Development and modelling of a semi-batch flotation apparatus.

Thesis (M.Sc.Eng.)-University of KwaZulu-Natal, Durban, 2008.When designing or optimizing flotation circuits in mineral processing plants, it is necessary to have accurate values of the flotation kinetics to ensure the correct mass pulls and material balances on the plant. Previous studies have shown that rate constants measured by single cell batch testing can cause a shift in the recovery—grade curve. The shift in the recovery—grade curve is the result of poor separation in conventional laboratory flotation devices. This project involved the development and modelling of a flotation device that provides a better separation than a conventional batch flotation cell. The device is called a semi—batch flotation apparatus (SBFA) because it simulates the operations of a pilot plant in a laboratory environment. It also provides dynamic data which facilitates the evaluation of model parameters. The SBFA tested a synthetic ore made from limestone, talc and silica. The synthetic ore was used as it was economical and easy to analyze. The results from the SBFA were compared to results obtained from conventional batch flotation tests; by using recovery—grade curves to assess the degree of separation achieved from both devices. The SBFA separated the limestone from the gangue (silica and talc) much better than the batch tests. For instance the final grade for a concentrate obtained from a single cell batch test was 20 % limestone while the final grade for a concentrate obtained from the SBFA was between 40 % and 70 % limestone. The improvements in separation can be attributed to the multistage design of the SBFA which has a pulp recycle between the stages. A model has been developed for the SBFA. The model fitted the experimental data well with a correlation coefficient close to unity. The cumulative recoveries predicted from the SBFA model was compared to the actual cumulative recoveries, by using a global set of parameters (&2 and RMAX)- The investigation showed that the model had problems in fitting the data for the early periods of the experiments because of the complex interaction between the stages in the SBFA