Abstract:
Dense medium separation (DMS) plants are typically used to beneficiate run-of-mine (ROM) coal in coal metallurgy. These plants normally make use of a dense medium cyclone as the primary processing unit. Because of the deviations in the ROM quality, the production yield and quality become difficult to maintain. A control system could benefit such operations to maintain and increase product throughput and quality. There are many different methods for developing a control system in a metallurgical operation; however, what is most fundamental is the use of a mathematical model to design a controller. For this reason, a first principle dynamic mathematical model has been developed for a DMS circuit. Each unit operation is modelled individually, then integrated together to form the complete system. The developed DMS circuit dynamic model is then used to simulate the process. It is also found that most models developed for DMS operations typically make use of steady-sate analysis and that very little literature is available on dynamic models of this kind. Difficulties that arise when validating a model in metallurgical processes are insufficient measurement points or the challenges in measuring certain variables, such as physical properties (e.g. particle size) or chemical components (e.g. ash percentage). This paper also explains how the Runge-Kutta approximation can be used in simulating DMS unit processes with intermediate online measurements that may be available. This can ultimately assist in verifying the accuracy of the simulation. One of the other problems that can occur when developing models from first principles is the estimation of model parameters. Specifically when non-linear state-space relationships are developed, one must ensure that there is a unique solution for the parameters in question. A method employing parameter identifiability is also presented in this dissertation to illustrate its use. In addition the process of estimating parameters is explained and illustrated. Copyright
