Property-Based Modelling at Fixed Chemistry: The use of a back-calculated induction time for predicting recoveries in flotation

Danoucaras, Anastasia (2011). Property-Based Modelling at Fixed Chemistry: The use of a back-calculated induction time for predicting recoveries in flotation PhD Thesis, Sustainable Minerals Institute, The University of Queensland.

       
Attached Files (Some files may be inaccessible until you login with your UQ eSpace credentials)
Name Description MIMEType Size Downloads
s40715645_phd_final.pdf s40715645_phd_final.pdf application/pdf 3.53MB 15
Author Danoucaras, Anastasia
Thesis Title Property-Based Modelling at Fixed Chemistry: The use of a back-calculated induction time for predicting recoveries in flotation
School, Centre or Institute Sustainable Minerals Institute
Institution The University of Queensland
Publication date 2011-07
Thesis type PhD Thesis
Supervisor Dr Sergio Vianna
Prof Anh Nguyen
Total pages 210
Total colour pages 10
Total black and white pages 200
Language eng
Subjects 0914 Resources Engineering and Extractive Metallurgy
Abstract/Summary Different research groups endeavour to model flotation with the goal of predicting its metallurgical performance (i.e. final grade and recovery). The approach adopted by the Australian Minerals Industry Research Association (AMIRA/P9) researchers for modelling flotation is to separate the pulp zone rate constant into two parameters; one accounting for the ore type and one accounting for machine characteristics. The parameter that represents the ore type is called the floatability (P) which is assumed to be conserved throughout the flotation circuit in the absence of regrinding, severe aggregation and/or oxidation and change in chemical environment. This is a simplification because in fact, it is a flotation response dependent on cell operating conditions such as gas rate and power dissipation, as well as the ore characteristics. It has long been recognised that the assumption of the conservation of P has limitations. Currently, P is a lumped parameter with no direct physical meaning. To obtain the value of P, time consuming batch tests are performed in conjunction with an industrial scale flotation survey. Any change to the ore characteristics such as a new head grade or particle size distribution, requires a new value of P to be found, which then requires another survey and additional batch tests. Furthermore, when operating conditions change too far from the initial survey that was used to obtain the P values, the model for recovery prediction is no longer valid. The aim of the thesis is to decouple P into component parts with physical meaning because when a measurable variable is incorporated into a model, it can be accounted for when it changes and so P can be predicted. The initial approach was to use an existing laboratory flotation based study (Vianna, 2004) investigating how the pulp zone rate constant of attached particles (and indirectly P), changed with particle size, degree of liberation and collector coverage. Vianna's (2004) results showed that the relationship between the rate constant and liberation had a linear region. Fundamental studies on the effect of hydrophobicity of particles on the rate constant were found to explain the gradients. It was found that for a given collector dose, the intermediate sized particles have the largest gradient of the composite particles. Fine particles are least affected by the increase in composition of galena in particles because it is size which is the dominant factor on their floatability. Coarse particles have a flatter gradient because they need a higher liberation value to enable them to float as strongly as the intermediate particles, at a fixed collector dose. The analysis indicated that there was a relationship between liberation and hydrophobicity. A fundamental model from the literature was chosen with variables that had physical meaning that could quantify the efficiency of collection (which will be shown to be the same as P) of a particle of a particular size and liberation class. In theory, most variables could be measured in-situ with the exception of induction time of a particle. In practice, viscosity and pulp density were not measured in the survey work but an approximation was found. Induction time is a measure of the hydrophobicity of a particle and can be measured in the laboratory but no technique exists for its measurement in a flotation cell. Hence, induction time of a particle size-by-liberation class was back-calculated. The resultant relationships showed the variation of induction time with liberation (or particle composition); however the back-calculated induction time (BCIT) is conceptually different to existing laboratory measurements. Once a model was chosen, it was necessary to validate the modelling approach using an industrial data set which was provided from a block survey of the BHP Billiton Cannington's lead rougher circuit located in Queensland, Australia. To validate the model, the assumption was made that the induction time, in the absence of staged collector addition, was conserved down the bank. The data from the first three cells of the circuit were used to obtain a fit for the induction time of each size-by-liberation particle class. The test of the model was then to use the fitted numbers of induction time within the model for efficiency of collection and predict the recoveries on a size-by-liberation basis of the fourth cell. The predicted recoveries were then compared against the experimental mass balanced ones and those predicted by the Modified Floatability Component Modelling (MFCMij) Approach. It was found that the two modelling approaches trended each other but over-predicted the recoveries. An extensive error analysis was conducted on the two modelling approaches. It was found that the MFCMij approach was less sensitive to the fluctuation of its parameters than that offered by the BCIT modelling approach. In particular, the error analysis showed that the bubble rise velocity would be a very significant parameter of Nguyen's model used in the BCIT modelling approach since small changes in the bubble rise velocity could lead to significant variations in the predicted rate constant. Finally, since the variation in the key flotation variables down the bank at the BHP Billiton Cannington's lead rougher circuit was not significant, the two approaches predicted similar flotation kinetics down the bank. The significance of the work is that it has found a way to relate induction time and liberation. If the relationship between induction time and liberation was well understood, then it would be possible to predict how the floatability P changes with liberation at fixed chemistry. The gain in no longer using P as a lumped parameter is that the model can account for changes in other variables because they are explicit.
Keyword flotation modelling
flotation efficiency
induction time
liberation
floatability
Additional Notes Should be printed in colour: Page numbers refer to the pdf document not the page number of the thesis, 66, 82, 108, 111, 112, 127, 153, 173, 174, 175

 
Citation counts: Google Scholar Search Google Scholar
Created: Thu, 26 Apr 2012, 09:50:39 EST by Ms Anastasia Danoucaras on behalf of Library - Information Access Service