired for environmental and navigational reasons, the authorities impose restrictions on the operation of the power plants. The water levels upstream of each power plant need to be kept within a given tolerance band by manipulating the turbine discharges of each power plant. Moreover, large variations in the turbine discharges need to be avoided. Application of a centralized MPC to the river cascade is computationally infeasible due to the large scale of the cascade. Therefore, a decentralized MPC scheme is developed, in which the cascade is divided into smaller subsystems and each subsystem is controlled with a local MPC scheme. In decentralized MPC, the computational requirements decrease, however, the performance might deteriorate due to the lack of coordination. In order to decrease the performance deterioration, a downstream communication is provided between the local MPC schemes: each local MPC communicates the predictions of its outflow to the downstream local MPC. Through simulations we show that employing the downstream communication strategy is sufficient to prevent a significant performance deterioration in decentralized MPC.


Type of Publication:

(03)Ph.D. Thesis

M. Morari

File Download:

Request a copy of this publication.
(Uses JavaScript)
% Autogenerated BibTeX entry
@PhDThesis { Xxx:2009:IFA_3513,
    author={A. Sahin},
    title={{Model Predictive Control for Large Scale Systems: Two
	  Industrial Applications}},
Permanent link