116964 research outputs found
Sort by
Ugh, Caradeaux : the trials and triumphs of staying true to your vision during revision
No full textThe following report describes the writing process for the feature film Oh, Caradeaux! including early draft comparisons, outlines, feedback, and a full feature draft. The report also details the conception of the idea as well as films that influenced the tone of the script, and a reflection of the various drafts of the story.Radio-Television-Fil
Sacred springs and places public : water, architecture, and Roman imperialism in Gaul
No full textThis dissertation studies how architectural changes made to indigenous settlements in Gaul in the first century BCE reflected changing local identities and served to remove local cultural and social attachments in favor of cultivating more Roman ones. I focus on two urban features that are not typically studied in conjunction: sacred springs and civic centers. I argue that separating the impact of these spaces on local lives engages in a binary of sacred versus secular which did not exist in the ancient world, and that by examining them in conjunction allows us to better understand the dynamic experience of local Gauls under Roman imperial control. Using two case studies, Bibracte and Glanum, I first create a diachronic survey of architectural changes made to their primary sacred springs and civic centers, before discussing how these architectural changes impacted the way locals could use and interact with the spaces. Next, I analyze to what extent we can recreate behaviors at each site; in particular, I find that evidence from the sacred springs indicate a continuity of religious rituals even in the face of major architectural disruption, while the behavioral impact on civic spaces is harder to reconstruct with our extant evidence. Each case study ends with an exploration of how the experience of these spaces changed following these architectural changes, and what impacts this had on local identity formation, especially because most of these major urban changes made at Bibracte and Glanum were facilitated by local elites as opposed to Roman intervention. I argue that any deliberate destruction and rebuilding of city structures in this context constitutes a type of cultural violence, therefore changes driven by local agents are still a facet of Roman imperialism. I conclude by comparing Bibracte and Glanum and how their unique histories with Rome and Roman conquest affected their urban fabrics, and end with a call for a continued focus on this topic by expanding the case studies to include other comparable sites around Roman Gaul to better assess the behavioral patterns of local communities under Roman imperial control in the first century BCE.Classic
Robust optimal task planning for autonomous mobile robots to maximize battery life
No full textThis thesis presents a comprehensive optimization framework for autonomous mobile robots (AMRs) aimed at minimizing battery degradation while ensuring efficient task completion under both deterministic and uncertain conditions. Battery degradation is modeled considering cycling losses during charging/discharging and calendar aging during idle periods. The framework addresses the inherent bilinear nature of the optimization problem through advanced linearization techniques, including piecewise linear approximations and McCormick envelopes, transforming NP-hard problems into tractable mixed-integer linear programs (MILPs). The research is structured in three progressive phases. First, a constrained optimal planning approach is developed using a rectangle method for piecewise linearization, focusing on deterministic scenarios to balance SOC management, charging rates, and task execution. Second, an enhanced deterministic model incorporates McCormick envelopes for tighter relaxations, with detailed numerical validation in single-task and warehouse environments. Third, the framework is extended to handle uncertainties in task intervals and distances via a two-stage robust optimization with chance constraints and sample average approximation (SAA), ensuring robustness against real-world variations.Mechanical Engineerin
A closed-chain approach to generating affordance trajectories
Get PDFThis thesis is motivated by Los Alamos National Lab’s efforts to develop autonomous robotic solutions for alpha radiation surveys by addressing the challenge of performing contact tasks and manipulation. Robots operating in unpredictable environments such as those present during radiation surveys require versatile hardware-and-task-agnostic frameworks to adapt to various tasks. A recent promising approach to address this challenge involves utilizing Screw Theory to model affordances. However, this framework operates online, computing end-effector velocities in real-time for small time steps, consequently lacking a comprehensive overview of the joint trajectory towards the goal. Additionally, it does not offer a method to control the gripper orientation along the affordance path. To address these limitations, we propose a novel framework that generates complete joint trajectories to execute affordances in real time. Our method treats the affordance and manipulator together as a closed-chain mechanism and introduces a new approach to solving closed-chain inverse kinematics. The closed-chain affordance model encapsulates task constraints and simplifies task definition. The framework is versatile, hardware-and-task-agnostic, grasp-invariant, and offers intuitive gripper orientation control along the affordance path. We validate our work through simulations of a UR5 robot and real-robot implementation on a manipulator-integrated Boston Dynamics Spot robot. For the tested real-world contact tasks (including moving a stool, opening a drawer, turning a valve, etc.) experiments demonstrate our ability to generate complete joint trajectories within a range of 0.008s to 0.35s, even in cases that involve precise gripper orientation control.Mechanical Engineerin
Engineering services in a mission critical environment : Engineering Services – Science and Technology Operations’ infrastructure support at Los Alamos National Laboratory
Get PDFAs an engineering team within a facilities-driven organization, Engineering Services – Science and Technology Operations (ES-STO), supports Los Alamos National Laboratory (LANL), playing a pivotal role in the U.S. nuclear stockpile mission. This report outlines ES-STO’s contributions through the installation of crucial systems such as HVAC units, compressors, and scientific specialty equipment, as well as providing expert consultation to optimize laboratory operations. ES-STO’s goal is to ensure that LANL's infrastructure and research facilities are aligned with mission-critical needs, supporting both operational efficiency and safety in the nuclear stockpile management and maintenance. This report discusses the installation processes, ongoing consultations, and the significant impact of our efforts on national security objectives.Mechanical Engineerin
Eagle Ford Miscible Fluid Injection Pilot Project Evaluation and Economic Evaluation Under Different Utilization Ratios
Get PDFEOR (Enhanced Oil Recovery) projects in the Eagle Ford formation have proven to be a cost-effective strategy
to offset incremental drilling costs, which average around $1 million per well (Phinds, 2021). By late 2024, miscible
fluid injection—primarily using natural gas—had been implemented across 40 leases operated by nine companies.
The success of these projects depends heavily on a clear understanding of production profiles and PVT (pressure-
volume-temperature) phase behavior (Grinestaff et al., 2020). This study developed a robust workflow to evaluate
production profiles before and after injection. Pre-injection Estimated Ultimate Recovery (EUR) was used to establish
a baseline natural decline, while post-injection EUR was assessed to measure production uplift.
To account for the complexities of lease-level production—including variable well start dates, shut-ins, and short
production histories—we conducted lease-level analysis to ensure compatibility with reported injection data and avoid
uncertainties from production allocation. We also examined basin-scale phase behavior variations and operational
factors such as injection volumes and changes in neighboring wells' gas-oil ratio (GOR) to better understand the
drivers of EOR performance and optimize future implementation.
EOR success is highly sensitive to oil and gas prices as well as utilization ratio (Jiang et al. 2019).. Traditional
definitions of utilization ratio—based on the fraction of recovered oil from the original oil in place—are not applicable
in unconventional plays due to uncertainties in in-place volumes, well stacking, and horizontal development
complexities (Magomadov et al. 2022). Therefore, we define a modified Utility Ratio (UR) as the ratio of recovered
oil to injected gas volume during the EOR period. This metric serves as a key indicator for performance evaluation.
A low UR raises important operational and economic questions: Could it reflect increased reservoir contact? Does it
add reserves or only boost short-term production? How does it impact Net Present Value (NPV)? And does EOR
implementation influence neighboring wells through GOR anomalies, suggesting altered drainage volumes?
To investigate these questions, we first analyzed UR scenarios using a subset of Eagle Ford leases with reported
annual injection and production data. We then assessed nearby wells within a defined radius to identify GOR changes
attributable to EOR activities. An improved decline curve analysis (DCA) workflow was developed—grounded in
literature and validated through our own tests—to separately evaluate pre- and post-injection trends and quantify
production uplift. We supplemented this with geological and reservoir properties from an in-house Eagle Ford geomodel, as well as completion data from S&P Global, to analyze the effect of these parameters on EUR both with
and without EOR.
For economic evaluation, we integrated historical and projected oil and gas prices using EIA scenarios to calculate
the NPV of each lease. Our analysis considered both optimistic and conservative operational conditions to determine
whether EOR implementation exceeded or underperformed compared to baseline declines. Results indicate that oil
EUR uplift due to EOR ranged from -41.5% to 69.6%, with an average of 10.1% and median of 8.4% based on
hyperbolic decline analysis. High UR leases consistently delivered positive NPV outcomes, while low UR leases were
more sensitive to commodity prices. The primary driver of NPV performance was the degree of production uplift,
supplemented by additional tax credit benefits from increased oil production.
This integrated technical and economic evaluation underscores the importance of factors such as injection volume,
utilization ratio, and access to previously untapped rock volumes. The findings provide actionable insights for
improving EOR planning, implementation, and overall performance across unconventional reservoirs.Bureau of Economic Geolog
Doctoral thesis recital (tenor trombone)
No full textUnable to determine.MusicName of supervisor not provided
Surrogate Modeling of Cell Dynamics in Agent-Based Models Using Transformers and PINNs
No full textAgent-based models (ABMs) capture individual behaviors and interactions, enabling the study of complex dynamics at the cellular scale. However, as the number of agents increases, computational cost rises sharply, limiting their feasibility for large scale studies in computational oncology. To address this, we compare transformers and physics-informed neural networks (PINNs) as surrogate models to replace the cell-movement component of our ABM. Since our ABM can generate unlimited training data, we evaluate each method's ability to generalize by testing predictions on unseen inputs. PINNs incorporate the underlying physics involved in the movement of cells, including the balance of forces acting on each cell. Transformers leverage attention mechanisms to identify spatial correlations and determine which neighboring cells influence a given cell’s motion. For both methods, we adjust key hyperparameters, including learning rate, number of layers, epochs, attention depth (for transformers), and loss functions (for PINNs). As we evaluate two modeling approaches across numerous hyperparameter combinations, the training process becomes computationally intensive and impractical to run on a personal computer. Access to TACC’s Lonestar6 system allowed us to perform these experiments efficiently using GPU nodes equipped with NVIDIA A100s, significantly accelerating our progress. Preliminary results on an unseen validation dataset demonstrate that the transformer model accurately identifies the neighborhood influencing each cell’s movement. On average, the predicted cell position deviates by only 13.2 ± 2.2 % of the cell radius (1.32 ± 0.22 µm from the ground truth). Similarly, the PINN-based approach effectively captures both the balance of forces and maximum cell velocity, achieving an average deviation of 25.4 ± 2.9 % of the cell radius (2.52 ± 0.29 µm) on the same validation set. These results support the feasibility of both modeling strategies. Ongoing work explores generalizability to higher cell counts than used in training. These methods provide a scalable framework for accelerating ABMs in oncology and other domains involving Newtonian motion.Texas Advanced Computing Center (TACC
Quantifying the distribution of abiotic transformation rate constants in low permeability clay zones for improved assessment of TCE impacts to groundwater at DoD field sites
No full textBack diffusion of trichloroethylene (TCE) from low-permeability zones (LPZs) remains a significant obstacle to achieving groundwater remediation goals at Department of Defense (DoD) sites. TCE in LPZs can gradually diffuse into adjacent and more permeable layers, leading to mass discharge rates that cause extracted groundwater to exceed maximum contaminant levels (MCLs). Previous studies have demonstrated that under anaerobic conditions TCE can be reduced to gases such as acetylene and butane by reduced iron minerals, while under aerobic conditions it can be oxidized to organic acids such as glycolic and formic acids via hydroxyl radicals. Hence, natural attenuation is suggested to be a plausibly promising long-term strategy to mitigate TCE back diffusion concerns in many contaminated sites. Building upon prior work, this study investigated abiotic TCE transformation in two low-permeability clays, St. Louis Ordinance Plant and Longhorn Army Ammunition Plant clays, by monitoring a 10-week diffusion vial experiment. Decreasing TCE concentrations, as well as the accumulation of transformation products confirmed abiotic dechlorination in both clays. Ferrous iron speciation showed spatial patterns suggesting its involvement in TCE transformation, with structural ferrous iron likely acting as the dominant long-term reductant. Some transformation products remained unidentified, potentially due to further transformation of organic acids to CO2 or reduced gases. Future research is needed to explore the potential for organic acid conversion to reduced gases in the presence of ferric iron and hydroxyl radicals to further elucidate mineralogical effects on TCE abiotic transformation pathways.Environmental and Water Resources Engineerin
Human Cognition: Powered by AI?
No full textWorks of science fiction frequently portray the notion of characters that operate as half human and half computers- denoted “cyborgs”. Imagine a world where these sorts of individuals exist and operate in society with superhuman abilities, both in regards to physicality and
cognition. Although this may sound far-fetched, there is no need to visualize such a concept; these sort of “cyborg-like” individuals already exist and may be ubiquitous in our future. In addition, their abilities may actually encompass a wide range of science fiction themes, including telepathic communication, movement, and interaction with reality. In other words, the ability to perform actions using thoughts is a proven reality, and it will only be built upon and perfected in the years to come.
The intersection of artificial intelligence (AI) with neuroscience has facilitated the development of brain-computer interfaces (BCIs), a revolutionary technology that integrates the human nervous system with external devices, softwares, and mechanical systems in hopes of enhancing motor and cognitive function.Polymathic Scholar