Imperial College London Featured PhD Programmes
University of Southampton Featured PhD Programmes
University of Reading Featured PhD Programmes

Machine learning approaches for ore deposit characterization in the Platreef (Bushveld Complex, South Africa)

This project is no longer listed on FindAPhD.com and may not be available.

Click here to search FindAPhD.com for PhD studentship opportunities
  • Full or part time
    Dr H Hughes
    Dr James Hickey
    Dr S Das
    Dr Chris Yeomans
  • Application Deadline
    No more applications being accepted
  • Funded PhD Project (European/UK Students Only)
    Funded PhD Project (European/UK Students Only)

Project Description

The Northern Limb of the Bushveld Complex (South Africa) is host to the world’s largest resource of platinum-group elements (PGE), along with significant nickel, copper and cobalt in a complex magmatic sulphide ore deposit. All of these resources are linked with environmentally-friendly technologies and energy usages in the automotive industry, with the PGE being essential components in catalytic converters, and Ni, Co and Cu critical metals in Electric Vehicle batteries. The Northern Limb of the Bushveld is likely going to play a large part in the switch to cleaner automotive technology.

The major deposit in the Critical Zone of the northern Bushveld (commonly known as the ‘Platreef’) rests directly on variable basement of Archaean and Paleoproterozoic units at surface. Down dip, it overlies Bushveld Complex Lower Zone rocks; themselves prospective for base metal sulphide mineralisation. The stratigraphy, structure, mineralization styles and metal budgets of the Northern Limb of the Bushveld Complex show important differences to the other limbs of the complex. The causes of these differences are poorly understood and no geological model exists to investigate them on anything more than a local scale.

The characteristics of PGE mineralisation and host lithologies varies significantly laterally and with depth across the Critical Zone of the Northern Limb. Further complexities result from diverse and discontinuous alteration assemblages overprinting the layered igneous rocks. These intricacies mean that attempts to correlate the mineralisation across the Critical Zone in the Northern Limb according to geology or grade have ultimately led to disparities in models and generalisation, limiting visualisation of the full complexity of the mineralising system and potentially obscuring important relationships.

Machine learning algorithms (MLAs) are a suite of computationally efficient tools that can deal with large, high-dimensional input datasets with non-Gaussian distributions. They are flexible and can operate using no prior data to find clusters of similar samples (unsupervised methods) or take a pre-determined set of labels and fit a classification based on this information (supervised methods). Provided the data are prepared appropriately, MLAs are able to generate robust results and recognise patterns across multidimensional space. In geoscience, geochemical datasets present some of the most complicated inter-relationships to interpret – MLAs may be able to highlight previously unidentified correlations between suites of elements previously missed by more conventional statistical methods.

This PhD project aims to use MLA to unravel complex industry-standard bulk geochemical datasets in order to understand clustering and important correlations for geology and mineralisation. This PhD provides an exceptional opportunity to integrate machine learning with traditional geochemistry applied to an extensive company borehole database.

Research approach:

[A] Use unsupervised MLAs on bulk rock geochemical data to identify clusters of data and towards a machine learning geochemical-based identification process.
[B] Identify key lithologies based on [A] and compare these to borehole logs and databases. These initial interpretations will lead to supervised classifications of units and assignment of lithological (or geochemical) zones within the ore deposit.
[C] In combination with grade data (i.e., grade shells), PGE geochemistry and concentrations of penalty elements and by-products, a supervised regression will be used towards a predictive model that may be applicable to mine planning, geometallurgy or future exploration.

Whilst this PhD will apply strictly to the Critical Zone of the Northern Limb, the studentship may identify methodologies that allow for development of MLA applied to geochemical/geological data in broader areas of the northern Bushveld.

Further information:
This PhD forms part of a large consortium project (NL4D) funded by Anglo American involving University of Exeter (CSM), Cardiff University and University of Leicester. The NL4D project runs for five years from January 2020 and will include three PhD studentships (one based at Exeter) working alongside senior researchers, postdocs and Masters students from all three universities. The PhD researcher will have access to labs at all three institutions and the opportunity to spend some extended period working at each. There will be regular interaction with Anglo American through fieldwork, internal reporting and workshops/meetings in South Africa and the UK. The PhD researcher will be encouraged to present their results at national and international conferences.

Funding Notes

The University of Exeter’s College of Engineering, Mathematics and Physical Sciences, in partnership with Anglo American, is inviting applications for a fully-funded PhD studentship to commence in September 2020. For eligible students the studentship will cover UK/EU tuition fees plus an annual tax-free stipend of at least £15,009 for 3.5 years full-time, or pro rata for part-time study. The student will be based at the Camborne School of Mines at the Penryn Campus of the University of Exeter, in Cornwall. A research, training and travel support grant will be provided for the student over the course of the project.



FindAPhD. Copyright 2005-2020
All rights reserved.