Mines Repository
Communities in Mines Repository
Select a community to browse its collections.
Recent Submissions
Publication Embargo Integrated approaches of systems engineering and graph analytics for mineral supply chains analysis(Colorado School of Mines. Arthur Lakes Library, 2025)The global push for decarbonization and supply chain resilience has placed increasing pressure on the mining industry to reduce its environmental impact while ensuring a steady supply of critical minerals. This dissertation integrates systems engineering and graph analytics to analyze three key aspects of mineral supply chains: (1) decarbonization of mining haulage systems, (2) emission impacts of iron ore processing, and (3) supply chain transparency and resiliency assessments using graph-based methods. The first study applies system dynamics modeling (SDM) to quantify greenhouse gas (GHG) emissions and economic trade-offs in open-pit copper mining haulage systems. It compares a diesel truck fleet with an electric overland conveyor system under different electricity grid emission factors. Results show that electrification can significantly reduce emissions when powered by low-carbon electricity sources. However, high capital expenditure (CAPEX) and regional energy mix variability present key challenges to widespread adoption. The second study focuses on iron ore processing systems in Canada and Peru, examining how geographical location and ore type influence emissions. Using system dynamics modeling, this study models the iron ore processing systems and quantifies the GHG emissions of those plants while considering the regional energy grids and the type of ores processed. Results indicate that processing magnetite ore in fossil fuel-reliant grids (e.g., Peru) generates significantly higher emissions than in hydropower-dominated grids (e.g., Canada). Additionally, the study assesses the long-term effects of Canada's carbon tax policy and energy transition strategies on iron ore processing emissions. Findings reveal that while carbon pricing effectively reduces emissions over time, its success depends on the availability of low-emission electricity and industry adaptation to regulatory changes. The third study reconstructs a complete mineral supply chain network using global trade transaction data and employs graph analytics to assess transparency and resiliency. This study provides a novel methodology to assess the transparency of the supply chain networks by a graph-based methodology. Furthermore, the study applies various graph analytics metrics to identify the critical nodes, bottlenecks, and geopolitical risks in supply chain operations and applies stress testing to assess the resiliency of the network by applying random disruptions to the supply chain network. Results show that supply chain disruptions can be quantified using graph metrics, providing a scalable method for enhancing risk management and emission tracking in global mineral supply chains. By integrating systems engineering and graph analytics, this dissertation provides a comprehensive data-driven decision-support framework for policymakers and industry stakeholders. The findings offer practical strategies for reducing mining emissions, optimizing processing systems, and enhancing supply chain resiliency and transparency, contributing to a more sustainable and transparent mineral industry.Publication Embargo Depositional process and mechanical properties of carbonate turbidites(Colorado School of Mines. Arthur Lakes Library, 2025)Carbonates in deep-water environments, resedimented by sediment-gravity flows, exhibit greater complexity and variability than siliciclastic, as carbonate grains display a broader range of settling velocities. This variation is influenced not only by grain size but also by shape and density. These factors are governed by diverse morphologies, skeletal structures, and mineral compositions, which ultimately affect sediment-transport processes. Turbidity currents are the key transport mechanism in generating deep-marine successions. However, studies about depositional processes and turbidite bed types have predominantly focused on siliciclastic systems. This research utilizes polished slab and thin section data from two locations representing a variety of slope sub-environments—the Mississippian Lake Valley Formation in the Sacramento Mountains, New Mexico, and the Permian Bell Canyon Formation in the Guadalupe Mountains, Texas. The study aims to enhance understanding of calciturbidite depositional processes (including both high-density and low-density turbidites) and investigates its impact on the mechanical properties through an assessment of fracture intensity. I specifically analyze grain-size distribution trends and grain orientations through quantitative methods, integrating observations to reconstruct hydrodynamic processes. This includes examining the relationship between flow velocities, bed thickness, aggradation rates, and the relative duration of deposition. Such interpretations help determine whether the deposits originated from high-density or low-density turbidity currents. Our findings have significant implications for predicting sedimentological and geomechanical characteristics in both conventional and unconventional hydrocarbon reservoirs, particularly within calciturbidites in carbonate slope settings.Publication Embargo Table manners at scale: introducing Lora Lens for efficient analysis & detoxification of language models(Colorado School of Mines. Arthur Lakes Library, 2025)Transformer-based language models have achieved significant advancements across numerous natural language processing (NLP) tasks. However, as these models grow in scale and complexity, ensuring interpretability and mitigating toxic outputs become increasingly critical challenges. This thesis addresses these issues by first analyzing the role attention heads play in propagating toxicity within models, leveraging a recently proposed interpretability tool known as Attention Lens. By decoding attention head outputs into human-interpretable tokens, we identify specific attention heads contributing disproportionately to toxic content generation and demonstrate that targeted interventions at the head level significantly reduce toxicity without requiring complete model retraining. Despite its effectiveness, Attention Lens faces severe limitations in scalability and computational efficiency. To overcome these limitations, we propose and implement the Lora Lens, an innovative adaptation of Attention Lens employing Low-Rank Adaptation (LoRA) to drastically reduce memory footprint and computational cost. Specifically designed for compatibility with large-scale models such as Llama 3 8B, Lora Lens integrates seamlessly with HuggingFace's transformers library and Microsoft's DeepSpeed framework, enabling efficient distributed training. Our results demonstrate that Lora Lens maintains the interpretative capabilities of Attention Lens while significantly enhancing its efficiency and scalability, allowing practical deployment on models with billions of parameters. Ultimately, this work contributes a practical, scalable interpretability technique, enabling researchers and practitioners to better understand, evaluate, and safely deploy large transformer models.Publication Embargo Resource recovery with membrane processes: overcoming limitations and increasing performance and economics(Colorado School of Mines. Arthur Lakes Library, 2025)Resource recovery from water and wastewater is of increasing importance due to stressors such as climate change, which decreases water availability, especially in the western United States. Several solutions to mitigate water stress have been proposed and are being implemented, and membrane processes offer unique benefits and high potential for enhancing both water supplies and recovery of valuable resources. This thesis examines two ways in which membrane processes can be utilized to increase resource recovery. The first focuses on increasing resource recovery in membrane desalination processes. The goal was to increase water recovery in a closed-circuit reverse osmosis system by implementing a scaling detection system, which enables a water recovery of over 90% from a gypsum-rich solution without scaling occurring by detecting crystal growth as it occurs in solution. From the highly concentrated brine, gypsum was recovered, which can be reintroduced to the economy. The second focus was on utilizing membranes for the recovery of nitrogen from nitrogen rich streams. Anaerobic digester centrate was collected, analyzed and a pretreatment process developed. Nitrogen was then recovered from the pretreated centrate through a membrane contactor process. This was done in lab-scale experiments at 1.5 L/min with up to 30 L batches to optimize conditions for pilot-scale testing at 75 L/min with over 5,500 L batches. Pilot-scale testing was then implemented with an ultrafiltration pilot system as a pretreatment process. The ultrafiltration system removed solids and foulants from the centrate, before nitrogen was recovered with a membrane contactor system. The pretreatment process combined with the membrane contactors system allowed for >80% of nitrogen recovery without nitrogen recovery performance declining. The economic viability of the pretreatment and membrane contactor processes for a 20 MGD wastewater treatment facility was investigated in a techno-economic analysis, with data from piloting applied for the economic evaluation of the process in a model. The results indicate large economic benefits for wastewater treatment facilities larger than 10 MGD. Sensitivity analysis of key parameters for the economic viability of the process for a 20 MGD wastewater treatment facility was conducted. Base (NaOH) cost, aeration demand, and electricity cost were found to have a large impact on economic viability of the process.Publication Embargo Directional synthesis and catalytic activity evaluation of rock salt nickel oxides for the oxygen evolution reaction in anion exchange membrane electrolyzers(Colorado School of Mines. Arthur Lakes Library, 2025)This thesis presents an in-depth view into the development process of transition metal oxide electrocatalysts to be used in Anion Exchange Membrane Water Electrolyzers (AEMWEs) for green hydrogen production. To begin, a literature review was done to gain a robust understanding of the different structures of transition metal oxide electrocatalysts and the enhancement strategies used to improve their catalytic activity. The work done in this thesis focuses on the crystal and interface engineering enhancement strategies in particular. This review ends by highlighting that rock salt transition metal oxides electrocatalysts have not been as extensively studied as their perovskite, spinel, and layered double hydroxide (LDH) counterparts. A synthetic study was done comparing rock salt nickel oxide nanomaterials synthesized with different exposed surface facets and crystalline properties using either solvothermal wet chemistry or molten salt solid state synthesis techniques. The sharp contrast in OER activity between samples was attributed to agglomeration issues common to molten salt synthesis procedures, mechanistic differences caused by different surface coverages, and potential issues of ink dispersion. Finally, a study of the effects of Fe incorporation into nickel oxide and hydroxide nanosheets produced through a microwave-assisted synthesis was done. The nanosheets were tested with two electrochemical setups: rotating disk electrode (RDE) and membrane electrode assembly (MEA) cell testing. The difference in activity trends among the oxide and hydroxide catalysts between the two testing systems provided further insight into the relationship between catalyst property and activity among the different testing environments. The results of this study indicated that Fe-incorporation can improve the OER activity of rock salt nickel oxide systems in ways different than the Fe-incorporation of NiFe-LDHs. Future experiments to support the insights gained in this thesis and improve electrocatalyst design knowledge are discussed.