The focus of this thesis is on building models by utilizing process information: from data, from our knowledge of physics, or both. The closer the model approximates reality, the better is the expected performance in forecasting, soft-sensing, process monitoring, optimization and advanced process control. In the domain of batch and continuous manufacturing, quality models can help in ensuring tightly controlled product quality, having safe and reliable operating conditions and reducing production/operation costs. To this end, first a parallel grey box model was built which makes use of a mechanistic model, and a subspace identification model for modeling a batch poly methyl methacrylate (PMMA) polymerisation process. The efficacy of such a parallel hybrid model in the context of a control problem was illustrated thereafter for reducing the volume of fines. Real-time implementation of models in many cases demand the model to be tractable and simple enough, and thus the parallel hybrid model was next adapted to have a linear representation, and then used for control computations. While the parallel hybrid modelling strategy shows great advantages in many applications, there can be other avenues of using fundamental process knowledge in conjunction with historical data. In one such approach, a unique way of adding mechanistic knowledge to improve the estimation ability of PLS models was proposed. The predictor matrix of PLS was augmented with additional trajectory information coming strategically from a mechanistic model. This augmented model was used as a soft-sensor to estimate batch end quality for a seeded batch crystallizer process. In a collaborative work with an industrial partner focusing on estimating important variables of a hydroprocessing unit, an operational data based input-output model was chosen as the right fit in the absence of available mechanistic knowledge. The usefulness of linear dynamic modeling tools for such applications was demonstrated. / Thesis / Doctor of Philosophy (PhD)

A run-of-mine (ROM) ore milling circuit is primarily used to grind incoming ore containing precious metals to a powder fine enough to liberate the valuable minerals contained therein. The ground ore has a product particle size specification that is set by the downstream separation unit. A ROM ore milling circuit typically consists of a mill, sump and classifier (most commonly a hydrocyclone). These circuits are difficult to control because of unmeasurable process outputs, non-linearities, time delays, large unmeasured disturbances and complex models with modelling uncertainties. The ROM ore milling circuit should be controlled to meet the final product quality specification, but throughput should also be maximised. This further complicates ROM ore grinding mill circuit control, since an inverse non-linear relationship exists between the quality and throughput. ROM ore grinding mill circuit control is constantly evolving to find the best control method with peripheral tools to control the plant. Although many studies have been conducted, more are continually undertaken, since the controller designs are usually based on various assumptions and the required measurements in the grinding mill circuits are often unavailable. / To improve controller performance, many studies investigated the inclusion of additional manipulated variables (MVs) in the controller formulation to help control process disturbances, or to provide some form of functional control. Model predictive control (MPC) is considered one of the best advanced process control (APC) techniques and linear MPC controllers have been implemented on grinding mill circuits, while various other advanced controllers have been investigated and tested in simulation. Because of the complexity of grinding mill circuits non-linear MPC (NMPC) controllers have achieved better results in simulations where a wider operating region is required. In the search for additional MVs some researchers have considered including the discrete dynamics as part of the controller formulation instead of segregating them from the APC or base-layer controllers. The discrete dynamics are typically controlled using a layered approach. Discrete dynamics are on/off elements and in the case of a closed-loop grinding mill circuit the discrete elements can be on/off activation variables for feed conveyor belts to select which stockpile is used, selecting whether a secondary grinding stage should be active or not, and switching hydrocyclones in a hydrocyclone cluster. Discrete dynamics are added directly to the APC controllers by using hybrid model predictive control (HMPC). HMPC controllers have been designed for grinding mill circuits, but none of them has considered the switching of hydrocyclones as an additional MV and they only include linear dynamics for the continuous elements. This study addresses this gap by implementing a hybrid NMPC (HNMPC) controller that can switch the hydrocyclones in a cluster. / A commonly used continuous-time grinding mill circuit model with one hydrocyclone is adapted to contain a cluster of hydrocyclones, resulting in a hybrid model. The model parameters are refitted to ensure that the initial design steady-state conditions for the model are still valid with the cluster. The novel contribution of this research is the design of a HNMPC controller using a cluster of hydrocyclones as an additional MV. The HNMPC controller is formulated using the complete nonlinear hybrid model and a genetic algorithm (GA) as the solver. An NMPC controller is also designed and implemented as the base case controller in order to evaluate the HNMPC controller’s performance. To further illustrate the functional control benefits of including the hydrocyclone cluster as an MV, a linear optimisation objective was added to the HNMPC to increase the grinding circuit throughput, while maintaining the quality specification. The results show that the HNMPC controller outperforms the NMPC one in terms of setpoint tracking, disturbance rejection, and process optimisation objectives. The GA is shown to be a good solver for HNMPC, resulting in a robust controller that can still control the plant even when state noise is added to the simulation. / Dissertation (MEng)--University of Pretoria, 2018. / National Research Foundation (DAAD-NRF) / Electrical, Electronic and Computer Engineering / MEng / Unrestricted

In this thesis, we discussed various possibilities of integrating scheduling decisions with information and constraints from Advanced Process Control (APC) systems in semiconductor Manufacturing. In this context, important questions were opened regarding the benefits of integrating scheduling and APC. An overview on processes, scheduling and Advanced Process Control in semiconductor manufacturing was done, where a description of semiconductor manufacturing processes is given. Two of the proposed problems that result from integrating bith systems were studied and analyzed, they are :Problem of Scheduling with Time Constraints (PTC) and Problem of Scheduling with Equipement health Factor (PEHF). PTC and PEHF have multicriteria objective functions.PTC aims at scheduling job in families on non-identical parallel machines with setup times and time constraints.Non-identical machines mean that not all miachines can (are qualified to) process all types of job families. Time constraints are inspired from APC needs, for which APC control loops must be regularly fed with information from metrology operations (inspection) within a time interval (threshold). The objective is to schedule job families on machines while minimizing the sum of completion times and the losses in machine qualifications.Moreover, PEHF was defined which is an extension of PTC where scheduling takes into account the equipement Health Factors (EHF). EHF is an indicator on the state of a machine. Scheduling is now done by considering a yield resulting from an assignment of a job to a machine and this yield is defined as a function of machine state and job state.

Increasing worldwide demand for petroleum motivates greater efficiency, safety, and environmental responsibility in upstream oil and gas processes. The objective of this research is to improve these areas with advanced control methods. This work develops the integration of optimal control methods including model predictive control, moving horizon estimation, high fidelity simulators, and switched control techniques applied to subsea riser slugging and managed pressure drilling. A subsea riser slugging model predictive controller eliminates persistent offset and decreases settling time by 5% compared to a traditional PID controller. A sensitivity analysis shows the effect of riser base pressure sensor location on controller response. A review of current crude oil pipeline wax deposition prevention, monitoring, and remediation techniques is given. Also, industrially relevant control model parameter estimation techniques are reviewed and heuristics are developed for gain and time constant estimates for single input/single output systems. The analysis indicates that overestimated controller gain and underestimated controller time constant leads to better controller performance under model parameter uncertainty. An online method for giving statistical significance to control model parameter estimates is presented. Additionally, basic and advanced switched model predictive control schemes are presented. Both algorithms use control models of varying fidelity: a high fidelity process model, a reduced order nonlinear model, and a linear empirical model. The basic switched structure introduces a method for bumpless switching between control models in a predetermined switching order. The advanced switched controller builds on the basic controller; however, instead of a predetermined switching sequence, the advanced algorithm uses the linear empirical controller when possible. When controller performance becomes unacceptable, the algorithm implements the low order model to control the process while the high fidelity model generates simulated data which is used to estimate the empirical model parameters. Once this online model identification process is complete, the controller reinstates the empirical model to control the process. This control framework allows the more accurate, yet computationally expensive, predictive capabilities of the high fidelity simulator to be incorporated into the locally accurate linear empirical model while still maintaining convergence guarantees.

Dans cette thèse, nous avons examiné différentes possibilités d'intégration des décisions d'ordonnancement avec des informations provenant de systèmes avancés des contrôles des procédés dans la fabrication de semi-conducteurs. Nous avons développé des idées d'intégration et défini des nouveaux problèmes d'ordonnancement originales : Problème d'ordonnancement avec des contraintes de temps (PTC) et problème d'ordonnancement avec l'état de santé des équipement (PEHF). PTC et PEHF ont des fonctions objectives multicritères.PTC est un problème d'ordonnancement des familles de jobs sur des machines parallèles non identiques en tenant compte des temps de setup et des contraintes de temps. Les machines non identiques signifient que toutes les machines ne peuvent pas traités (qualifiés) tous les types de familles d'emplois. Les contraintes de temps nommés aussi Thresholds sont inspirées des besoins de l'APC. Elle est liée à l'alimentation régulière des boucles de contrôle de l'APC. L'objectif est de minimiser la somme des dates de fin et les pertes de qualification des machines lorsqu'une famille de jobs n'est pas ordonnancée sur la machine donnée avant un seuil de temps donné.D'autre part, PEHF est une extension de PTC. Il consiste d'intégrer les indices de santé des équipements (EHF). EHF est un indicateur associé à l'équipement qui donne l'état de la. L'objectif est d'ordonnancer des tâches de familles de jobs différents sur les machines tout en minimisant la somme des temps d'achèvement, les pertes de qualification de la machine et d'optimiser un rendement attendu. Ce rendement est défini comme une fonction d'EDH et de la criticité de jobs considérés. / In this thesis, we discussed various possibilities of integrating scheduling decisions with information and constraints from Advanced Process Control (APC) systems in semiconductor Manufacturing. In this context, important questions were opened regarding the benefits of integrating scheduling and APC. An overview on processes, scheduling and Advanced Process Control in semiconductor manufacturing was done, where a description of semiconductor manufacturing processes is given. Two of the proposed problems that result from integrating bith systems were studied and analyzed, they are :Problem of Scheduling with Time Constraints (PTC) and Problem of Scheduling with Equipement health Factor (PEHF). PTC and PEHF have multicriteria objective functions.PTC aims at scheduling job in families on non-identical parallel machines with setup times and time constraints.Non-identical machines mean that not all miachines can (are qualified to) process all types of job families. Time constraints are inspired from APC needs, for which APC control loops must be regularly fed with information from metrology operations (inspection) within a time interval (threshold). The objective is to schedule job families on machines while minimizing the sum of completion times and the losses in machine qualifications.Moreover, PEHF was defined which is an extension of PTC where scheduling takes into account the equipement Health Factors (EHF). EHF is an indicator on the state of a machine. Scheduling is now done by considering a yield resulting from an assignment of a job to a machine and this yield is defined as a function of machine state and job state.