Performance Evaluation of Basic Flotation Kinetic Models Using Advanced Statistical Techniques

ABSTRACT It is well evident that mathematical models play a key role in any time-dependent kinetic process; Froth flotation is a process resulting from the interaction of air bubbles, mineral surface, and chemical reagents. Flotation is a kinetic phenomenon where residence time is usually studied to reckon the flotation performance in terms of grade and recovery. The use of mathematical models is not only limited to performance assessment but they also find their application in designing and optimization of flotation circuits. Hence, it is justifying why froth flotation is being studied so intensively in mathematical terms. Literature review reveals that many models co-relating recovery (y), time (t), rate constant (a), and ultimate recovery of component (R) have been developed in the past; however, the practical applicability and correct methodology to select these models still needs to be studied. In the present investigation seven models were evaluated by comparing the model output with experimental data obtained from coal flotation experiments at different conditions of chemical reagents (collector and frother dosage) and pulp density. The paper also describes the application of classical statistical tools, such as least square method, sum of square error test, standard error test, F-test, test for Jarquebera, test statistic for skewness and kurtosis, and AIC value test, to assess the deviation of model output from experimental data. Results indicate that most of the studied models, if properly calibrated, work well for coal flotation data set. No single model was found to be uniformly the best for all the data set studies. The models were ranked according to their scores obtained on performing statistical tests on them.