U.S. Department of Energy has identified a set of isotopes for which accurate nuclear reaction cross sections are required, particularly for medically significant and scientifically relevant applications. One such isotope is zirconium-89 (Zr-89), which is widely used in immuno-positron emission tomography (PET) for the detection of breast, kidney, and brain cancers. A detailed understanding of its production pathways is therefore essential for optimizing and expanding the availability of Zr-89. 

Using photonuclear reactions, the Zr-90(γ,n)Zr-89 reaction was first modeled with the FLUKA Monte Carlo simulation code and subsequently studied experimentally at the Idaho Accelerator Center at Idaho State University. A variable-energy electron accelerator operating between 8 MeV and 23 MeV was used to measure the induced activity and excitation function of the reaction. The resulting excitation function was then used to extract the reaction cross section. 

Paste backfill stope filling involves highly transient, non-Newtonian, multi-phase flow processes that govern material deposition, pressure development, and fill performance, yet these mechanisms are not adequately captured by traditional design approaches. This research applies Computational Fluid Dynamics (CFD) to systematically investigate paste backfill behavior during stope filling. A transient, multi-phase CFD framework incorporating non-Newtonian rheology is used to simulate paste inflow, free-surface evolution, and air displacement within complex stope geometries. The simulations quantify flow patterns, velocity and shear rate distributions, and segregation-prone zones influencing fill uniformity. Time-dependent static and dynamic pressures acting on barricades are resolved to assess structural loading throughout the filling process. Parametric analyses of inlet location and filling sequence are conducted to evaluate their effects on consolidation behavior and free water generation. This research will present CFD model setup, governing equations and comparative filling scenarios, together with pressure–time histories and final fill profiles. Results demonstrate that CFD provides a rigorous, physics-based tool for optimizing paste placement strategies, improving barricade design reliability, and enhancing stope stability in underground mining operations. 

Surgical gesture prediction, phase prediction, and skill assessment are often treated as independent tasks, which limits the ability of models to capture shared structure and interdependencies across these related objectives. We propose a unified generative multimodal framework based on a Discrete Denoising Diffusion Probabilistic Model (D3PM) that jointly performs surgical gesture prediction, surgical phase prediction, and surgeon skill assessment within a single architecture. The framework first learns a gesture prediction model that integrates visual, linguistic, and action features and achieves strong performance on surgical gesture recognition. Transfer learning is then used to adapt the trained gesture model for phase prediction by replacing the gesture prediction head with a phase prediction head. Surgeon skill assessment is enabled through multi-component regression heads that predict Global Rating Scale (GRS) scores. We evaluate the framework on the JHU-ISI Gesture and Skill Assessment Working Set (JIGSAWS) dataset across suturing, needle passing, and knot tying tasks. Experimental results demonstrate strong performance across all objectives, with transfer learning configurations for phase prediction improving training efficiency by at least 35 percent while maintaining competitive accuracy. Overall, the proposed framework is computationally efficient due to its unified design and supports joint understanding of surgical behavior and automated skill assessment, with strong potential to generalize across tasks and surgical settings. 

Quadruped robots are increasingly deployed in challenging and unstructured environments, where robust and adaptable locomotion strategies are essential. This work presents a unified two-stage trajectory optimization framework for quadruped jumping under varying gravity conditions. The proposed approach decomposes the problem into an outer-loop planner that generates center-of-mass (CoM) trajectories based on ballistic motion, and an inner-loop optimizer that enforces full-body dynamics and contact constraints. A key advantage of the outer-loop formulation is that it does not depend on detailed robot parameters, leveraging gravity-driven dynamics to enable generalizable trajectory planning across different environments. The inner-loop optimization refines these trajectories to produce dynamically feasible motions while ensuring accurate tracking and stability during takeoff, flight, and landing phases. To improve numerical accuracy and consistency, trapezoidal collocation is incorporated into the trajectory optimization. The framework is evaluated in simulation using a quadruped robot model under multiple gravity conditions, including Earth, Mars, Moon, and intermediate cases. Results demonstrate that the proposed method can generate stable and repeatable jumping motions with high position accuracy across a wide range of environments. Additionally, the use of trapezoidal collocation leads to smoother trajectories and improved dynamic consistency. This work provides a flexible and scalable approach to quadruped locomotion, with potential applications in planetary exploration and energy-efficient robotic mobility. 

Orthohantaviruses are zoonotic RNA viruses maintained in natural reservoir hosts, primarily rodents, where they establish persistent and typically asymptomatic infections. However, certain viral strains are capable of causing severe disease in humans, including hemorrhagic fever with renal syndrome (HFRS) and hantavirus cardiopulmonary syndrome (HCPS). Distinguishing pathogenic viral sequences from reservoir-associated sequences remains an important challenge for understanding viral evolution, predicting zoonotic risk, and improving surveillance of emerging infectious diseases. In this study, we plan to investigate whether modern protein language models can effectively identify pathogenic viral sequences from protein sequence data. A curated dataset containing reservoir-associated and pathogenic viral protein sequences was constructed and used for supervised learning experiments. Three state-of-the-art protein representation models—ESM, ProtBERT, and ProstT5—will be evaluated for their ability to encode biologically meaningful sequence features and classify pathogenic sequences. Each model will be used to generate protein embeddings that will subsequently be applied to downstream classification tasks. The models will be compared based on their performance in distinguishing pathogenic sequences from reservoir-associated sequences using standard evaluation metrics. This study will provide a comparative analysis of transformer-based protein language models for viral pathogenicity prediction. Identifying which model will most effectively capture pathogenic sequence signatures may improve computational approaches for the early detection of high-risk viral strains and may contribute to the development of bioinformatics tools for emerging virus surveillance. 

The deployment of autonomous systems in underground mining operations is critical for enhancing personnel safety and the effectiveness of rescue missions in high-risk environments. This presentation will explores the sensing and hazard detection capabilities of the Husky Unmanned Ground Vehicle (UGV) and evaluates several detection methodologies, including deployable "egg-inspired" sensor nodes for environmental monitoring of gas and temperature and a theoretical "Temperature Net" for directional heat prediction. The primary focus is the development and testing of a "Distance and Temperature Point Grid" system, which utilizes a robotic arm equipped with a narrow field-of-view thermal sensor and a single-point LIDAR to sweep and localize thermal threats. Experimental results indicate that the arm-sweep method offers the most practical solution for active hazard avoidance, successfully identifying the distance and scope of heat sources. These advancements aim to improve the Husky UGV's situational awareness and autonomy in disaster-recovery applications. 

There is a pressing need for sustainable alternatives in concrete construction as the manufacture of Portland cement accounts for 7-8% of the world's anthropogenic CO2 emissions. Despite its high silica content and recyclability, millions of tons of waste glass are produced globally each year and dumped in landfills. In this study, WGP was examined as a potential partial substitute for Portland cement in both extrusion-based 3D-printed concrete and traditionally cast concrete. ASTM C109 techniques were used to make concrete mixtures with 0–30% WGP in order to assess the development of compressive strength under air-curing conditions. Compressive strength increased with increasing WGP content up to about 15%, which demonstrated the maximum strength among the tested compositions, according to the data. In order to comprehend changes in pore structure, hydration products, and microstructural evolution, representative mixes were further characterized using water absorption tests, FTIR, BET surface area analysis, Micro-CT, XRD, and SEM-EDS. Mixtures comprising up to 15% WGP were modified for extrusion-based concrete 3D printing in the study's second section since this range showed good mechanical performance. Stable extrusion and shape retention were achieved through iterative mix optimization, and microstructural and phase characterization techniques were used to further examine the printed samples. Overall, this work shows that WGP can be used as a sustainable cement substitute while also investigating its use in concrete additive manufacturing. It also provides insight into the strength development, microstructural mechanisms, and printability of concrete modified with WGP. 

This study presents a heat transfer analysis of an industrial rotary kiln (L = 50 m, D = 3.6 m) used for gold ore roasting at a Ghanaian processing plant. Using 15 operating observations, the peak roasting temperature (T_peak) is expressed as a function of fuel rate, ore feed rate, gas flow properties, kiln geometry, and surface emissivity.

Dimensional analysis via the Buckingham Pi theorem reduces 11 physical variables to 6 dimensionless groups, with T_peak·c_p·ṁ_ore/Ḃ_fuel as the dependent dimensionless temperature. Data confirms fuel rate as the dominant driver of peak temperature (R² = 0.99) and a tight coupling between peak temperature and thermal efficiency (R² = 0.97). The optimal operating window is identified as 780–790°C, corresponding to fuel rates of 85–89 L/hr.

Produced water (PW) generated during oil and gas production poses significant environmental and management challenges due to its large volumes and complex composition. It typically contains dissolved and dispersed hydrocarbons, high salinity, and trace metals, requiring rigorous treatment before discharge or reuse to protect water resources, ecosystems, and human health.

In this study, a Direct Contact Membrane Distillation (DCMD)–crystallization system was evaluated for high salinity PW desalination, followed by membrane adsorption for selective lithium recovery. PW from the San Juan Basin in southeastern New Mexico, with a total dissolved solids concentration of 120,000 mg/L, was treated using DCMD operated with a 60°C feed and a 20°C permeate stream. The system achieved a permeate yield of approximately 60%, producing a concentrated brine of 259 g/L and salt rejection greater than 99.9%, confirming robust separation performance under harsh conditions.

The concentrated brine was then processed using a functionalized membrane adsorbent for lithium extraction, yielding an adsorption capacity of 17.55 mg/g and a desorption capacity of 12.48 mg/g, indicating effective recovery and regenerability. Overall, the DCMD–crystallization and membrane adsorption approach offers a promising pathway for simultaneous water purification and critical mineral recovery from PW, supporting more sustainable oil and gas operations. 

This project is conducted under the guidance of Ali Barenji at New Mexico Tech. It presents the development of an autonomous robotic pit stop system capable of performing a pitstop with minimal human intervention. The primary purpose of this research is to design and validate an integrated robotic framework that combines perception, kinematics, and motion planning to automate a manipulator to complete a complex time-critical task.

We hypothesize that a modular robotic system equipped with vision-based object detection and specialized end effectors can reliably execute tire replacement tasks in a structured environment with accuracy comparable to human operators. To test this hypothesis, the system is designed to detect and localize tires, wheel hubs, tire racks and tools in order to navigate between predefined workstations within an assembly area.

The experimental plan involves implementing the system in the RViz and Gazebo simulation environment where perception, planning and control modules are integrated using ROS2 and evaluated. A nut runner end effector is used to tighten and loosen a centerlock nut while a tire-handling end effector enables grasping and transporting of tires. The work is divided among team members: kinematic modeling (An Nguyen), software integration (Gifty Quayson), visual perception (Fahad Mannan) and motion planning (Seth Diaz) using MoveIt2. Preliminary results demonstrate that the system can successfully identify key components including planning feasible trajectories and execute tire manipulation tasks within simulation. These findings support the feasibility of autonomous pit stop operations and highlight the potential for robotic systems in high-speed precision-critical applications.

This research aims to propose a hypothesis about a sustainable drone vertiport system for Mars which will be feasible to perform the exploration and data collecting tasks for a longer period. In this hypothesis we plan to deploy the drone vertiport with adjustable landing gears from a certain height containing drones and equipped with renewable power generation system with the help of solar energy. In this case, the vertiports will be aided by temporary parachutes and thrusters which will guide the vertiport to have a vertical and proper landing. The vertiport will be equipped with weather station, solar panels for generating power to support the system, and capsules with contact-based charging systems to charge and house the drones. 

 Understanding how and why rocks fail is essential for improving safety in underground

mines, where sudden rock collapses or seismic events can put workers and infrastructure at

risk. In many cases, failure occurs along faults, which are natural weak surfaces in the rock

where movement can happen.  This study investigates how the geometry of these faults, whether a single fault or a system of interacting faults, controls how rocks slip under stress. Laboratory experiments were performed on rock samples subjected to confining pressure to simulate underground conditions. During testing, acoustic emissions were recorded. These signals are similar to very small “micro-earthquakes” and provide early clues about internal damage in the rock before visible failure occurs. Because these signals are complex and numerous, we used clustering, a machine learning technique that automatically groups similar patterns, to identify different stages of rock deformation.

The results show that more complex fault networks produce more variable and distributed

damage, while simpler faults lead to more concentrated and predictable failure.

Importantly, when complex fault systems fail, they release greater amounts of energy

compared to single-fault systems, indicating a higher potential for destructive events. This

work demonstrates how combining laboratory experiments with data-driven methods can

improve early warning systems and stability assessment in underground mining and other

geotechnical applications.

Mechanoluminescent emission in ZnS-PDMS composites was investigated as a strain-sensing method because it offers a way to measure deformation through emitted light rather than attached sensors. High-speed imaging was paired with digital image correlation (DIC) so that optical emission could be directly related to measured strain during cyclic loading. The video data were processed into sequence-based features, including RGB intensity and frame-to-frame gradients, to capture both intensity changes and the evolution of emission over time. A preliminary model showed that the optical signal contained enough information to recover overall strain trends. To better represent the non-uniform deformation of elastomeric materials, a final spatially resolved, amplitude-aware LSTM model was developed using DIC-aligned strain data. The model captured cyclic strain behavior, phase progression, and the characteristic multi-peak emission response during loading and unloading, while also retaining predictive capability for previously unused strain conditions. These results show that mechanoluminescent materials can serve as selfindicating strain sensors and that data-driven modeling provides a practical way to link optical response to mechanical deformation in ZnS-PDMS composites. 

In this paper, we present novel mechanical computing logic gates, which can provide real-time photonic feedback during computational operations, by designing mechanical-luminescent metamaterial (MLM). Finite element analysis is performed to optimize the design of MLM, and candidate MLM prototype logic gates (i.e., AND, OR, BUFFER) are fabricated by printing a customized mechano-luminescent (ML) zinc sulfide (ZnS)-functionalized resin using a stereolithography 3D printer. One color camera is employed for multiphysics analysis of the MLM prototypes to establish direct connections between mechanical forces and the resulting photonic responses of logic states. The method enables mechanical computing systems to detect errors and verify states independently through self operating processes which work in conditions that exceed electronic sensor limitations. Mechanical computing systems with self-aware capabilities emerge through the combination of ML materials with metamaterial structures. 

Underground mine pillars are subjected to sustained and cyclic stresses that can cause time-dependent deformation, progressive damage, and eventual instability. This study compares sandstone behavior under creep, fatigue, and alternating creep-fatigue loading to determine which condition causes the earliest failure, which produces the greatest damage at failure, and whether cracking stages can be identified using unsupervised clustering. Specimens were tested with continuous acoustic emission (AE) monitoring, and mechanical response, AE parameters, and b-value were analyzed. K-means clustering was applied to AE parameters to identify damage populations. Results show that fatigue caused the earliest failure and produced the highest AE activity, indicating the most intense rock damage. Clustering revealed three common damage populations across all loading conditions: initial compaction, subcritical crack growth, and critical crack growth. Fatigue and alternating creep-fatigue also showed an additional unloading-related population, while alternating creep-fatigue further divided the low-damage regime into two populations, interpreted as initial compaction and re-compaction. These results highlight the strong influence of loading path on time-dependent damage evolution in rock. 

This project is inspired by how dandelion seeds stay aloft for extended periods of time using a light and porous shape that creates a stable and detached vortex ring above the seed. The main question is how geometry and porosity, through their control of attached versus detached wakes, set the strength of thermal forcing on drag and unsteadiness at low Reynolds numbers. An axisymmetric porous‑media model of a true dandelion seed uses a Darcy–Brinkman description with non‑isothermal Boussinesq flow. A three‑dimensional model of a dandelion‑inspired and radially slotted conical frustum uses a buoyant low‑Mach solver with both steady Reynolds‑averaged and transient large‑eddy simulations. The natural model supports a detached vortex ring whose drag and structure change only weakly when the seed is heated or cooled. This suggests a wake that is robust to thermal forcing from, for example, solar heating that depends on coloration. In contrast, the bioinspired frustum generates an attached wake that remains closely coupled to the body. Therefore, changes in ambient temperature and thermally driven density differences from surface coloration can strongly reorganize the wake and cause large changes in drag and unsteady motion. The main conclusion is that geometry‑induced wake topology controls how effective thermal forcing can be. In summary, engineered devices can be designed either to resist environmental heating or to use it as a tuning mechanism for performance and stability. 

Hydrogen sulfide is a highly toxic and corrosive gas commonly encountered in oil and gas operations, posing serious risks to human health, infrastructure, and the environment. Due to these hazards, stringent regulations limit the venting and flaring of H₂S in many jurisdictions, promoting alternative management strategies such as acid gas injection, where H₂S is co-injected with carbon dioxide into subsurface formations for safe disposal and potential long-term storage. This study investigates the effects of reservoir conditions on acid gas–brine–rock interactions, with a focus on their influence on CO₂ mineralization processes. A numerical core-scale modeling approach is employed to evaluate the coupled effects of varying H₂S–CO₂ mole fractions, brine salinity, reservoir pressure, and temperature on reactive transport behavior. The simulations capture key geochemical processes, including mineral dissolution, precipitation, and changes in fluid composition. Results indicate that the presence of H₂S significantly modifies reaction pathways and kinetics compared to pure CO₂ systems, influencing both mineral assemblages and trapping mechanisms. Variations in salinity, pressure, and temperature further impact the extent and rate of mineralization, highlighting the importance of site-specific reservoir conditions in predicting long-term storage performance. This work provides a core-scale perspective on the geochemical behavior of acid gas systems and offers insights for optimizing AGI operations while ensuring environmental safety and storage integrity. 

Ultrasonic testing enables nondestructive detection of internal voids in structural materials, however mining environments pose unique challenges. High signal attenuation in cemented ore bodies and pervasive electrical noise from heavy equipment limit the effectiveness of the conventional commercial instruments. This research addresses the need for a robust, application-specific platform capable of reliable void detection under these demanding conditions. The hypothesis is that a custom hardware platform utilizing high-voltage bipolar excitation and a purpose-designed low-noise receiver can achieve sufficient signal penetration and noise rejection to enable void detection in mining-relevant materials. To test this, a modular platform is being developed. The transmitter employs a full H-bridge driver delivering 100Vpp bipolar pulses in 8-cycle bursts at 40 kHz to maximize acoustic energy transfer. The receiver chain comprises a transmit/receive protection stage, a low-noise amplifier, and a multi-stage bandpass filter centered at 40 kHz to suppress out-of-band interference. Experimental validation uses high-bandwidth oscilloscopic measurement to verify subsystem performance against design specifications. Preliminary results confirm stable pulse generation with consistent amplitude and burst characteristics. Receiver front-end characterization are expected to demonstrate effective noise filtering and amplification suitable for subsequent integration. This work is expected to establish a validated hardware baseline for future void detection experiments. The modular, low-cost platform will contribute to the field of nondestructive evaluation by demonstrating a pathway to application-specific instrumentation for challenging industrial environments. 

Inertial Measurement Units (IMUs) are instruments used in navigation that provide specific force and angular velocity measurements using an accelerometer and a gyroscope. These measurements are integrated over time to estimate a body's changing position, velocity, and attitude. While IMUs have good short-term performance, the integration of these noisy and biased measurements lead to estimations diverging over time. This leads to the use of external measurements from systems such as Global Navigation Satellite Systems (GNSS) to provide highly accurate readings that converge system estimations. This project will utilize a navigation simulation to consider a body equipped with multiple IMUs traveling along a trajectory. The system will apply convex optimization methods to select an optimal weight for weighted sensor fusion. Using an optimal weight for the measurements will allow calculations to be done with a higher degree of accuracy, thus improving navigation estimations. Constraints will be set on the system so that the true values of the IMU measurements are known, allowing for measurement accuracy can be calculated. The convex optimization problem will be formed on this premise. This can be done during stationary initialization, where both the angular velocity and specific force are known. The optimization will be performed on a per-axis basis, allowing for weighting of each individual measurement component. The ultimate goal of the project is to utilize simulation to test the effects of utilizing convex optimization for IMU-fusion on a navigation system.