Contribution au développement d'une méthode de contrôle des procédés dans une usine de bouletage

Abstract

This thesis, a collaborative effort between École de technologie supérieure and ArcelorMittal Company, presents the development of a methodology for monitoring and quality control of multivariable industrial production processes. This innovation research mandate was developed at ArcelorMittal Exploitation Minière (AMEM) pellet plant in Port-Cartier (Quebec, Canada). With this undertaking, ArcelorMittal is striving to maintain its world class level of excellence and continues to pursue initiatives that can augment its competitive advantage worldwide. The plant`s gravimetric classification process was retained as a prototype and development laboratory due to its effect on the company`s competitiveness and its impact on subsequent steps leading to final production of iron oxide pellets. Concretely, the development of this expertise in process control and in situ monitoring will establish a firm basic knowledge in the fields of complex system physical modeling, data reconciliation, statistical observers, multivariate command and quality control using real-time monitoring of the desirability function. The hydraulic classifier is mathematically modeled. Using planned disturbances on the production line, an identification procedure was established to provide empirical estimations of the model`s structural parameters. A new sampling campaign and a previously unpublished data collection and consolidation Policy were implemented plant-wide. Access to these invaluable data sources has enabled the establishment of new thresholds that govern the production process and its control. Finally, as a substitute for the traditional quality control process, we have implemented a new strategy based on the use of the desirability function. Our innovation is not in using this function as an indicator of overall (economic) satisfaction in the production process, but rather in proposing it as an “observer” of the system`s state. The first implementation steps have already demonstrated the method`s feasibility as well as other numerous industrial impacts on production processes within the company. Namely, the emergence of the economical aspect as a strategic variable that assures better governance of production processes where quality variables present strategic issues.

Date	20 Oct 2014
Original language	French
Awarding Institution	École de technologie supérieure
Supervisor	Antoine Tahan (Supervisor) & Michel Ruel (Co-supervisor)