Date of Award
Doctor of Philosophy (PhD)
Dr. M.C. de Malherbe
Professor R. Newcombe
An approach to study control problems in complex systems of interacting, mixed variable processes is presented. Discrete event simulation is combined with continuous simulation to provide a versatile method for system modelling and analysis. Provisions are made to include a human model into the model of a real system which allows an accurate reproduction of systems in which human control is an integral part. A special model display unit was built to provide feedback to the model operator, which can be readily connected to a hybrid computer. The display unit permits operation of the model by process people with no detailed knowledge of simulation which is an added advantage.
A copper smelter system was modelled and simulated on a hybrid computer which includes discrete as well as continuous processes and requires human control to integrate the process to achieve a maximum output. The system thus represents a good example for a variety of similar systems encountered in industry. Semi-empirical models of smelting processes were developed based on existing knowledge of the processes. Using the display unit the evolutionary development necessary for some models was possible.
The simulation model was calibrated and validated to accurately reproduce the real system behaviour. Based on simulation experiments on improved operating technique for controlling aspects of the smelting process was tested and an increase in production was predicted. This operating technique was tested in plant trials and adopted while the predicted production increase was verified.
The validated model was then used to develop heuristic decision and forecasting algorithms optimal control of model operations. Based on the realistic model diffrent decision strategies could be tested by observing their success in controlling the simulation model. The control algorithms were refined to give a satisfactory control of the simulation model, without the need of human intervention. The smelter model with the heuristic control can thus serve as a useful tool in developing an on-line control system for the real system, which could be used as a decision making aid to operating personnel.
Ströbele, Kurt Anton, "Modelling and Dynamic Optimisation of Mixed Variable Processes Using Hybrid Simulation Techniques" (1973). Open Access Dissertations and Theses. Paper 1027.