Ton Backx

Missing point estimation in models described by proper orthogonal decomposition

This paper presents a new method of missing point estimation (MPE) to derive efficient reduced-order models for large-scale parameter-varying systems. Such systems often result from the discretization of nonlinear partial differential equations. A projection-based model reduction framework is used where projection spaces are inferred from proper orthogonal decompositions of data-dependent correlation operators. The key contribution of the MPE method is to perform online computations efficiently by computing Galerkin projections over a restricted subset of the spatial domain. Quantitative criteria for optimally selecting such a spatial subset are proposed and the resulting optimization problem is solved using an efficient heuristic method. The effectiveness of the MPE method is demonstrated by applying it to a nonlinear computational fluid dynamic model of an industrial glass furnace. For this example, the Galerkin projection can be computed using only 25% of the spatial grid points without compromising the accuracy of the reduced model.

Development of a Model-Based Controller for a Three-Way Catalytic Converter

The performance of a three-way catalytic converter under transient operation can be improved by controlling the level of oxygen stored on ceria at some optimal level. A model-based controller with the model estimating the level of ceria coverage by oxygen, can achieve this goal. A simple, dynamic model is based on step responses of the converter and and is used to train the controller off-line. The controller is a neuro-fuzzy approximation of a model predictive controller. Thus, it retains a high performance while being less computationally involving. The system performance has been experimentally tested by a specially designed, highly transient test cycle.

Integration of Model Predictive Control and Optimization of Processes: Enabling Technology for Market Driven Process Operation

Abstract This paper discusses changes in the market of chemical processing industries and its consequences for process operation. The problem of continuously declining capital productivity is tackled, which has been faced by the chemical processing industries during the past decades. In the paper a motivation is given on how new technologies in the area of model predictive control and dynamic process optimization can contribute to improve business performance of the chemical processing industries in the changed markets. These new technologies have to support dynamic operation of plants at intended dynamics. Extensions and changes required to technologies that are currently applied in the oil refining industries and in some petrochemical industry applications are discussed. These extensions and changes are necessary to meet the specific requirements of the chemical processing industries.

A high performance model predictive controller:: application on a polyethylene gas phase reactor

Abstract This paper describes the development of a new model predictive control technology INCA ® that enables a high performance demand driven operation in the chemical process industry. The technology sustains optimal grade changes, maintains tight quality control and leads to low application development and implementation costs. An application on a polyethylene gasphase reactor is discussed.

Vehicle Application of Model-Based Catalyst Control

Abstract Standard stoichiometric control of a three-way catalytic converter is often not sufficient to meet new, very stringent, emission standards. This paper presents vehicle based testing of a model-based controller that takes into account dynamics of the catalytic converter. By applying such a controller higher efficiency of the catalytic converter under dynamic operation is achieved, which ultimately leads to lower emissions and a possibility to reduce the cost of the system.

Plantwide Economical Dynamic Optimization: Application of a Borealis Borstar Process Model

Abstract A novel plantwide dynamic optimizer PathFinder has been applied to a dynamic model of a Borealis Borstar process. PathFinder optimizes dynamic paths subject to a merely economical criterion. Introduction of process constraints allows for a gradual migration from the currently used transition towards a more optimal transition. Special care has been taken to integrate the optimizer with on-line control tools. The results show a significant improvement in added value during a grade transition.

Scheduling of energy flows for parallel batch processes using max-plus systems

A common approach to increase reliability and efficiency of manufacturing systems is optimal scheduling. In this paper, we consider scheduling problems for a specific type of systems where batch processes are running in parallel. The goal is to schedule the interaction between the parallel processes, in the form of energy flows, to minimize a desired objective. We make use of a max-plus approach in the formulation of the scheduling problem and use this formulation in the optimization subsequently. This framework has been used to formulate and optimize a case study for the production of calcium silicate stones.

Optimal Trajectories for Grade Change Control: Application on a Polyethylene Gas Phase Reactor

Abstract This paper describes the development of a Grade Transition optimizer PathFinder that enables a high performance demand driven operation in the chemical process industry. The technology makes use of a dynamical rigorous model based optimization of an economic criterion along a grade change trajectory. An application on a large grade slate for a polyethylene gasphase reactor is discussed.

Combining First Principles Dynamic Modelling and Plant Identification for Design of an Industrial MPC Application to a Polymer Reactor

Abstract This paper presents the design ot model predictive control systems on the basis ot combined use of first principles dynamic simulation models and process identification techniques to develop the models ultimately applied in the control system. As an example the application to a fluidised bed gas phase polymerisation reactor is discussed. The use of both modelling approaches is motivated from the requirements on the models imposed by the controller at one hand and the restrictions in process operation related to testing at the other hand. Aspects related to moving from the simulation environment to the plant environment to bridge the gap between true polymer reactor dynamics and the dynamics represented by the simulation model are discussed. Suggestions are given for further improvement of the approach.

Economically Optimal Grade Change Trajectories: Application on a Dow Polystyrene Process Model

Abstract A novel dynamic optimizer PathFinder has been applied to a dynamic model of a Dow polystyrene production facility at Tessenderlo, Belgium. PathFinder optimizes grade transitions subject to an economic cost function. Introduction of process constraints allows for a gradual migration from the currently used transition towards a more optimal transition. The results show a significant improvement in added value during a grade transition.