The Robert M. Buchan Department of Mining Graduate Theses

Permanent URI for this collection


Recent Submissions

Now showing 1 - 5 of 129
  • Item
    Efficient Stope Sequencing Optimization Under Grade Uncertainty Using Genetic Algorithms
    (2024-05-08) Kheirparast, Soheil; Mining Engineering; Sari, Yuksel Asli; Ortiz, Julian M.
    Long-term mine planning refers to the optimization of strategies to exploit ore from a deposit efficiently and economically. It indicates which part of the orebody to extract, when to extract it, and where to send it in a way that the profitability of the operation is maximized, while complying with a set of constraints. However, the inherent grade uncertainty, stemming from limited drilling data, arises as a key challenge in mine planning. If not properly managed, this uncertainty can lead to deviations from production targets, and misleading financial expectations of the project, resulting in profit loss and a potential risk of mine shutdown. To address these challenges, stochastic mine planning leverages equiprobable realizations of the orebody to develop robust mine plans. One widely reported challenge of stochastic mine planning is the computational complexity of solving the mine scheduling problem, particularly when considering stochastic variables. Therefore, there is a need to develop efficient solution approaches for the problem to mitigate this complexity. The first part of this thesis provides an in-depth review of mine planning optimization models under grade uncertainty and their solutions methods. The review not only emphasizes the benefits of incorporating uncertainty into mine planning but also identifies the limitations and challenges encountered in existing studies. In the second part, the focus shifts to the integration of grade uncertainty into underground mine scheduling, particularly targeting strategic stope sequencing in sublevel stoping operations. The objective is to maximize the expected NPV while minimizing its variability across various equiprobable realizations of the stope layout. To efficiently solve the multi-objective optimization problem, a meta-heuristic approach based on Genetic Algorithms (GA) is developed to obtain quality solutions in a short timeframe. Moreover, the GA’s performance is enhanced to efficiently increase its searchability of the solution space. GA’s efficiency coupled with performance improvements significantly reduced the runtime, allowing for the generation and evaluation of many sequences with various parameters and risk levels for more robust and informed decision making. Finally, in the last part of this thesis, a framework is introduced to mitigate the computational complexity associated with the utilization of geostatistical simulations in stochastic mine planning applications, including risk assessment and management. The presented framework strategically decreases the number of grade realizations required for robust and informed decision making, thereby mitigating runtime issues without compromising the accurate representation of existing uncertainty.
  • Item
    Predicting Turbocharger Failures in Mining Haul Trucks
    (2024-04-30) Kalyoncuoglu, Nuri; Mining Engineering; Sari, Yuksel Asli
    This thesis presents an innovative approach for predicting turbocharger failures in mining haul trucks, leveraging the power of machine learning algorithms to improve predictive maintenance strategies within the mining industry. The haul trucks have a critical role in mining operations, and a turbocharger failure within the haul truck can cause significant costs. This research aims to minimize downtime, improve safety, and enhance operational efficiency. By incorporating machine learning models, this thesis introduces a novel predictive maintenance framework to monitor equipment conditions in real-time and identifies patterns that precede failures by analyzing sensor data from turbochargers, enabling preemptive maintenance actions. The study thoroughly examines turbocharger technology, common failure causes, and the role of various components. It also provides a detailed review of machine learning techniques applied to predictive maintenance such as ensemble learning, support vector machines, neural networks, and advanced anomaly detection methods. Two case studies form the core of the analysis, demonstrating how supervised and unsupervised learning models are applied for predicting failures and detecting anomalies. These models make use of sensor data to forecast the behavior of turbocharger systems, detecting both anomalies and potential failures. This approach allows for timely maintenance decisions, reducing unnecessary maintenance operations and preventing catastrophic failures. The thesis makes several contributions to the field, such as the use of machine learning models for predicting turbocharger failures, the development of new metrics for monitoring turbocharger health, and the identification of key sensors that affect turbocharger performance.
  • Item
    Grade Control Optimization in Open Pit Mining using Supervised Machine Learning Approach
    Potakey, Noble Eyiram; Mining Engineering; Ortiz, Julian
    Ore and waste classification in grade control is a critical aspect of mining operations to define which material is considered as ore or waste. Geostatistical techniques such as kriging and conditional simulation are used to estimate grades between sample points, followed by the application of a cut-off grade to discriminate the materials. Misclassification occurs when the estimated grades are not consistent with the actual grade distribution leading to ore loss and waste dilution. With the rapid evolution of computers and technology, including advancements in data processing and predictive capabilities, a machine learning aided classification and a corresponding evaluation system are proposed. In this research, Support Vector Classification (SVC) is used as a machine learning approach to optimize material classification, alongside an industrial approach based on Ordinary Kriging (OK). Sequential Gaussian Simulation (SGSIM) is used to simulate the ground truth against which the two grade control models (SVC and OK) are benchmarked across two different open pit projects. Both models demonstrated decent results, achieving close to 90% accuracy in their predictions. However, the Kriging model exhibited approximately 8.5% misclassification of blocks, while the Support Vector Machine (SVC) showed a slightly higher misclassification rate of about 10.5% across the two scenarios. The impact of sampling errors, another source of misclassification, is evaluated by introducing five levels of random noises to the original data. On average, the performances of both Kriging and SVC models reduced by about 4 % when sampling errors were introduced into the data. An economic evaluation of the lost ore due to misclassification in both models showed lower metal losses in OK than in SVC model, which increased in OK as the sampling error level increased.
  • Item
    Biological Oxidation of Refractory Sulfidic Ores at Circumneutral pH
    Hein Alvial, Guillermo A.; Mining Engineering; Ghahreman, Ahmad
    The industrial process for gold recovery, comprising acidic biological oxidation (bio-oxidation) and cyanidation, has demonstrated competitive effectiveness in dealing with the refractoriness of sulfidic gold ores. Nevertheless, many aspects could still be improved to strengthen the process, most prevalently the complexity associated with pH control and neutralization of acidic slurries preceding cyanidation, which results in high cost and generation of new waste streams. Consequently, performing bio-oxidation at circumneutral pH with an in-situ neutralization supported only by culture medium conditions offers a novel biotechnological opportunity to facilitate the standard process and enhance environmental protection. This research reports the capacity of unexplored microorganisms in conventional mining applications to oxidize sulfidic gold ores at circumneutral pH. First, the application of two neutrophilic bacteria, Thiobacillus thioparus, and Starkeya novella, was investigated, and optimal bacterial growth conditions and their capacity to oxidize different matrices were determined. A culture medium containing 4.5 g/L and 0.9 g/L thiosulfate as the initial energy source favored the oxidation of a sulfidic ore, with a maximum of 27.2% and 14% using 1% w/v after 10-day treatment assisted by Thiobacillus thioparus and Starkeya novella, respectively. Subsequently, the fungus Phanerochaete chrysosporium was selected as another biological engine to improve sulfide oxidation. The optimal conditions of the fungal treatment revealed that an initial pH of 5.8 resulted in 23.1% sulfide oxidation using 5% w/v following 14-day treatment. D-optimal response surface methodology suggested a modified culture medium consisting of 12.86 g/L glucose, 2.20 g/L malt extract, 1.67 g/L yeast extract, and 0.49 g/L MgSO4·7H2O to enhance microbial activity and reach 28.7% sulfide oxidation. pH-controlled batch cultures in the pH range of 5.8 to 7.0 showed that a rise in pH was detrimental to microbial activity and, thus, sulfide oxidation. Higher sulfide oxidation was accomplished by replenishing the microbial culture in a 42-day multi-stage bio-oxidation, attaining 51.1% sulfide oxidation in the modified culture medium. Furthermore, corn steep was studied to replace standard ingredients, resulting in 40.6% sulfide oxidation when substituting 1.67 g/L yeast extract. Conventional cyanidation was also performed in oxidized samples. Ultimately, implications and limitations of this research are summarized.
  • Item
    Reprocessing of the Cantung Mine Tailings
    Collins, Arik; Mining Engineering; Gibson, Charlotte; Johnson, Anne
    Tailings, a waste product generated from the processing of ore, constitute a large, long-lasting portion of a mine’s environmental footprint. The Cantung Mine, located in Northwest Territories, Canada, is a defunct mine with over 4 million tonnes of tailings. The tailings contain sulphide minerals, which can oxidize and produce acid rock drainage if left in their current state. Due to the mine’s proximity to the Flat River, any acid rock drainage produced by the tailings has the potential to impact the surrounding environment. This thesis investigates methods to reprocess and separate the Cantung tailings into two distinct fractions: a high-mass fraction of tailings that are not acid generating and an acid-generating low-mass concentrate containing sulphide minerals that can be handled and stored separately. By separating the mine’s tailings, benign waste might be filtered and stored in dry-stacks, reducing the environmental and structural risks posed by sub-aqueous tailings storage units. Mineralogical work determined that pyrrhotite was the main sulphide mineral present in the tailings. Scoping level flotation tests determined that the reagents sodium isopropyl xanthate, sodium hexametaphosphate, and Aero 6493 had the greatest influence on flotation results. A Box-Behnken experimental design was conducted to optimize the flowsheet. Grinding, magnetic separation, and flotation were employed to recover up to 86.5% of the total sulphur in a concentrate weighing 29.8% of the initial mass. The low-sulphide tailings contained 1.96% sulphur, which would reduce the impact of acid rock drainage if implemented. Analysis of the low-sulphide tailings determined that the remaining sulphur was primarily found in the -38 micron fraction, which is difficult to recover by flotation. Preliminary flotation tests were completed in an attempt to recover copper and tungsten in the tailings not recovered during the mine’s operation. Both metals were unable to be recovered at high grades, with further research required to determine appropriate recovery methods. The findings of this thesis demonstrate the effectiveness of reprocessing mine waste from the Cantung Mine to reduce the potential of future environmental impacts.