Dear colleagues of hand-lens and geological pick – picota (sp.) or marteau (fr.) – in the next 5 years we could be witnessing a paradigm change with respect to the geometallurgical data of the Spatial (or block) Model in comprehensive geological works. Recently, Python and Matlab algorithms have been implemented and trained for determining, classifying and simulating textures in mineral deposits (Koch P-H., 2017).

Nowadays, engineers are working to find the relationship of textural classes with metal recovery and grade elaborating Process Models in specialized software (HSC Sim, Outotec). In the midterm, there will be greater corporative confidence in this software primarily used in developed mining countries (Canada, Sweden, Finland, etc.) as they will start using dynamic simulation, i.e.: mass balance reached in time.

We will soon be simulating particles, liberation and mineral associations by particle-size intervals, allowing concentrate production optimization following the “mineral approach” of Geometallurgy level 3. Core logging automation (Hylogger, Corescan, etc.) was only the beginning of the great deal of technological toys that we will be playing with. In fact, todays cost per meter is becoming more accessible in price and portability is ridiculously good. As has been observed many times before in this industry the technological acquisitions start with consultancy firms, junior companies and giant mining companies. However, the great majority of companies in developing countries today seem to be in the middle of an indecisive gauss curve, like if they were waiting for prices and analysis times to decrease further. Overall, my perception is that these new techniques help us for gaining sampling and analytic representability.

A paradigm of the past century, that is already disappearing in a critical mass of companies, is that Mining is a ‘miners’ problem (is TNT better than the geologist for finding the ore lode?). This is something that rarely applies and, if so, only in a very short time span in mineral deposits with exquisite grades or homogeneous mineralogy. In Peru, the geologist that changed this paradigm was Don Alberto Benavides de la Quintana (1920-2014, 93 years old) and I could write another article telling (or rather translating) the story of what did he do. Today’s most successful and responsible companies use Geostatistics and block models that include critical variables affecting mineral processing (the “bulk rock” concept; Canchaya S.). Nonetheless, this is also going to change.

I (Marco Acevedo, ESR2) am working in a PhD in Trinity College Dublin (Ireland) that studies 2D Correlative Microscopy, i.e.: combining microscopy images Big Data (optical, SEM, EDS, EPMA, LA-ICP-MS mapping, etc.). I have planned elaborating a software solution in Python that can navigate and process spectral data from images (pixels) efficiently (in a laptop) and following a petrographic workflow. Primarily, it is intended for scaling up LA mapping (mm scale) to entire thin sections (cm scale). In doing so, it will allow me to give geometallurgical feedback at the scale of trace elements distributed in sulphide grains. Additionally, joining my new software output with simulation of particles in HSC Sim (flowsheet streams) will allow us to have a new dimension in mineral processing predictability of metal concentrates (bonus and penalty elements) and much more.

My partner (Pratama Istiadi, ESR3) is working in a PhD in Luleå University of Technology (Sweden), parallel to mine and that goes even further, innovating in the midterm (5-10 years). He is looking forward to develop X-ray computed micro tomography (µXRT) in 3D applied to not just Geology, but also Geometallurgy. His PhD could correct the stereological error in mineral wt. % estimations of 2D mapping (my project) of thin sections to give a better 3D representation of particle shape in the head rock materials with liberation considerations. Subsequently these 3D particles could then be simulated with mineral processing software, allowing us to forecast productions accurately.

This way, we are looking forward to give Geology a quantitative sense that Engineers can understand. Likewise, the Geometallurgy branch advances at an increasing pace towards obtaining more detail and smaller scale features in mineral characterization studies. For example, process engineers (e.g.: Outotec) are trying to adapt the analytical data of Mineral Liberation Analysis (MLA) to unitary models of comminution and flotation that also use proxies of the different stages of mineral processing (ball and SAG mills, flotation cells, thickeners, etc.). In this context, the geological validation of Machine Learning algorithms applied in our PhD projects also plays an important role and we could greatly benefit from the contribution of mine geologists contributing to our databases (“crowd sourcing”).

I am hoping that through communication, ex-colleagues and friends in the industry start gaining interest, anticipate paradigm changes and know how to open their gates to technological transference. We all need to update traditional practices with greater levels of automation and detail in feasibility studies (incl. laboratory, pilot and industrial tests). A good example is Aitik Mine (Sweden) that has the smallest production cut-off in the world for an open pit operation.

In our research group (@MetalIntelligence, Horizon 2020, EU) we are working together to shift the paradigm in mining industries. We work to better analyze and quantify data obtained from geologists, which allow us to build better production planning, process modelling, and even training of the operators. Please explore our website for more details of this exciting project! Cheers!


Leave a Reply

Your email address will not be published. Required fields are marked *

This site uses Akismet to reduce spam. Learn how your comment data is processed.

By continuing to use the site, you agree to the use of cookies. more information

The cookie settings on this website are set to "allow cookies" to give you the best browsing experience possible. If you continue to use this website without changing your cookie settings or you click "Accept" below then you are consenting to this.