Abstract: This study focuses on the drying process of pasta in a convectional air drying oven. A model predictive control algorithm is designed, based on a dynamic model describing the mass transfer between pasta and the surrounding air. The multiple input single output control algorithm minimizes a cost function over a prediction horizon which represents the deviation of the process from a desired reference track, thereby incorporating the working limitations of the oven. The performance of the controller and the influence of the prediction horizon are examined in this paper, showing minimal deviation between process and reference track.

Abstract: This paper aims at promoting cooperation between drying and control communities in the future. Indeed, with a review of 71 relevant publications all dealing with control aspects in drying, this paper shows that the use of control tools really started to emerge in drying applications only since 1979. In a second phase started around 1998, new trends based on more advanced concepts have also appeared in drying control. This paper clearly shows that control in drying is more and more a reality and that many opportunities exist to enhance industrial performance via efficient control of the operation.

Abstract: This paper is intended to show the comparison of two closed-loop controllers performances for a catalytic reverse flow reactor (RFR): a linear quadratic regulator (LQR) and a model predictive control (MPC) strategy that both use a high-gain observer. The aim of the RFR is to reduce, by catalytic reaction, the amount of volatile organic compounds (VOCs) released in the atmosphere. The particularity of this process is that the gas flow inside the reactor is periodically reversed in order to trap the heat released during the reaction inside the process. Very few papers are dealing with the control of the RFR. The new multivariable optimal control issue tackled here is to confine the hot spot temperature within two temperature limits, in order to ensure complete conversion of the pollutant and to prevent catalyst overheating, while optimizing the process yield and the consumption of electrical power. Both control laws aim to optimize the tuning of the dilution and the internal electric heating. The model considered here for control of the RFR is obtained from a countercurrent pseudo-homogeneous partial differential equation (PDE) model. The feed concentration (acting as an important input disturbance) and the temperature profile in the RFR are estimated on-line using a high-gain observer based on three temperatures measurements. In order to maintain the safe use of the reactor, the estimate state is injected in the LQR whereas the MPC is based on the estimated input disturbance. Simulation results allow comparing performances of the LQR and the MPC.

Abstract: A neural network based strategy for detection of feedstock variations in a continuous pulp digester is presented. A feedforward two-layer perceptron network is trained to detect and isolate unmeasured variations in the feedstock. Training and validation data sets are generated using a rigorous first principles model. The most important issue discussed here is the design of the data set required to train the artificial neural network. Efficiency and limitation of such an approach are demonstrated using simulations.

Abstract: This paper is devoted to the multiple input multiple output (MIMO) model predictive control (MPC) of a catalytic reverse flow reactor (RFR). The RFR aims to reduce, by catalytic reaction, the amount of volatile organic compounds (VOCs) released in the atmosphere. The peculiarity of this process is that the gas flow inside the reactor is periodically reversed in order to trap the heat released during the reaction. The objective of this paper is to propose a solution to avoid the limitation seen in a previous study for the single input single output (SISO) control case. It was dealing with the impossibility to avoid degradation of the catalytic elements due to the excessive heating induced by the exothermic reaction. In order to overcome this issue, a ratio of the inlet (cold) gas flow is now bypassed into the central zone. This allows introducing a second manipulated variable: the dilution rate. The phenomenological model considered here for the MIMO MPC of the RFR is obtained from a rigorous first principles modeling. The resulting accurate nonlinear partial differential equation (PDE) model can be a drawback for the model based controller implementation. To overcome this issue, we use a MPC strategy previously developed: it combines a two phase approximation of the PDE model in an internal model control (IMC) structure. This strategy allows using a less accurate model and a less time-consuming control algorithm. Finally, efficiency of the control approach is shown in simulation.

Abstract: A general framework for a partial differential equation (PDE) model predictive control problem is formulated. A first principle model of the system, described by a semi-linear PDE system with boundary control, is employed in a model predictive control (MPC) framework. Here, the aim is to determine, off-line (i.e. without process measurement), the theoretical optimal behavior of the process that will be used during on-line MPC. Input and output constraints are handled in the optimization task using a non-linear programming method. This strategy is evaluated for the optimization of processing temperatures during the manufacture of thick-sectioned polymer composite laminates. The off-line optimization task consists of determining the optimal temperature profile, otherwise known as the cure cycle. Moreover, for this particular process, the existence of a feasible constrained optimization problem is discussed through the design of a constraint bound.

Abstract: This paper deals with the experimental control of an infrared drying process of a water based epoxy-amine painting. This approach is based on a unidirectional diffusional modelling of infrared drying phenomena where both heat and mass transfers under shrinkage conditions are accounted for. The control problem is concerned with the tracking of any given trajectory for one of the characteristics (i.e. the temperature or the mean water content) during the drying cycle. This is solved using the well-known model predictive control framework where the non-linear diffusional model is directly used in the control formulation. Experimental results show the efficiency of the trajectory tracking. This method can be extended for more general constrained control problem.

Abstract: Piece maitresse dans l’approche de commande predictive non lineaire sous contraintes, l’algorithme d’optimisation doit determiner une commande satisfaisant les contraintes et ceci le plus rapidement possible en vue d’une application en temps reel. On montre ici le lien explicite entre la faisabilite de la commande predictive et la convergence de l’algorithme d’optimisation. Une condition suffisante est proposee pour garantir la convergence de l’algorithme d’optimisation et de ce fait la faisabilite de la commande predictive non lineaire sous contraintes. Une application en temps reel illustre cette approche.

Abstract: It is now recognized that model predictive control (MPC) is an interesting alternative for real-time control of industrial processes. In the meantime, some problems do still remain in progress: for theoretical aspects, the a priori guarantee of the stability and for the practical aspects, the guarantee of sufficient time to solve to optimization problem at each sampled time positions. In this paper, we propose a global method that aims to reduce the on-line calculation time due to the PDE model based optimization task resolution. It is addressed for a particular class of systems not very often studied in this context: systems described by partial differential equations (PDEs) which are, in the present case, nonlinear and parabolic. In order to decrease the computational burden, the nonlinear PDE system is solved off-line. Then, a linearized PDE model around the previous off-line behavior is used to find the optimal variations for the on-line predictive control. The real-time control application given is concerned with a infrared drying process of painting film.

Abstract: This paper deals with the control of a catalytic reverse flow reactor. The aim of this process is to reduce, by catalytic reaction, the amount of volatile organic compounds (VOCs) released in the atmosphere. The peculiarity of this process is that the gas flow inside the reactor is periodically reversed in order to trap the heat released during the reaction inside the process. This allows use of the reactor in heat saving mode. The goal of this work is to provide a model predictive control (MPC) framework to significantly enhance the poor overall performance currently obtained through the actual control strategy. It is directly addressed for the nonlinear parabolic partial differential equations (PDEs) that describe the catalytic reverse flow reactor. In the context of the application of MPC to this particular distributed parameter system, we propose a method that aims to reduce the on-line computation time needed by the control algorithm. The nonlinear model is linearized around a given operating trajectory to obtain the model to be solved on-line in the approach. MPC strategy combined with internal model control (IMC) structure allows to use less accurate and less time consuming control algorithm. The efficiency of the method is shown in simulation for this SISO system.

Abstract: This paper presents a detailed fundamental model of a continuous pulp digester developed for simulation and control. Most modeling efforts in the past have either emphasized reaction kinetics and energy transfer assuming a prescribed flow behavior, or modeled digester hydraulics with simplified reaction kinetics. In this work, we augment the well-known extended Purdue model (Wisnewski et al., 1997) by incorporating axial momentum transport, an immediate consequence of which is ability to simulate chip level and its impact on the Kappa number profile. An essential contribution is the ability to model chip compaction. Interaction between digester hydraulics and kinetics is illustrated via simulation examples. Examples include scenarios that show incipient plugging of the digester vessel. The current work also explores and illustrates the use of higher order finite-difference methods to reduce numerical dissipation and dispersion. Ability to describe chip velocities enables rigorous tracking of transition front within the digester during feedstock grade transitions. Operational difficulties encountered during hardwood to softwood transition are explained using the model.

Abstract: This paper deals with the boundary control of distributed parameter systems. We give, for non-academical example, an explicit form of stability and feasability conditions to an output feedback boundary control, and we generalize the internal model control (IMC) structure to infinite dimensional system. The framework is directly presented via a MIMO heat exchanger pilot process example. Some simulations results of asymptotic traking with a variable set point are given.

Abstract: L’objectif de ce travail est de montrer la robustesse d’une loi de commande prédictive en temps réel d’un procédé de cuisson de peintures en poudre sur un support plan et métallique sous infrarouge. En dépit d’un modèle de connaissance non parfait dû à des propriétés radiatives (statique et dynamique) de l’émetteur complexes à modéliser et à mesurer en ligne, cette commande en boucle fermée permet de contrôler expérimentalement l’état du feuil de peinture qui est étroitement lié à la maîtrise de l’évolution des températures pendant la cuisson, en particulier au sein de la couche de peinture.

Abstract: Since 1998 [3], the yearly mean publication rate of papers using a first principle model for control has been multiplied by 11! This increase is quiet similar to the increase of the control papers in drying based on optimal control strategies, like MPC. With 60 000 products dried and 100 dryer types used worldwide [1], a real potential exists. With control tools like MPC, it is expected that the industrial drying operation will continue to improve its energy efficiency while enhancing product quality and reducing the negative environmental impact of dryers.

Abstract: Cette etude presente le developpement d’une commande de procede de cuisson de peinture en poudre sous infrarouge en utilisant une structure de commande predictive a modele interne. Le modele de cuisson, issu d’etudes precedentes, est utilise pour predire le comportement de la peinture en reponse a differentes intensites radiatives imposees a la surface du film, la cuisson etant suivie au travers de la mesure de temperature de l’ensemble support + peinture. Cette commande a ete implementee sur un four de cuisson de laboratoire et testee pour des poursuites de trajectoire en temperature.

Abstract: This paper presents a parametric sensitivity based controller for on line optimal model parameter identification using constrained closed loop control tools and an observer. In optimal input design problem, analytical solution exists for few particular cases based on a relatively simple model. The approach proposed here may be used for a process based on a continuous model in the time domain, with two assumptions on the observability and the general structure of the model. The new proposed approach is to solve a model predictive control problem coupled with an on line process parameter estimation at each time using an observer. A dynamic parametric sensitivity model (derived from the process model) is also used on line to get the parametric sensitivity that has to be optimized. Both optimal input and estimated model parameter are therefore obtained on line. The case study presented here is a powder coating curing process where the main thermal parameter to identify influences the powder curing. First simulation results show here the efficiency of the approach in the control software (MPC@CB) developed under Matlab.

Abstract: This paper presents the application of a model based predictive control strategy for the primary stage of the freeze drying process, which has not been tackled until now. A model predictive control framework is provided to minimize the sublimation time. The problem is directly addressed for the non linear distributed parameters system that describes the dynamic of the process. The mathematical model takes in account the main phenomena, including the heat and mass transfer in both the dried and frozen layers, and the moving sublimation front. The obtained results show the efficiency of the control software developed (MPC@CB) under Matlab. The MPC@CB based on a modified levenberg-marquardt algorithm allows to control a continuous process in the open or closed loop and to find the optimal constrained control.

Abstract: In emulsion polymerization processes, the surfactant concentration plays an impor- tant role in the latex stability, reaction kinetics and particle size distribution (PSD). Controlling the free surfactant concentration in the aqueous phase ensures the stability of the latex and the control of the micellar nucleation rate. The PSD is determined by particle nucleation, growth and stability which are highly nonlinear behaviors. The PSD affects the polymer rheology, adhesion, optical properties and mechanical strength. This work deals with the model predictive control (MPC) of free surfactant concentration using the surfactant feed rate as a control variable. The used strategy is a global method that aims to reduce the on-line calculation time due to the partial differential equations (PDE) model based optimization task resolution. In order to decrease the computational burden, the nonlinear PDE system is solved off-line. Then, a linearized PDE model around the previous off-line behavior is used to find the optimal variations for the on-line predictive control.

Abstract: This paper deals with a reusable simulation computer code (MPC@CB ). The original program was developed under Matlab for single input single output (SISO) model predictive control (MPC) for any constrained optimization problem (trajectory tracking, processing time minimization...). The control structure is an adaptation of MPC with internal model control (IMC) structure. The algorithm was applied and validated for different processes. It was adapted in this work for multiple input multiple output (MIMO) constrained systems.

Abstract: This paper deals with the first principle model based predictive control of the primary drying stage of a freeze drying of solutions in vials. In the proposed control approach, any online control problem concerned with a constrained optimization during the primary drying stage may be stated. It is solved using a special model predictive control framework where the model is used in the controller formulation. Our first simulation results show here the efficiency of the control software developed (MPC@CB) under Matlab. MPC@CB may be easily used for any other process and any constrained control problem.

Abstract: This paper deals with the control of a powder coating radiative curing process by infrared flow. It is based on a unidirectional dynamic modelling where both heat transfert and cure phenomena are accounted for within the thicknessl. The control problem is the constrained optimization of the curing cycle. This is solved using a special model predictive control framework: it is designed such that the calculation time is smaller that the sampling time (a few seconds), in spite of the need to solve the non-linear partial differential equation based model involved during the online constrained optimization task resolution. Simulation results show here the efficiency of the control software developed (MPC@CB) under Matlab. MPC@CB may be easily used for any other constrained control problem.

Abstract: This paper aims to compare 2 optimal control laws for a catalytic reverse flow reactor (RFR), which topic has not often been tackled. The RFR aims to destruct volatile organic compounds (VOCs), which are noxious products and are the source of a lot of problems: pollution, acid rains, woods wasting, green house effect and health hazards. Due to the public regulations, the VOCs discharge in the atmosphere becomes therefore more and more limited. The RFR is characterized by a periodic cycle of treatment of the VOC flow, allowing trapping the heat of reaction inside the core of the RFR. This allows using the RFR in an autothermal mode. The control problem tackled here is to maintain the RFR is this mode, such that the RFR extinction is avoided, and such that the catalyst elements are not destroyed. Effects of the influencing stochastic variations of the inlet pollutant concentration (the input disturbance) have to be accounted for by the optimal constrained tuning of the dilution rate and the internal electric heating. Optimal closed-loop performances of a linear quadratic regulator (LQR) and a model predictive controller (MPC) are compared through simulations.

Abstract: Based on a review of 71 relevant publications all dealing with control aspects in drying, this paper aims to promote more works in common between drying and control communities. First, it is shown that the use of control tools really started to emerge in drying applications only since 1979. In a second phase started around 1998, new trends based on more advanced concepts have also appeared in drying control. It is underlined how control in drying is more and more a reality and that many opportunities exist to enhance industrial drying performance via efficient control of the operation.

Abstract: This paper is devoted to the MIMO control of the catalytic reverse flow reactor (RFR) which aims to reduce the amount of volatile organic compounds (VOCs) released in the atmosphere. The RFR is characterized by the periodic reversal of the gas flow that aims to trap the heat of reaction inside the RFR. The control issue is to confine the hot spot temperature inside an envelope (in order to ensure complete conversion of the pollutant and to prevent catalyst overheating) in spite of stochastic variations of the inlet pollutant concentration (the input disturbance). The manipulated variables (dilution rate and internal electric heating ) have to be optimized. Closed-loop performances of the LQR and the MPC are compared through simulations.

Abstract: Piece maitresse dans l’approche de commande predictive non lineaire sous contraintes, l’algorithme d’optimisation doit determiner une commande satisfaisant les contraintes et ceci le plus rapidement possible en vue d’une application en temps reel. On montre ici le lien explicite entre la faisabilite de la commande predictive et la convergence de l’algorithme d’optimisation. Une condition suffisante est proposee pour garantir la convergence de l’algorithme d’optimisation et de ce fait la faisabilite de la commande predictive non lineaire sous contraintes. Une application en temps reel illustre cette approche.

Abstract: This paper deals with the model predictive control of an infrared drying process of a water based epoxy-amine painting. During the drying cycle, the control problem is to optimize the use of the process under constraints. This approach is based on a nonlinear dynamic unidirectional diffusional model of the infrared drying phenomena where both heat and mass transfers under shrinkage conditions are accounted for. To validate our approach, simulation results presented here deal with the minimization of the processing time while accounting for a constraint specified on a difference of humidity inside the sample.

Abstract: This paper deals with the constrained optimal control of an infrared drying process of a water based epoxy-amine painting. This approach is based on a nonlinear dynamic unidirectional diffusional model of infrared drying phenomena where both heat and mass transfers under shrinkage conditions are accounted for. The control problem is to minimize the processing time while accounting for any constraints for some of the characteristics (i.e. the temperature and the humidity profile) during the drying cycle. This is solved using a model predictive control framework where the nonlinear diffusional model is directly used in the control formulation. Such advanced controller requires the use of a soft-sensor (observer) in order to reconstruct the entire humidity profile according to the infrared flow applied, the measured temperature and the model.

Abstract: This paper deals with the experimental control of an infrared drying process of a water based epoxy-amine painting. This approach is based on a unidirectional diffusional modeling of infrared drying phenomena where both heat and mass transfers under shrinkage conditions are accounted for. The control problem is concerned with the tracking of any given trajectory for one of the characteristics (i.e the temperature or the mean water content) during the drying cycle. This is solved using the well known model predictive control framework where the nonlinear diffusional model is directly used in the control formulation. Experimental results show the efficiency of the trajectory tracking. This method can be extended for more general constrained control problem.

Abstract: A strategy for detection of feedstock variations in a continuous pulp digester is presented. A Gaussian Radial Basis Function Neural Network is used to infer these unmeasured variations. The absence of plant data motivates the development of training set data. The efficiency and limitation of the approach is demonstrated with a first principles model.

Abstract: A trajectory tracking strategy is presented for processes described by a parabolic nonlinear distributed parameter model. A model-based predictive control approach, combined with an internal model control structure and a state estimation method is extended to this kind of process. On-line requirements such as computational time and constraint satisfaction are outlined and discussed. This method is then implemented for the control of a drying process and experimental results are discussed.

Abstract: Diagnostic strategies for Fault Detection and Isolation in a continuous pulp digester are presented. Several methodologies for Fault Detection and Isolation are compared for accuracy and real-time implementation potential. A Gross Error Detection methodology for biased and noisy measurements is initially examined. Next, a Gaussian Radial Basis Function neural network approach for detection of product quality related changes is also used. Changes in feedstock composition using an additional neural network approach is also developed. The efficiency and limitations of all methods are demonstrated using a first principles model.

Abstract: This paper presents a detailed fundamental model of a continuous pulp digester developed for simulation and control. Most modeling efforts in the past have either emphasized on reaction kinetics and energy transfer assuming a prescribed flow behavior, or modeled digester hydraulics with simplified reaction kinetics. In this work, we augment the well-known Purdue model by incorporating axial momentum transport, an immediate consequence of which is ability to simulate chip level and its impact on the Kappa number profile. Model performance is illustrated using simulation.

Abstract: This paper presents a detailed fundamental model of a continuous pulp digester developed for simulation and control. Most modeling efforts in the past have either emphasized on reaction kinetics and energy transfer assuming a prescribed flow behavior, or modeled digester hydraulics with simplified reaction kinetics. In this work, we augment the well-known Purdue model by incorporating axial momentum transport, an immediate consequence of which is ability to simulate chip level and its impact on the Kappa number profile. Model performance is illustrated using simulation. The current work also explores and illustrates the use of higher order finite difference methods to reduce numerical dissipation and dispersion.

Abstract: This paper deals with the model predictive control of processes. The new step is the use of a distributed parameter system instead of a lumped parameter system. The internal model control structure is also used to solve the trajectory tracking problem. The internal model is obtained from the linearization of the initial set of nonlinear partial differential equations about the desired trajectory. Finally, the control problem stated as a constrained optimization problem is solved by a control algorithm. Experimental results presented here show the efficiency of this control strategy.

Abstract: This work deals with a model-based approach to non-linear (NL) parabolic PDE control. We address two problems in the optimization and control of the autoclave lay-up curing process. The first is to find the best process trajectory corresponding to the shortest possible operating time. The second is to determine the process control required to track the optimized trajectory within the constraints of the system. In previous work, a simulation based approach was used called local criteria optimization (LCO). Here the former problem is stated as a constrained optimization problem. A predictive type approach with a penalty method is solved by the Levenberg-Marquardt (LM) algorithm. For the tracking problem, the well-known Internal Model Control (IMC) structure is used for the predictive control strategy. Finally, we show that a time-varying linear model around the optimal profile can be used for the on-line optimization.

Abstract: This paper deals with the nonlinear distributed parameter process control. The aim is to realize a predictive type control that minimize an optimization criterion under some constraints. The well-known internal model control (IMC) structure is used to make the predictive control strategy where the internal model is a time-varying linearized model around a tracking trajectory. This nonlinear distributed parameter system control is developped and presented through a practical distributed control application : an infrared drying process control.

Abstract: L’objectif principal de ce travail a ete de developper des solutions pour le contrôle de la distribution de la taille des particules (DTP) lors d’un procede de polymerisation en emulsion. La DTP affecte les proprietes rheologiques, et optiques du polymere ou encore sa stabilite. Pour ces raisons la DTP apparaît comme une caracteristique importante a contrôler. Le modele decrivant l’evolution de la DTP est regi par des equations aux derivees partielles (EDP) non lineaires. L’etude numerique du modele a permis de dresser un modele de commande. Deux types de strategie de commande predictive par modele interne ont ete definis, l’une basee sur un modele de commande non lineaire et l’autre basee sur un modele lineaire temps variant. La DTP etant difficilement mesurable en ligne, notre choix s’est porte vers d’autres mesures, les concentrations de tensioactif libre et de monomere. Il a ete possible de definir des profils de concentration afin d’obtenir un produit final avec une DTP specifique. Les resultats de simulations ont montre la faisabilite de cette approche.

Abstract: Ce travail se situe dans le cadre de la commande des systemes a parametres repartis regis par des equations aux derivees partielles non lineaires. On cherche ici a elargir l’utilisation de la strategie de commande predictive du cadre classique de systemes decrits par des equations differentielles ordinaires a celui où les systemes sont decrits par une classe d’equations non lineaires aux derivees partielles (EDP). L’interêt de l’utilisation de la strategie de commande predictive est de pouvoir resoudre divers problemes de conduite tout en tenant compte explicitement de diverses contraintes de fonctionnement. Nous avons choisi une demarche qui a consiste a considerer directement des cas reels issus du Genie des Procedes, où de nombreux modeles de connaissance sont decrits par des EDP fortement non lineaires : un procede de sechage de peinture par infrarouge (donnant lieu a une application reelle), un procede de fabrication de materiau composite et un procede de destruction de gaz polluants par catalyse. La particularite de ce travail concerne le fait de proposer une structure adaptee de commande predictive par modele interne, celui-ci etant determine autour d’une trajectoire soit predefinie, soit generee en tenant compte de diverses contraintes de fonctionnement. Dans le contexte de resolution du probleme d’optimisation pose par l’approche predictive, nous proposons un algorithme hors-ligne (generation de trajectoire) et un autre en ligne.