Modelling of MF2 Platinum Concentrators with Blended Feeds
| dc.contributor.author | Mneno, Dumile Peabo | |
| dc.contributor.supervisor | Higginson, Antony | |
| dc.contributor.supervisor | Brooks, Kevin | |
| dc.date.accessioned | 2025-11-13T08:54:11Z | |
| dc.date.issued | 2025 | |
| dc.description | A research report submitted in fulfillment of the requirements for the Master of Science, in the Faculty of Engineering and the Built Environment, School of Chemical and Metallurgical Engineering, University of the Witwatersrand, Johannesburg, 2025 | |
| dc.description.abstract | This study presents a data-driven approach to modelling mill-float-mill-float (MF2) platinum concentrator performance when processing blended feeds of Merensky and UG2 ores. Conventional phenomenological models are often limited by data availability and delayed laboratory results, hindering timely decision-making in complex mineral processing environments. Using real plant data from Two Rivers Platinum, this research seeks to understand how the MF2 circuit is operated with and without blended feeds and apply machine learning (ML) algorithms to predict ore mineralogy and plant performance. After comprehensive data cleaning and feature selection, four ML models: Linear Regression (LR), Random Forest (RF), Extra Tree (ET), and k- Nearest Neighbours (kNN) were developed and evaluated. The ET algorithm achieved superior accuracy in predicting ore mineralogy, while RF was most effective in forecasting key plant performance indicators. Despite constraints in data granularity, particularly with weekly mineralogical data, the models demonstrated robust performance and potential for integration into production planning. These findings underscore the value of ML in enhancing metallurgical plant responsiveness and optimising recovery in MF2 circuits handling variable ore blends. The research advocates for more granular data collection to unlock the full potential of predictive modelling and highlights machine learning as a scalable tool for process control in the platinum mining sector using blended feeds. | |
| dc.description.submitter | MM2025 | |
| dc.faculty | Faculty of Engineering and the Built Environment | |
| dc.identifier.citation | Mneno, Dumile Peabo. (2025).Modelling of MF2 Platinum Concentrators with Blended Feeds [Master`s dissertation, University of the Witwatersrand, Johannesburg]. WIReDSpace. https://hdl.handle.net/10539/47578 | |
| dc.identifier.uri | https://hdl.handle.net/10539/47578 | |
| dc.language.iso | en | |
| dc.publisher | University of the Witwatersrand, Johannesburg | |
| dc.rights | © 2024 University of the Witwatersrand, Johannesburg. All rights reserved. The copyright in this work vests in the University of the Witwatersrand, Johannesburg. No part of this work may be reproduced or transmitted in any form or by any means, without the prior written permission of University of the Witwatersrand, Johannesburg. | |
| dc.rights.holder | University of the Witwatersrand, Johannesburg | |
| dc.school | School of Chemical and Metallurgical Engineering | |
| dc.subject | UCTD | |
| dc.subject | MF2 | |
| dc.subject | Merensky-UG2 blend | |
| dc.subject | Extra Trees | |
| dc.subject | Random Forest | |
| dc.subject | k-NN | |
| dc.subject.primarysdg | SDG-9: Industry, innovation and infrastructure | |
| dc.subject.secondarysdg | SDG-12: Responsible consumption and production | |
| dc.title | Modelling of MF2 Platinum Concentrators with Blended Feeds | |
| dc.type | Dissertation |