Precisely measuring seismic arrival times is a labor-intensive task but is critical for both earthquake monitoring and subsurface imaging. Recently published deep learning models have demonstrated superior performance compared to traditional automatic approaches for picking arrival times. Although existing deep learning models have shown promising results, further advancements are necessary as their performance is not yet satisfactory especially when applied to new regions and station networks. Increasing model size has led to improved performance in other machine learning applications. We aimed to investigate whether enlarging deep learning models can increase performance on accepted benchmarks. We trained three models of varying sizes, small (1X), medium (4X), and large (16X), using globally distributed local and regional earthquake signals and background noise waveforms from a benchmark dataset, Stanford Earthquake Dataset. Our results indicate that the largest model (PickerXL) outperforms both the smaller models and Seisbench implementation of the PhaseNet model, which has the same number of parameters as our small model. The PickerXL model’s enhanced capacity to extract complex patterns from seismograms contributes to its superior arrival picking abilities compared to the smaller model.

Microscopy imaging techniques are instrumental for characterization and analysis of biological structures. As these techniques typically render 3D visualization of cells by stacking 2D projections, issues such as out-of-plane excitation and low resolution in the z-axis may pose challenges (even for human experts) to detect individual cells in 3D volumes as these non-overlapping cells may appear as overlapping. A comprehensive method for accurate 3D instance segmentation of cells in the brain tissue is introduced here. The proposed method combines the 2D YOLO detection method with a multi-view fusion algorithm to construct a 3D localization of the cells. Next, the 3D bounding boxes along with the data volume are input to a 3D U-Net network that is designed to segment the primary cell in each 3D bounding box, and in turn, to carry out instance segmentation of cells in the entire volume. The promising performance of the proposed method is shown in comparison with current deep learning-based 3D instance segmentation methods.

The energy sector is currently undergoing a transition towards increased utilization of green energy technologies. The green energy transition relies heavily on metals, such as aluminium, chromium, cobalt, copper, lithium, manganese, nickel, rare earth elements (REEs), silicon, tin, titanium, tungsten and zinc, among others. However, this transition occurs within the context of: (1) a geographical concentration of known mineral deposits and downstream capability; (2) a demand that vastly exceeds supply; (3) a strong drive to mitigate environmental and energy concerns; and (4) an increasing level of geopolitical conflicts. Consequently, the energy transition is not straightforward, as it intensifies material demand, market and geopolitical competition. This is especially true for lithium which is pivotal in this transformation. Concerns driven by material access, energy sustainability and national sufficiency are increasingly resulting in national and super-national geopolitical activities, such as resource nationalisation, forming strategic or trade alliances, encouraging near- and friend-shoring, promoting material circularity, and accelerating green technology research and deployment. This study examines the global impact of the green energy transition, from the perspective of the mineral value chain, including downstream products, its implications on the projected demand, and geopolitics. There are many potential outcomes in the future, depending on: (1) the pragmatic outcomes of the energy transition, which is only empirically realizable through implementation; (2) the strength of independence and global stability; and (3) the rise of regional or friendly trade blocs. In particular, this study examines a future scenario in which there is an emergence of an OPEC-style organisation for green energy minerals and metals (GEMMs), focusing on lithium as an example, because: (1) it has a clear and essential role in the green energy transition; (2) it is geographically concentrated in a manner that facilitates production coordination; and (3) it is overwhelmingly consumed by developed nations but supplied by developing nations. An organisation built around lithium could be a prototype to other GEMM markets. Consequently, we propose that it is possible to leverage existing circumstances and propose that such an organisation, here termed Green Energy-Mineral Exporting Countries, or "GEMEC", could serve as a collaborative platform to enhance geopolitical positioning, maximise economic benefits through coordinated production and export policies, and address the environmental, social and governance challenges associated with the green energy transition.

The green energy transition is aimed at mitigating the impact of climate change. Yet, the current emphasis on ‘green’ is narrowly centred around decarbonisation, or CO2 reduction, often side-lining the roles of other gases, such as sulphur hexafluoride (SF6) and PCF-14 (CF4), which have a respective 24,300- and 7380-times higher global warming potential than CO2 on a time horizon of a century. In addition, any energy transition is a complex affair that simultaneously impacts the environmental, economic and social systems, with significant system-level interactions. For example, the material requirement for renewable energy is known to be substantial, at a factor of 15 times greater than natural gas-based energy for offshore wind generation, and almost 7 times greater for solar. The resulting increased competition for materials is reducing the appetite for global collaboration. In addition, the global capacity to deploy renewable energy technology or participate in climate change mitigation is geographically variable and no single solution is universally viable. This study examines an expanded definition of ‘green’ energy and proposes a beyond-decarbonisation approach that is more comprehensive and globally inclusive, in pursuit for a sustainable transition. The increased diversity of our approach promises advantages such as heightened global collaboration, diminished geopolitical tension, improved energy access, expanded market opportunities, and socio-environmental co-benefits. Strategies pivotal to this approach involve understanding the role of carbon-based energy systems in the transition, amplifying renewable resources, augmenting cross-sector energy efficiency, implementing effective carbon markets, and integrating nature-based as well as carbon removal technologies. Moreover, it is imperative to implement all-cost and all-benefit monitoring and evaluation systems to optimise existing decarbonisation methods systematically. This could entail the use of composite metrics that normalise the gain in climate change mitigation against economic, social or environmental metrics. Addressing societal apprehensions requires a focus on pragmatic and fair outcomes, geopolitical stability, market impacts, developmental objectives, effective public engagement, and recognition of the role of enterprises. Policymakers are important in fostering global synergy by implementing policies that encourage international collaboration, investment, enterprise engagement, institution fortification, and cross-sector policy integration.

The Rustenburg Layered Suite of the Bushveld Complex has been extensively exploited for platinum group elements (PGEs), mainly from the Merensky and UG2 Reefs. An exploration drilling programme on the Booysendal Platinum producing Mine on the Eastern Limb of the Bushveld Complex revealed the occurrence of the UG2 Reef split-facies. In comparison to the normal UG2 reef (where the chromitite seam is not separated by pyroxenite), the split reef facies is defined by the splitting of the chromitite layers into multiple layers that are separated by pyroxenite middlings. This is anticipated to pose new challenges during metallurgical processing of the ore. To fully comprehend the mineralogical and metallurgical variations of the UG2 Reef split-facies types, geometallurgical test work was carried out on drill core samples collected from the mine. This was done to determine the bulk modal mineralogy, milling performance, mineral liberation and the flotation response of the various UG2 Reef split-facies types identified. Preliminary results of the study are presented and discussed in this paper.

The Bushveld Igneous Complex (BIC) hosts the largest reserves of Platinum Group Metals (PGMs) in the world with most of the Platinum Group Element (PGE) mineralization occurring within the Merensky Reef, UG2, Platreef and Middle Group Chromitites reef.This research sought to investigate the PGMs locked in the silicates of the UG2 ore from the Eastern Limb.As most of the PGMs are locked in the silicate stream, these are disposed in the tailing dams.The silicate stream was characterized using the XRF, XRD, SEM and fire assaying through nickel sulphide collection.1kg of the sample was milled in the rod mill to 80% passing 53 µm and floated in a D12 Denver flotation cell to recover the PGMs from the silicate stream based on Response Surface Methodology.It was found that the major mineral phases in the silicate stream were quartz, sphalerite, pyrrhotite, chalcopyrite and chromites.The size of the sulphide mineral that hosted the PGMs in the silicate was 20 µm.The highest concentrate grade obtained after flotation was 19.19 g/t of Pt.The as-received sample also had 0.95 g/t of Au which was recovered.

Reprocessing of historical tailings storage facilities (TSFs) has gained momentum worldwide to address the demand for critical metals and minerals. Limited information about these repositories is available, hence the need to fill this knowledge gap. A tailings dam of the Murchison Greenstone Belt (MGB) was characterized to quantify the presence of economic antimony in terms of modal mineralogy, relative mineralogy and their associations (i.e. co-existing phases), and the deportment of antimony in various antimony-bearing minerals. Holes were drilled from the top to the bottom of the tailings dam to acquire samples for chemical and mineralogical characterization. A FEI 600F Mineral Liberation Analyzer (MLA) was used to characterize this tailings dam, accompanied by ICP-MS, XRF and LECO analyses. The MLA results, complemented by ICP-MS analyses revealed the occurrence of considerable amounts of Sb in this TSF. The TSF is mainly comprised of quartz, magnesite, chlorite and dolomite. Chapmanite and stibnite are the most abundant antimony-bearing minerals, along with other antimony minerals such as schafarzikite, senarmontite and berthierite. Significant antimony is mainly deported in stibnite, schafarzikite and senarmontite, with minor deportment into berthierite, cervantite, and other antimony phases.

The Assen Fe ore deposit is a banded iron formation (BIF)-hosted orebody, occurring in the Penge Formation of the Transvaal Supergroup, located 50 km northwest of Pretoria in South Africa. Most BIF-hosted Fe ore deposits have experienced post-depositional alteration including supergene enrichment of Fe and low-grade regional metamorphism. Unlike most of the known BIF-hosted Fe ore deposits, high-grade hematite (> 60% Fe) in the Assen Fe ore deposit is located along the lithological contacts with dolerite intrusions. Due to the variability in alteration levels, identifying the lithologies present within the various parts of the Assen Fe ore deposit, specifically within the weathering zone, is often challenging. To address this challenge, machine learning was applied to enable the automatic classification of rock types identified within the Assen Fe ore mine and to predict the in-situ Fe grade. This classification is based on geochemical analyses, as well as petrography and geological mapping. A total of 21 diamond core drill cores were sampled at 1 m intervals, covering all the lithofacies present at Assen mine. These were analyzed for major elements and oxides by means of X-ray fluorescence spectrometry. Numerous machine learning algorithms were trained, tested and cross-validated for automated lithofacies classification and prediction of in-situ Fe grade, namely (a) k-nearest neighbors, (b) elastic-net, (c) support vector machines (SVMs), (d) adaptive boosting, (e) random forest, (f) logistic regression, (g) Naïve Bayes, (h) artificial neural network (ANN) and (i) Gaussian process algorithms. Random forest, SVM and ANN classifiers yield high classification accuracy scores during model training, testing and cross-validation. For in-situ Fe grade prediction, the same algorithms also consistently yielded the best results. The predictability of in-situ Fe grade on a per-lithology basis, combined with the fact that CaO and SiO 2 were the strongest predictors of Fe concentration, support the hypothesis that the process that led to Fe enrichment in the Assen Fe ore deposit is dominated by supergene processes. Moreover, we show that predictive modeling can be used to demonstrate that in this case, the main differentiator between the predictability of Fe concentration between different lithofacies lies in the strength of multivariate elemental associations between Fe and other oxides. Localized high-grade Fe ore along with lithological contacts with dolerite intrusion is indicative of intra-basinal fluid circulation from an already Fe-enriched hematite. These findings have a wider implication on lithofacies classification in weathered rocks and mobility of economic valuable elements such as Fe.

The mineralogy and texture of Ni-sulfide ores at the Nkomati nickel mine are highly variable, and this results in often erratic nickel recovery at the mine. The variability of the ore presents an opportunity to study the influence of grind size on the flotation-based recovery of Ni in highly heterogeneous sulfide ores, which would be applicable to this ore type at many other mines worldwide. In view of this, a process mineralogy investigation was conducted on thirteen mineralogically and texturally different nickel-sulfide ores from the Nkomati Nickel Mine, with a view on the influence of grind size on the flotation performance of pentlandite. Ore types presented include medium- and high-grade variants of the bleb, disseminated, massive, semi-massive, and net-textured sulfide ores of the Main Mineralized Zone (MMZ), as well as disseminated chromite-rich nickel sulfide ore and massive chromitite ore of the Peridotitic Chromitite Mineralized Zone (PCMZ). Laboratory scale metallurgical test work, comprising of sequential grinding and bench-top flotation testing of the ores, was conducted in combination with quantitative mineralogical investigation of the flotation feed and associated flotation products, using a FEI 600F Mineral Liberation Analyzer. The ore types under consideration require a variety of grind sizes (i.e., milling times) in order to attain optimal recovery of nickel through flotation. This is predominantly controlled by ore texture, and also partly by the abundance of the major constituent minerals in the ore, being pyroxenes, base metal sulfides, and chromite. Liberation of pentlandite is directly correlated with grind size (milling time), which is also positively correlated with the level of nickel recovery through flotation. A grind size of P80 at 75 µm results in the highest concentrate nickel grades of 7.5–8.1% in the PCMZ ores’ types which is the current grind for the PCMZ ores at Nkomati. A grind size of P77 at 75 µm yields the best overall pentlandite liberation, Ni recoveries of 84–88% and grades of 5.3–5.6% in the MMZ ores. This holds the potential to produce the best overall pentlandite liberation, nickel grades, recoveries from blending the MMZ and PCMZ ore types, and milling the composite ore at a target grind of P80 at 75 µm.

Geological and geophysical models are essential for developing reliable mine designs and mineral processing flowsheets. For mineral resource assessment, mine planning, and mineral processing, a deeper understanding of the orebody's features, geology, mineralogy, and variability is required. We investigated the gold-bearing Black Reef Formation in the West Rand and Carletonville goldfields of South Africa using approaches that are components of a transitional framework toward fully digitized mining: (1) high-resolution 3D reflection seismic data to model the orebody; (2) petrography to characterize Au and associated ore constituents (e.g., pyrite); and (3) 3D micro-X-ray computed tomography (µCT) and machine learning to determine mineral association and composition. Reflection seismic reveals that the Black Reef Formation is a planar horizon that dips < 10° and has a well-preserved and uneven paleotopography. Several large-scale faults and dikes (most dipping between 65° and 90°) crosscut the Black Reef Formation. Petrography reveals that gold is commonly associated with pyrite, implying that µCT can be used to assess gold grades using pyrite as a proxy. Moreover, we demonstrate that machine learning can be used to discriminate between pyrite and gold based on physical characteristics. The approaches in this study are intended to supplement rather than replace traditional methodologies. In this study, we demonstrated that they permit novel integration of micro-scale observations into macro-scale modeling, thus permitting better orebody assessment for exploration, resource estimation, mining, and metallurgical purposes. We envision that such integrated approaches will become a key component of future geometallurgical frameworks.

This work proposes to optimize an additive manufacturing AM process to reduce energy and printing cost. The polymer AM process is inherently dependent on the time-temperature history of each layer to maintain geometric tolerances and mechanical integrity. Our preliminary study shows that regression-based layer time control model using thermal images could result in up to 30% build time reduction for simple geometries. This proposed work would use high-performance computing (HPC) to couple the data-driven model with thermal simulation for better predicting layer temperature profiles, improving throughput of large-scale additive manufacturing, and reducing its energy cost.We have developed a method to optimize a layer deposition time (a.k.a. layer time) for large-scale AM via physics-based simulations. A long layer time leads to an over-cooled surface on which a new layer is deposited, and therefore, it may result in a weak bonding or debonding between layers, cracking, or warping. A short layer time leads to a high temperature of the structure due to insufficient cooling, and therefore, the structure may not be stiff enough and may collapse during manufacturing. Therefore, it is important to estimate the optimal layer time in additive manufacturing for a high-quality product. The temperature of a top layer right before deposition is recommended to be slightly higher than the glass temperature of the material. A temperature cooling was approximated to an exponential function of time, and the optimized layer time was obtained based on a target temperature while maintaining a minimal printing time. The material used is carbon fiber-reinforced polycarbonate (CF/PC), and the large-scale deposition system used is LSAMTM from Thermwood Corporation. Three different layer time cases were used for experiments, and a series of thermal images were obtained via an infra-red (IR) camera during the entire AM processes. AM process simulations were performed using a finite element method and the temperature profiles from the simulation were in good agreements with those from experiments. The layer time optimization was performed based on the temperature profiles from the simulations. A layer temperature with the optimal layer time was confirmed as the target temperature through simulation.In addition to the development of a layer time optimization method, we have developed a numerical framework for AM simulation with element activations in sync with toolpath, based on an open source finite element framework, DEAL.II.A major portion of this work was presented at SAMPE 2022 Conference and Exhibition on May 2022, and published in Proceedings of SAMPE 2022.

Cyclic DARTS (CDARTS) is a Differentiable Architecture Search (DARTS)-based approach to neural architecture search (NAS) that uses a cyclic feedback mechanism to train search and evaluation networks concurrently. This training protocol aims to optimize the search process and evaluate the deep evaluation network comprised of discretized candidate operations. However, this approach introduces a loss function for the evaluation network dependent on the search network. The dissimilarity between the evaluation network’s loss function used during the search and retraining phases results in a search network that is a sub-optimal proxy for the final evaluation network accessed during retraining. We present a revised approach that removes the dependency of the evaluation network weights upon those of the search network. In addition, we introduce a modified process for relaxing the search network’s zero operations that allows these operations to be retained in the final evaluation networks.

Mechanized mining methods adopted in platinum-group elements (PGE) industry leads to mobile machinery leakages which result in mine sludge contamination by hydrocarbons.This paper aims at investigating the mineralogical characteristics of the contaminated mine sludge in comparison to the pristine ore.Two types of samples were used in the study were pristine PGE ore and contaminated PGE mine sludge.The samples were analysed using FTIR, XRF, XRD, SEM-EDS, Fire assaying, ICP-OES and Malvern PSD analyser for elemental composition, mineral composition, particle size distribution and the presence of hydrocarbon functional groups.FTIR results indicated the presence of a single C-H bonds in the contaminated mine sludge, which is the hydrocarbons functional group and pristine PGE ore was found to be free from an indication of such contaminants.The fire assaying results revealed that the mine sludge contained a total of 9.32g/t 4E (5.68 ppm Pt, 2.95ppm Pd, 0.6ppm Rh and 0.09ppm Au) and the pristine ore 4.74g/t 4E (2.64ppm Pt, 1.63ppm Pd, 0.42ppm Rh and 0.05ppm Au).SEM-EDS further confirmed that the sludge is indeed richer in PGE than pristine PGE ore.Mineral phases identified by XRD included millerite, chalcocite, aluminium oxide, chalcopyrite, pyrite, pyrrhotite, pentlandite, anorthite, enstatite and magnesium chromite and they were common to both samples.Particle size analysis revealed that the hydrocarbon contaminated mine sludge and the pristine ore had a P80 of 169 and 252µm respectively.Both samples had approximately 4% of fines (<10 µm).Remedial of hydrocarbon contaminants would lead to high grade PGE recovery.

The quest for steady primary supplies of critical raw materials (CRMs) creates significant waste, which is inevitably generated at each phase of mining and mineral processing. Waste from extraction, separation and refinement of non-renewable natural resources is accumulated globally and creates not only environmental hazards but also economic possibilities. Mine waste management is an expensive and prolonged task but unavoidable. Mine tailings, especially historical ones, can contain economically feasible resources, and given the right condition, such tailings could be reutilised through a waste valorisation concept. A prominent example are the Witwatersrand gold mine tailings in South Africa, which have been reused in small-scale projects. Tailing reutilisation is only possible if a sound classification, sampling and resource modelling framework is established to thoroughly and accurately profile the economic, environmental, health and geometallurgical aspects. Our study on valorisation of mine waste is presented in two parts: Here, in Part I, we focus on the essential components of a mine waste valorisation framework that includes the characterization and development of a systematic sampling framework for consolidated mineralised tailings. The development of a mine waste valorisation framework will hopefully enable worldwide reduction and reutilisation of mine waste.

Mine waste can create long-term and occasionally catastrophic environmental degradation. Due diligence of mine waste in the form of monitoring and maintenance requires a constant supply of societal resources. Furthermore, mine waste is unlikely to disappear with current mining methods and instead, it is more likely to accumulate at a faster rate due to decreasing primary ore grades and increasing societal demands. However, mine waste can be a societal asset, as it can offer an alternative source of partly critical raw materials (CRMs) that can augment primary sources and provide an opportunity to mitigate supply-risk while ensuring sustainability and easing geopolitical tensions. Cobalt is a critical raw material that is largely a by-product of mining of copper, nickel and platinum-group element ores. It is an element that the renewable energy and high-tech sectors critically depend on and for which no reasonable substitutes currently exist. The majority of the global cobalt production stems from the Central African Copperbelt. Published cobalt production figures for the Central African Copperbelt were used to evaluate cobalt tailings from the Central African Copperbelt. As part of a waste valorisation framework that focuses on primarily on the technical aspects of mine waste valorisation, this study assesses the application of key geostatistical methods, such as kriging and conditional simulation, followed by uniform conditioning, to evaluate the resource potential in a hypothetical copper-cobalt tailing deposit from the Central African Copperbelt. The results indicate that methods such as traditional algorithmic kriging, sequential Gaussian simulation and uniform conditioning are highly effective tools in resource modelling of mine waste. The resource assessment framework component developed in this study makes it possible to systematically characterise, profile and model any mine waste storage facility and thus supplements other framework components discussed in an accompanying paper to maximise mine waste utilization.

Neuromorphic systems allow for extremely efficient hardware implementations for neural networks (NNs). In recent years, several algorithms have been presented to train spiking NNs (SNNs) for neuromorphic hardware. However, SNNs often provide lower accuracy than their artificial NNs (ANNs) counterparts or require computationally expensive and slow training/inference methods. To close this gap, designers typically rely on reconfiguring SNNs through adjustments in the neuron/synapse model or training algorithm itself. Nevertheless, these steps incur significant design time, while still lacking the desired improvement in terms of training/inference times (latency). Designing SNNs that can mimic the accuracy of ANNs with reasonable training times is an exigent challenge in neuromorphic computing. In this work, we present an alternative approach that looks at such designs as an optimization problem rather than algorithm or architecture redesign. We develop a versatile multiobjective hyperparameter optimization (HPO) for automatically tuning HPs of two state-of-the-art SNN training algorithms, SLAYER and HYBRID. We emphasize that, to the best of our knowledge, this is the first work trying to improve SNNs’ computational efficiency, accuracy, and training time using an efficient HPO. We demonstrate significant performance improvements for SNNs on several datasets without the need to redesign or invent new training algorithms/architectures. Our approach results in more accurate networks with lower latency and, in turn, higher energy efficiency than previous implementations. In particular, we demonstrate improvement in accuracy and more than 5 × reduction in the training/inference time for the SLAYER algorithm on the DVS Gesture dataset. In the case of HYBRID, we demonstrate 30% reduction in timesteps while surpassing the accuracy of the state-of-the-art networks on CIFAR10. Further, our analysis suggests that even a seemingly minor change in HPs could change the accuracy by 5 − 6 ×.