Simulation models for mineral processing plants

"Simulation is the process of designing a computerized model, of a system or procedure, for the purpose of understanding its behaviour and developing strategies to control its operation. Simulation models are rapidly becoming an effective tool for mineral processing plants. Powerful software and fast hardware are more readily available than a basic understanding of the theory of mathematical probability and applied statistics that process simulation models demand.Simple spreadsheet software can be used to develop simulation models that take into account any set of conditions at a mineral processing plant. In addition to an unbiased estimate for each variable in the set on which two-and three-product formulas are based, reliable variance estimates for all variables are required. How to estimate variances, effectively and at the lowest possible cost, and how to compute variances for simple functions such as the metal content of a quantity of ore or concentrate, or more complex functions such as the percentage recovery at a mineral processing plant, are the key to useful simulation models.IntroductionSimulation models for systems and procedures have become one of the most powerful and effective applications for computers( 2.6). Simulations are applied to a wide range of scientific and engineering disciplines. Generally, a model is designed to investigate the effects of changes in and interactions between a set of variables before a system is in operation or a procedure is implemented, by simulating weeks, months, and even years of operation, in a few minutes of computer time. Along with practical and useful applications always comeabuse and misuse. For a simulation process cannot be more reliable than the methodology that was applied to design the model, to estimate variables and variances, to check performance, and to verify validity. A mineral processing plant is a dynamic system that changes continuously as a function of time but within probabilistic constraints that can be estimated by applying statistical tools and techniques to the variables and variances that interact in the system.A model should be designed to simulate not only the set of variables but also the variance for each variable. Models in which constant variances are used to generate normally distributed random numbers, are less realistic than models in which the variances are simulated first, and then used to simulate the variables."