The application of a machine vision system to relate to froth surface characteristics to the metallurgical performance of a PGM flotation process

This dissertation describes the use of a machine vision system to explore the relationships between the structure of the flotation froth surface, flotation performance and selected operational parameters in a flotation operation aimed at the beneficiation of Platinum Group Minerals. The machine vision system employed made use of the Fast Watershed Transform to resolve the froth surface into a distribution of bubble sizes. This distribution was then used to describe changes occurring in the structure of the flotation froth surface. An initial batch flotation programme concentrated on the influences that changing the frother and depressant dosage levels had on a flotation system treating a sample of Merensky ore, which originated from Rustenburg Platinum Mine's Amandelbult Section. It was found that both depressant and frother dosage had marked influences on the metallurgical performance of the flotation system. Furthermore, it was found that changes in these operational variables resulted in changes in the structure of the flotation froth surface, which were quantified by the machine vision system. These results highlighted the potential of the machine vision system to identify when either the frother or the depressant dosages had moved towards levels that would negatively influence flotation performance. A plant-based experimental campaign was then conducted on the Amandelbult Merensky Primary Rougher Flotation Section, in which the frother and depressant levels were manipulated. In terms of metallurgical performance of the flotation system, it was found that a strong relationship existed between the grade of concentrate being recovered and the rate of mass yield from the flotation unit. Furthermore, it was found that increasing the frother dosage led to an increased rate of mass yield and a decrease in concentrate grade, while increasing the depressant led to a decreased rate of mass yield at an increase in concentrate grade. As both of these variables influence metallurgical performance, a strategy that enables the automatic identification of which reagent dosage had deviated from the desired level would clearly be of great value. A strategy based on the observation of bubble size distributions at the top and bottom of the flotation bank, and which is capable of identifying whether either the frother or the depressant dosages are changing, is suggested in this work.

Grantor dc:publisher.institution

Department of Chemical Engineering

Year dc:date.issued

2000

Author dc:creator

• Sweet, Craig Gaven

Advisor dc:contributor.advisor

• Bradshaw, Dee

Language dc:language.iso

eng

Handle dc:identifier.uri

http://hdl.handle.net/11427/17885

OAI identifier oai:identifier

oai:open.uct.ac.za:11427/17885