Efficient modeling and control of crushing processes in minerals processing
Modeling and simulation is a tool to explore and increase the understanding of a phenomenon. This thesis focuses on developing models of crushers and equipment used in the mining industry. Specifically, the focus is on a branch of modeling called time dynamic modeling which is a model that gives an output as a function of time.
The work is divided into three areas: physical modeling, control modeling, and circuit modeling. Physical modeling deals with how to develop high fidelity unit models of equipment, in this thesis, a model of a jaw crusher and of an HPGR are presented. These models are aimed to be predictive and should predict the process variables under a specific set of operating conditions. The models are developed with the process parameters that are used in the physical unit, in the case of the HPGR, roller speed, and hydraulic pressure. The parameters within the models are parameters with units and have real physical meaning; for example, a dimension of the machine.
The topic of control modeling focuses on how to apply the knowledge from modeling in the control domain to improve operations. An example of setting up a model predictive controller and using it to control a crushing circuit simulation is demonstrated. Model predictive control is an optimal control strategy that can be used to drive the circuit towards a specific goal. As the demand is increased on the mining companies to perform better these types of controllers and operation improving actions are important. This thesis aims to target some of the challenges involved in improving plant operation and control.
Within circuit modeling, a broader perspective is taken to study the operations of an entire circuit or plant. The study presented in this thesis focuses on how sensitive a plant is to variations and how the plant design itself will affect the plant's ability to cope with variations. The approach has been to simulate faster and to use less complex models many times to determine limits and ranges. The method shows potential to understand a circuit better before it is built.
The outcome of the research is a better understanding of how to model machinery, such as the HPGR and the jaw crusher. By developing high fidelity models, insights are gained on how to move between the different modeling domains. The knowledge is useful for studies of circuits, and how to set up optimal controllers. Especially controllers that require models of a specific type or models that have to be fast to simulate.