Published ETD Collection

Permanent URI for this collectionhttps://hdl.handle.net/10657/2

Browse

Recent Submissions

Now showing 1 - 20 of 5909
  • Item
    Inflammation and Wound Healing in the Mouse Cornea
    (2023-12) De La Cruz, Angie S; Burns, Alan R.; Redfern, Rachel; Harrison, Wendy W.; Rumbaut, Rolando E.
    Corneal abrasion elicits an inflammatory cascade prompting inflammatory cells to exit the vasculature and accumulate at the limbus. Extravasated platelets, being non-motile, remain at the limbus while neutrophils migrate toward the wound. Lam and colleagues showed platelets are crucial for wound healing, suggesting CD18 as a possible mechanism for platelet extravasation. Other studies suggest platelets promote macrophage phagocytosis changing macrophage phenotype (M1, pro-inflammatory); potentially influencing the course of inflammation. The limbus has a large population of perivascular macrophages suggesting possible phagocytosis of platelets after injury. The relative distribution of macrophage phenotype within the cornea and contribution to the inflammatory cascade remains unclear. The Aims of this research were: 1) Determine the role of CD18 on platelet extravasation in this mouse model of inflammation, focusing on two relevant cell types that express CD18: PMNs and mast cells 2) Determine if macrophage numbers, phenotypes and distributions change to reflect the pro-inflammatory state of the injured cornea during the first 24 hours following a central epithelial abrasion. Following corneal abrasion, we discovered extravascular RBCs alongside extravasated PMNs and platelets. Platelet and RBC extravasation requires adequate levels of CD18, evidenced by reduced extravasated platelets and RBCs in mice with low CD18 expression. Ultrastructural observations of engorged limbal venules showed platelets and RBCs passing through endothelial pores. Venule engorgement is reduced when mast cell degranulation is absent or reduced. Mast cell-deficient mice and mice with antibody-induced depletion of circulating PMNs showed reduced venule engorgement, and extravasation of PMNs, platelets, and RBCs. Extravasated RBCs and platelets are phagocytosed by perivascular macrophages. The overall distribution of macrophages after injury decreases at the limbus and increases paralimbus to wound center. Reduction of limbal macrophages correlated with decreased M2 macrophages in the anterior stroma. This dissertation introduces RBCs and macrophages as active participants in the inflammatory response. We conclude in the injured cornea: (1) platelet and RBC extravasation depends on CD18, PMNs, and mast cell degranulation (2) macrophage redistribution occurs due to decreased M2 macrophages at the limbus (anterior stroma), followed by increases in M1 and M2 phenotypes across the cornea.
  • Item
    Modeling and Stability Analysis of Parallel Connected Inverters in Micro-grid Systems
    (2023-12) Gaddala, Ravi Kumar; Rajashekara, Kaushik; Krishnamoorthy, Harish S.; Huang, Hao; Fan, Lei; Pan, Miao
    In the future grid, more distributed generation resources will replace traditional synchronous generators. As a result, smaller, localized generation units are integrated near the consumption sites, resulting in the use of inverter-based generation to meet voltage and frequency requirements. It is important to note that, despite the many advantages of inverters, they are vulnerable to stability and resonance problems in many applications, including rooftop solar inverters, MW-scale solar inverters, electric traction systems, offshore wind farms, and high-voltage direct current transmission lines. For these reasons, modeling of power electronic converters, particularly inverters, has become an increasingly important part of analyzing the potential impact of their widespread adoption on future power grids. Furthermore, in order to establish a resilient and reliable grid infrastructure, it is crucial that investigations are conducted in advance, focusing on resonance and stability phenomena. In response to these challenges, this thesis proposes small signal DC impedance models for a grid-tied converter operating in an open loop, closed loop dq current control, and DCLV (DC link voltage control) control loop with consideration of PLL dynamics. It is noticed that both the magnitude and phase plots of DCIMs show surges and dips around the PLL bandwidth. Because of this, when DCIMs interact with DCNIs close to the PLL bandwidth, they cause oscillations in the DCLV, which results in unstable operation of the converter. The research presented in this thesis is further extended by proposing impedance models (i.e., observing from the DC side) for grid-forming converters' open-loop and closed-loop load voltage control. DC link voltage stability analysis has been investigated using the proposed DC impedance models. Based on the impedance stability analysis, it can be concluded that the variations in the parameters of the outer loop controller are responsible for the unstable operation of the grid-forming converter. In addition, it is evident that the changes in DCNI (Equivalent impedance of converters connected at the DC network terminals) result in the instability of the overall system. This thesis also proposes a comprehensive state space modeling and Eigenvalue-based stability analysis of islanded microgrids that incorporate all control loops and parameters of the networks to ensure equal and proportional reactive power sharing. Also, the proposed modeling and analysis accurately predict the effect of various control loop parameters on the stability of an islanded AC microgrid. Additionally, the limiting values of different control parameters are presented, facilitating the design of effective control strategies. The proposed modeling and analysis are validated using Typhoon Hardware in-the-loop testbed, and the results are presented. Finally, this thesis developed and presented detailed small-signal state-space models specially designed for islanded microgrids with IMC. Time-domain simulations have validated these models to ensure their fidelity and applicability. The proposed state-space models were tested for robustness by performing a load change test. Furthermore, a control strategy that combines line impedance compensation with virtual impedance control has been introduced to achieve equal active and reactive power sharing among converters, particularly in the presence of differing cable impedances.
  • Item
    High Performance Sensorless Controlled PMSM Drive with Long Cable for Subsea Applications
    (2023-12) Singh, Virendra; Rajashekara, Kaushik; Krishnamoorthy, Harish S.; Huang, Hao; Chen, Jiefu; Fan, Lei; Selvaraj, Goutham
    The increasing interest in subsea oil and gas as an energy source has led to a demand for highly efficient and reliable motors and their control. These motors are essential components of subsea applications like drilling, pumping, and boosting, which are necessary to extract natural gas and oil. The Permanent Magnet (PM) motors, known for their higher efficiency and high-power density, are increasingly preferred over traditional induction motors in these applications. It is important to consider that subsea systems, characterized by the inclusion of a sinewave filter, transformer, and long cable experience several control challenges. The voltage drops are a common issue across these components, while the passive elements of the sinewave filter specifically introduce magnitude & phase shifts, and the transformer is susceptible to core saturation problems. This dissertation presents innovative sensorless control strategies for PM motors to address the complexities associated with subsea system components. These strategies address the full spectrum of operational speeds while considering the constraints imposed by the sinewave filter, transformer, and the long cable. The sensorless starting of a PM motor is one of the key control challenges under heavy loads due to resistive voltage drops across the system components and transformer core saturation. In this thesis, an enhanced Volts-per-Hertz (V/Hz) control strategy that compensates for voltage drops and avoids transformer core saturation, ensuring reliable startup even under substantial loads is proposed. Moreover, the issue of temporary reverse speed at startup is another challenge that can cause the loss of synchronism. This has been addressed by incorporating an Initial Position Detection (IPD) methodology paired with V/Hz and a voltage compensation technique to accurately estimate the rotor's initial position to start the motor without speed reversal. Additionally, the accuracy of the closed-loop sensorless vector control highly depends on the system parameters. To address this, a Model Reference Adaptive System (MRAS) based online parameter estimation technique is incorporated to adjust the control variables while the motor is in operation. A multi-loop sensorless vector control technique is also employed to mitigate the effect of the sinewave filter, ensuring the system's stability. The final control scheme relies on High-Frequency Signal Injection (HFSI) based position estimation to start the motor in a closed-loop vector control from zero speed. The HFSI-based estimation technique must also account for various challenges associated with the sinewave filter and cable. As a low-pass sinewave filter is connected at the inverter terminal, which necessitates the careful selection of the injection frequency to prevent loss of signal information. The effectiveness of these control strategies for PM motors is validated through controller hardware in the loop (C-HIL) real-time simulations using Typhoon HIL-604 and Texas Instruments digital signal processor.
  • Item
    Wireless System for Long-term and Real-time Subsurface CO2 Monitoring
    (2023-12) Li, Xiaoliang; Shan, Xiaonan; Chen, Jiefu; Jackson, David R.; Mo, Yi-Lung; Wang, Hanming; Shan, Xiaonan
    The data and power transfer systems for long-term underground CO2 sequestrationmonitoring are normally based on wire-line cable, which will lead to a high potentialleakage path through casing and cement annulus in high-temperature, high-pressure, andhash underground environments. In this dissertation, a novel wireless communication andpower transfer system has been developed for real-time underground CO2 monitoring. Thesystem includes an array of toroidal transceivers winding around the highly conductivecasing string for wireless power transfer to the deep subsurface, which helps to maintainwell integrity and reduce potential leakage by eliminating the need to perforate the casingor an umbilical in the cement annulus. The metal casing’s amplification effect significantlyenhances the wireless power transfer efficiency and communication performance, whichprovides a highly conductive power/electric current pathway instead of omnidirectionalwireless radiation loss in the subsurface. The toroidal transceiver’s design has beenoptimized to improve the received signal, and our results show significant improvementsin wireless power transfer efficiency. Using the optimized design, we can receive 1 to 10 %power transfer efficiency at 800 meters deep using only one toroidal transceiver with 1Acurrent as input. Compared with other wireless antenna designs, such as the helix coilantenna, our system has shown 26,000 times power transfer efficiency improvement. Usingthe optimized design, for this 800 m long subsurface wireless system, the channel capacitycould improve about 14,000 times from 0.35 bps to 5 kbps, and the energy efficiencyimproved 109times from 10-3bit/J to 106bit/J. For experiments, a lab-scale system is built,and our experimental receiving voltage measurements support the simulation results. Thisscaled-down wireless communication system with USRP as transceivers was tested atvidifferent symbol rates and the bandwidth from the voltage signal spectrum andconstellation figures were compared and matched.Furthermore, we tried to use this system not only for leakage monitoringcommunication but also for CO2 migration monitoring. We tested the model with andwithout the CO2 plume, we could see the difference response from the voltage signal. Atthe same time, we build a new model in COMSOL to do the time-domain reflectometry(TDR), the difference of properties can also be detected.
  • Item
    Multiport Energy Router based Integration of Renewable Energy Sources for Offshore Grid
    (2023-12) Karbozov, Arnur; Rajashekara, Kaushik; Krishnamoorthy, Harish S.; Li, Xingpeng; Shi, Jian; Nguyen, Hien Van
    Integrating Renewable Energy Sources (RES) is a potential solution to improve the reliability and cost of the electrical offshore architecture for harnessing deep-water hydrocarbon resources. However, conventional subsea electrical architectures based on HVAC transmission and distribution require bulky line frequency transformers and have reactive power losses in the system that increase with the longer step-out distance. In turn, subsea HVDC cables are more efficient, lighter, and cost-effective per unit of transmitted power for long distances than HVAC cables. Furthermore, integrating RES via HVDC or MVDC systems requires voltage regulation without reactive power compensation. However, a DC microgrid with several power sources needs a proper architectural design. To address this issue, the dissertation proposes an HVDC offshore architecture that utilizes wind turbines, battery storage systems, and an onshore grid to power subsea oil and gas loads. The architecture is based on the proposed Multiport Energy Router (MER), a modular DC-DC converter that interconnects the sources above. MER is realized by a series-parallel combination of multiport DC-DC Triple Active Bridge (TAB) converters. The converters are critical to enhance the proposed system's power density, reliability, and efficiency. The TAB modules in MER have an inherent capability of reaching zero voltage source (ZVS) switching due to the port inductor or leakage inductor transformer. This feature assists in eliminating the turn-on losses along with the EMI issues. However, the soft-switching can be compromised depending on port voltages and operating power region variation. To address this issue, a control algorithm for efficiency optimization of TAB by selecting optimal control variables is proposed. The algorithm improves the TAB converter efficiency by extending the ZVS range and minimizing the circulating current in the transformer windings. The TAB converter in MER is a multi-input multi-output (MIMO) system. The conventional PI controller for TAB utilizes a static decoupling matrix, which introduces potential operational challenges for the converters. The dissertation proposes a Model Predictive Control (MPC) for TAB, which is preferred for MIMO systems. The proposed MPC controller can handle fast dynamics, incorporate various constraints, and have straightforward digital implementation.
  • Item
    Understanding the Dynamics of Complex Nanoparticle and Polymer Solutions Using Molecular Simulations
    (2023-12) Kotkar, Shivraj Bhagwatrao; Palmer, Jeremy C.; Conrad, Jacinta C.; Robertson, Megan L.; Howard, Michael P.; Poling-Skutvik, Ryan
    Understanding nanoparticle dynamics in polymer solutions holds significance for drug delivery and enhanced oil recovery applications. Deviation from the generalized Stokes-Einstein relation occurs when nanoparticle and polymer sizes are comparable. We employ hybrid molecular dynamics-multiparticle collision dynamics (MD--MPCD) simulations to investigate nanoparticle dynamics in semidilute solutions of ring and linear polymers. Nanoparticle diffusivities agrees with predictions from a polymer coupling theory [Cai, Panyukov, and Rubinstein, Macromolecules 44, 7853 (2011)], indicating coupling to segmental relaxations for both polymer architectures. Short-time nanoparticle dynamics exhibit subdiffusive behavior, deviating from coupling theory, instead closely tracking polymer subdiffusive exponents. The strong coupling of nanoparticle dynamics to polymer center-of-mass motions holds for both architectures. We also explore the impact of ring polymer stiffness on nanoparticle dynamics, observing deviations from coupling theory predictions and a vanishing coupling between nanoparticle dynamics and polymer center-of-mass motions with increased stiffness. In our subsequent study, we delve into the dynamics of polymers grafted onto spherical nanoparticles. Mean-square displacements of monomers near the grafting surface show an intermediate plateau, signifying confined dynamics akin to neutron spin-echo experiment reports. This confined dynamics disappears beyond a specific radial distance from the nanoparticle surface, dependent on polymer grafting density. We demonstrate that this dynamical confinement transition adheres to theoretical predictions for the critical distance associated with the structural transition from concentrated brush regime to semidilute brush regime.
  • Item
    High-Strength Reinforcing Bars (HSRB) and TMS 402: A First Study on Grade 80 (550 MPa) Bars
    (2023-12) Khalid, Omar Nazar Abdulhamid; Kalliontzis, Dimitrios; Belarbi, Abdeldjelil; Mo, Yi-Lung; Baxevanis, Theocharis
    In 2019, ACI 318 introduced a major code change by incorporating the usage and applications of high strength steel reinforcing bars (HSRBs). The adoption of HSRBs was motivated by several factors, including the reduction in steel congestion and construction cost, gain in member strength, and reduction in the carbon footprint of reinforced concrete buildings. However, code adoption of HSRBs has lagged in masonry design, which is attributed to the absence of relevant analytical and experimental data. Currently, TMS 402/602 defines the maximum allowable design stress for reinforcing bars at 60,000 psi. This research presents the first study on using Grade 80 (550 MPa) bars in the context of TMS 402/602. An analytical investigation was performed to examine the feasibility of using these bars with consideration to the ASD/SD provisions and the potential for reduction in material weight and cost in comparison with Grade 60 bars. The analytical investigation was followed by an experimental study of lap-splice tests for Grade 80 ASTM A615 and ASTM A706 reinforcing deformed bars. The experimental program included twenty-two contact lap-splice experiments in concrete and clay brick masonry, with test variables being the bar size, the lap-splice length, and the ASTM designation of the bars. Experimental and analytical results corroborated the technical feasibility of embedding Grade 80 bars in structural masonry with reasonable lap-splice lengths, following revisions to the design formulas of TMS 402/602. The need for future research is emphasized toward testing of large-scale masonry members reinforced with Grade 80 bars in flexure and shear.
  • Item
    Phase and Rheological Behavior of Colloidal Particles with Polymer-Induced Bridging Interactions
    (2023-12) Gallegos, Mariah; Conrad, Jacinta C.; Palmer, Jeremy C.; Karim, Alamgir; Shaffer, Devin L.; Louie, Stacey M.
    Suspensions containing micron-sized colloidal particles and polymers are widely used in many commercial and industrial applications such as paints, consumer products, pharmaceuticals, and separation processes. The addition of polymer to the suspensions can cause varying interactions between the colloidal particles and the polymers. The types of interactions that arise from the addition of polymer can affect the phase and rheological of these suspensions. The suspension attractive strength is affected by polymer size, polymer dispersity, polymer charge, polymer concentration, type of colloid-polymer interaction, solution pH, and salt concentration - also affecting the phase and rheological behavior. By understanding how these parameters affect the interactions between the colloidal particles and polymers can provide insight on flocculation processes such as wastewater treatment and creation of dense markers for indirect detection of specific cell surface molecules. In this thesis, we developed a bridging colloid – polymer model system in which the bridging strength can be controlled via the solution pH. The bridging strength as a function of polymer concentration, particle volume fraction, and solution pH showed to affect the formation of clusters and colloidal networks (Chapter 2). Next, we explored the effects of polymer molecular weight on our pH – tunable bridging model system at low particle volume fraction at varying normalized polymer concentration in the free volume without changing the solution pH. We found that the polymer size affected the size of cluster formation and steric stabilization limit in our colloid – polymer model system. At low polymer molecular weight, small clusters formed at low normalized polymer concentration which decreased in size as normalized polymer concentration increased until particles appeared nearly-hard-sphere due to steric stabilization. At high polymer molecular weight, large dense clusters formed at low normalized polymer concentration in which the size of the clusters decreased as normalized polymer concentration increased. Although a decrease of cluster size was observed, polymer concentrations were not large enough for all the particles to be sterically stabilized (Chapter 3). Next, we examined how polymer size affects the rheological behavior on solid-like bridging suspensions using the same colloid – polymer model system. We investigated the oscillatory shear rheology of high particle volume fraction suspensions at constant normalized polymer concentration across polymer molecular weights 130 and 450 kDa. We ran shear stress and flow sweeps on these suspensions to obtain the linear viscoelastic range and range of yield stress (Chapter 4). Finally, we provide a summary of the work presented in this thesis as well as a discussion of open concepts and inquiries to be still investigated in Chapter 5.
  • Item
    Extensional Rheology of Colloidal Particles and Polymer Mixtures
    (2023-12) Soetrisno, Diego D; Conrad, Jacinta C.; Krishnamoorti, Ramanan; Karim, Alamgir; Ferdowsi, Behrooz; Sharma, Vivek
    Colloid-polymer mixtures are ubiquitous in many different applications ranging from cosmetic products to energy materials. During manufacturing or application, these mixtures often undergo extensional deformation. Understanding how colloid-polymer mixtures flow under elongational stress is important to control and optimize the processability of these materials. We study the extensional flow properties by characterizing the capillarity-driven pinching dynamics of colloid-polymer mixtures using a dripping-onto-substrate (DoS) protocol. Methacrylate copolymer particles with acrylamide copolymer brushes are suspended in a refractive-index- and density-matched mixture of glycerol-water with NaCl added to screen the electrostatic repulsions. A non-adsorbing polymer, polyacrylamide, is added at varying molecular weight and dispersity. Addition of polyacrylamide induces depletion attractions between the colloids. The extensional properties is characterized by calculating the extensional relaxation time and the filament lifespan from the pinching dynamics. In polyacrylamide solutions without particles, the concentration dependence of extensional relaxation time is controlled by coil-stretch hysteresis. The scaling exponent of the extensional relaxation time with concentration increases with polymer size, which is attributed to the screening of excluded volume interactions by the presence of coil-stretch hysteresis under extensional flow. In colloid-polymer mixtures, the presence of particles does not affect the extensional relaxation times. The concentration-dependent scaling of extensional relaxation times collapses when they are scaled with free volume polymer concentration, suggesting that the polymer elastic properties control the extensional time scale. The filament lifespans of polymer solutions and colloid-polymer mixtures, when normalized by the filament lifespan of the corresponding fluid without polymer, follow a master curve as a function of free volume concentration. These results provide insight into the role of polymer in the extensional rheology of colloid-polymer mixtures.
  • Item
    Machine Learning-based Event Data Mining in Healthcare and Manufacturing
    (2023-12) Li, Mai; Lin, Ying; Feng, Qianmei; Lim, Gino J.; Fu, Wenjiang; Han, Zhu
    Event data, encompassing real-world occurrences with specific topics, locations, and timeframes, plays a pivotal role in diverse fields such as healthcare, manufacturing, and business, etc. Efficient and accurate event data mining is vital for deriving valuable insights and improving decision-making processes. However, event data mining faces unique challenges that require the development and integration of advanced machine learning techniques. These challenges include the risk of privacy leakage in multi-source event data mining, the difficulties in capturing various characteristics of events, and the complex dependencies among multichannel multitype events generated from multivariate time series. This dissertation addresses these challenges and contributes to event data analytics in the healthcare and manufacturing domains. First, we propose novel multi-source event data mining approaches that protect sensitive information while retaining analytical utility. Second, we develop advanced machine learning techniques for heterogeneous multi-output prediction of events, enabling accurate prediction of various characteristics simultaneously. Third, we introduce innovative methodologies for discovering complex dependencies among multichannel multitype events, providing insights into their intricate relationships and underlying mechanisms in multivariate time series. By integrating and developing state-of-the-art machine learning-based techniques, this research advances the field of event data mining, offering significant benefits to the healthcare and manufacturing sectors by enhancing decision-making processes and overall performance.
  • Item
    Machine Learning and Unstructured Data Analytics for Digital Marketing
    (2023-12) Xie, Wen; Han, Zhu; Pan, Miao; Fan, Lei; Overgoor, Gijs; Chen, Ming; Nguyen, Hien Van
    The internet has been a fertile ground for content creation, with people generating vast amounts of data daily. The proliferation of unstructured data can hold significant potential for practitioners and managers, as it may reflect the public's interests and behaviors. Analyzing this data, however, presents challenges for traditional marketing and business researchers. The emergence and evolvement of machine learning (ML), data science (DS), and sophisticated computational tools has given rise to novel approaches that can address these challenges. This dissertation investigates three prominent ML and DS applications within marketing research, demonstrating their capacity to effectively process and interpret complex unstructured data—including images and videos—for business insights and consumer welfare. The first study explores the integration of object detection algorithms with eye-tracking technology to dissect the consumer shopping experience online and on mobile platforms. This combination of techniques offers fresh perspectives on consumers' visual attention, yielding practical insights for digital advertising strategies. The second study employs image segmentation alongside Bayesian analysis to explore skin-tone representation in brand visuals on social media. The proposed approach refines the strategic management of marketing communications. In the final study, I apply rigorous data science methods to scrutinize user engagement with advertising on a short-video social platform. Informed by processing fluency theory, this analysis reveals nuanced patterns of consumer behavior under various contexts, providing essential insights to enhance algorithms for ad ranking and recommendation systems. Collectively, these studies demonstrate the significant capacity of ML and engineering techniques to transform marketing research by offering deeper, data-driven insights into unstructured data within the digital landscape.
  • Item
    Deep Learning Enhanced Multi-physics Joint Inversion
    (2023-12) Hu, Yanyan; Chen, Jiefu; Wu, Xuqing; Han, Zhu; Jackson, David R.; Hu, Wenyi
    Joint inversion has drawn considerable attention due to the availability of multiple geophysical datasets, ever-increasing computational resources, development of advanced algorithms, and its ability to reduce inversion uncertainty. A key issue of joint inversion is to develop effective strategies to link different geophysical data in a unified mathematical framework, where the information obtained from different models can complement each other. Traditionally, structural similarity constraints are pursued by joint inversion algorithms using manually crafted formulations (e.g. cross gradient). In this dissertation, we introduce a novel approach: a Deep Learning Enhanced (DLE) joint inversion framework. Within this framework, structural similarity is enforced using a well-trained deep neural network (DNN), enhancing the quality of joint inversion results through iterative improvements. The DNN is seamlessly integrated into existing independent inversion workflows, with the flexibility to extend its application to various multi-physics scenarios without requiring structural modifications. Within the DLE joint inversion framework, several key contributions are made. First, we design a double-channel U-Net for the simultaneous inversion of 2D DC resistivity data and seismic travel time. Extensive numerical experiments validate the efficacy of this method. Importantly, this learning-based approach exhibits impressive generalization capabilities when applied to datasets featuring diverse geological structures, sensing configurations, and nonconforming discretization. Second, we harness the power of deep perceptual loss as a regularization technique to further enhance structural similarity. Successive networks are trained with deep perceptual constraints, derived from a pre-trained network specializing in edge detection. The robustness of this approach is verified through experiments involving layered subsurface models, demonstrating its ability to jointly invert three types of geophysical data, including induced polarization data. Third, we simplify the DLE framework and apply it to tackle the challenging 3D joint inversion of magnetic and gravity gradient data. The proposed method is rigorously evaluated through synthetic and field cases, affirming its effectiveness and computational efficiency. In summary, this dissertation contributes to the advancement of multi-physics joint inversion by introducing the Deep Learning Enhanced framework. This innovative approach enhances both the accuracy and efficiency of geophysical inversion, promising broader applications and improved outcomes in the field.
  • Item
    Smart Ultrafine Cement Grouts Development, Characterization, and Modeling Multiple Behaviors and Testing the Capturing of Carbon Dioxide (CO2) Using Recycled Additives.
    (2023-12) Elsayed, Khalad Elsayed Mohamed; Vipulanandan, Cumaraswamy; Mo, Yi-Lung; Li, Hong-Yi; Lim, Gino J.; Chen, Yuhua
    Smart cement is a highly piezo, thermo, and chemoresistive sensor that has many applications in infrastructures such as bridges and oil wells. However, the limitation of using smart cement in grouting applications such as maintenance, repairing, and sealing joints of various infrastructures can lead to some difficulties in monitoring the structures during the maintenance. Therefore, in this study, a new smart ultrafine (UF) cement grout sensor was developed to be used for real time monitoring with high sensing performance which can be implemented in grouting applications as a repairing and maintenance material to minimize failures. Smart ultrafine cement grout was developed with higher water-to-cement and binder ratios by modifying it with different green materials. Also, the rheology, curing, shrinkage, setting times, and piezoresistive behaviors of the smart ultrafine cement grouts were tested and also modeled. In order to identify easily monitorable and dependable sensing properties, a series of experiments were conducted using electrical impedance, and resistivity was identified as the critical property to monitor. The piezoresistive behavior of the ultrafine cement grouts was substantially improved by adding 0.05% of carbon fiber. Also, the curing of the smart ultrafine cement grout was monitored with resistivity changes and also modeled. The piezoresistive strain at peak stress varied from 217% to 277% based on the curing times and water to cement ratios. Also, the stress-piezoresistive strain behavior was predicted using Vipulanandan p-q model and the artificial neural network (ANN) model. By incorporating the green materials in the smart ultrafine cement grouts significantly increased its piezoresistivity compared to unmodified smart UF cement grout. Also, the smart ultrafine cement grouts rheology was modeled using the ANN, Herschel Bulkley and Vipulanandan models. The findings indicated that the inclusion of green materials led to a decrease in shear stress, whereas an increase in temperature resulted in an elevation of shear stress. In this study also, methods for capturing CO2 using various recycled materials in water were investigated. The results showed that certain materials can capture the CO2 in the water very fast and several parameters were monitored including the resistivity which turned out to be very sensitive.
  • Item
    Engineering and Characterizing Escherichia coli for Enhanced Alkane Conversion
    (2023-12) Nguyen, Nam; Cirino, Patrick C.; Conrad, Jacinta C.; Lin, Yuheng; Fujita, Masaya; Varadarajan, Navin
    Rising global energy demand and geopolitical tensions have renewed interest in expanding domestic production of natural gas liquids (NGLs) like propane and butane. These alkanes possess high energy density but present transport challenges over long distances. Industrial bioconversion of NGLs into liquid fuels using traditional processes like Fischer-Tropsch (FT) is capital intensive. As an alternative, certain microbes can enzymatically activate alkanes with methyl-alkylsuccinate synthase (MAS), enabling more economical bioprocessing. However, native MAS-containing bacteria have limited genetic tractability. A majority of the research described in this dissertation focused on establishing functional MAS expression systems in the well-studied host Escherichia coli. Mas genes from anaerobic bacteria were heterologously expressed in E. coli under strict anaerobic conditions to maintain activity. Systematic optimization of media composition, cofactors, and culturing conditions improved alkylsuccinate production from 50μM to 200μM. Bioconversion of propane, butane, and hexane by the recombinant MAS was confirmed via GC-MS. Characterization of the MAS complex revealed a non-essential subunit, enabling comparison to the related benzylsuccinate synthase. Additionally, E. coli strains were engineered to enhance interactions with hydrophobic substrate. Surface proteins including fimbriae, flagella, and curli were engineered for inducible and tunable expression. Optimizing expression levels prevented toxicity while improving interactions with alkane emulsions and biofilm formation. In summary, this dissertation establishes the foundation for engineering E. coli capable of activating SCAs with enhanced interactions to hydrophobic substrates.
  • Item
    ENHANCING MALWARE DETECTION THROUGH BEHAVIORAL MODELLING AND FEATURE LEARNING
    (2023-12) El Aassal, Ayman; Huang, Stephen; Leiss, Ernst L.; Shi, Weidong; Conklin, Wm. Arthur
    The growing frequency of data breaches and cyberattacks due to malware infections in recent years highlights the significance of ongoing research in malware detection. Malicious software, or malware for short, often undergoes numerous mutations to avoid detection by signature-based antivirus software. The abundance of malware variants has made the task of detection increasingly complex. Mainstream cybersecurity vendors favor static analysis methods due to their speed and scalability in assessing incoming files, generating their signatures, and cross-referencing them with a database of recognized malicious signatures for detection. However, this form of analysis is susceptible to obfuscation methods where hackers modify malware code in superfluous ways to generate a new signature that is not yet recognized by antiviruses. That is why this work focuses on analyzing the run-time execution of programs to extract their behavior and identify them as malware or benign. This dissertation addresses the persistent challenge posed by the ever-evolving malware variants by introducing a framework designed to capture the run-time behavior of programs through graph modelling and deep learning methods. The proposed approach parses the log of native functions called by a program during its execution. This parsing process enables the creation of Behavior Call Graphs (BCGs) using a novel methodology emphasizing the connections between these native functions. Graph structures offer the ability to effectively represent intricate relationships within the data, facilitating the extraction of relevant information that might be challenging to capture otherwise. This research employs two different methods to analyze these BCGs. The first involves extracting domain expert features, while the second leverages deep learning algorithms to generate the features automatically. However, it's worth noting that conventional deep learning methods like Neural Networks and Convolutional Neural Networks are not designed to handle graphs as input. To address this limitation, we adopted feature learning algorithms that automatically embed graph structures into feature vectors within a multi-dimensional space. This dissertation validates the effectiveness of these approaches in analyzing BCG generated from Windows and Android applications to identify and capture the malicious behavior of malware variants. This research is helpful for companies and software publishers to test the safety of uploaded or shared applications and prevent malware from spreading to their end users.
  • Item
    Mapping Hydrocarbons and Rare Earth Elements at Various Scales through Imaging Spectroscopy
    (2023-12) Gadea, Otto C; Khan, Shuhab D.; Sisson, Virginia B.; Castagna, John P.; Krupnik, Diana
    Remote sensing techniques can play a critical role in geologic interpretation for mineral and energy exploration in unusual deposits. Hyperspectral cameras measure how a geophysical variable changes across wavelengths in the visible to short-wave infrared portion of the spectrum to map surface compositional variation. For two sites in this investigation, close-range ground-based imaging spectroscopy of hand samples with centimeter to sub-millimeter spatial resolution is used to establish preliminary information about the spatial distribution of natural resources within a geological formation. This information can then be compared with mineralogical maps of lateral rock exposures derived from airborne and spaceborne data. The first site is Fort McMurray in Alberta, Canada. The fluorescence characteristics of a bituminous sandstone depend on varying concentrations of light and heavy hydrocarbons, enabling imaging spectroscopy to distinguish zones of optimal yield for crude oil extraction. In the first chapter, multiple images of tar sand samples are collected under different wavelengths of ultraviolet illumination and normalized with respect to the fluorescence patterns of Spectralon diffuse reflectance material. Three classification methods are used to distinguish between bitumen, Spectralon, and a non-fluorescent slate background. Spectral indices useful for indicating concentrated bitumen in tar sands are proposed. The second site is the Sulfide Queen mine in Mountain Pass, California, which contains economic deposits of bastnäsite within a carbonatite body intruding a metamorphic host rock that have been mined for rare earth elements (REEs) critical to components in high technology devices. In the second chapter, a new spectral index based on reflectance measurements from hyperspectral data collected under visible illumination is created to map the relative REE abundances across three museum samples known to consist primarily of bastnäsite from the mine. In the third chapter, that index and its two new modified versions are applied to eleven ore samples of varying composition from known geolocations across the mine. To determine how changes in spectral patterns across different scales affect the perceived distribution of rare earth elements, these three indices are also applied to data collected by eight airborne and spaceborne imaging spectrometers over the Sulfide Queen mine and the nearby Ivanpah Dry Lake.
  • Item
    Developing Deep Learning Models for Depression Detection in Texts
    (2023-12) Aigbe, Steve Aibuedefe; Eick, Christoph F.; Chen, Guoning; Tsekos, Nikolaos V.; Yuan, Xiaojing
    Depression is a major mental health disorder affecting a significant portion of the world population. Methods mostly being employed for depression detection are clinical interviews and questionnaire surveys where psychiatric assessment tables are used to establish mental disorder prognosis. Analyzing texts written by an individual can serve as an additional knowledge source to diagnose depression. Consequently, using deep learning models to detect depressed and non-depressed individuals based on social media posts, by analyzing the words being posted, has become the focus of recent research. The lack of big-sized depression-labeled datasets for training models for depression detection in texts is a major challenge. Also, selecting a data augmentation (DA) method to augment the available small-sized datasets is difficult. So, we developed a methodology, named DAMEVAL, for the evaluation of DA methods for text classification. In DAMEVAL, we proposed a set of evaluation measures and benchmark NLP datasets for the evaluation and comparison of DA methods to create a reference for easier selection of DA methods by users. In this dissertation, we extracted and analyzed the textual depression symptoms indicators present in texts posted in online forums and the distribution of these indicators with respect to depressed and non-depressed social media users. Also, we computed weights, using the TFIDF method, based on the extracted depression symptoms’ indicators present in users' posts. Subsequently, we introduced a weighted deep learning model named DEP-BERTCNN, based on the computed depression indicators’ weights, for depression detection in text in online forums. DEP-BERTCNN uses a combination of a pre-trained BERT language model, an attention model and convolutional neural network to classify forum users as depressed or non-depressed. The DEP-BERTCNN model was trained and evaluated on the large-scale Reddit Self-reported Depression Dataset (RSDD). Our results outperform several baseline methods for depression detection in texts, demonstrating the effectiveness of combining deep learning model with linguistic indicators associated with depression symptoms. In summary, we aim to develop a beneficial system that can easily be used to detect depression in texts and enable policy makers to respond to mental health escalations easily and promptly.
  • Item
    Retention of Enhanced and Induced Superconductivity at Ambient Pressure Through High-Pressure Quenching
    (2023-12) Bontke, Trevor; Chu, Paul C. W.; Hosur, Pavan; Meen, James K.; Ren, Zhifeng; Ting, Chin-Sen
    In the past eight years impressively high superconducting critical temperatures (Tcs) have been reported in numerous materials. Among these include instances of anomalously high Tcs that approach, and in some contested reports, meet and exceed room temperature (RT), pushing the field to new heights. Unfortunately, achieving such impressive critical temperatures requires ultra-high external pressures, rendering them unviable for commercial use. Therefore, one of the most significant challenges remaining in the field of superconductivity is to retain the high Tc phases induced by pressure while lowering or removing it completely. We have therefore employed a pressure quenching technique to retain high-pressure-induced/-enhanced superconducting phases in Bi, FeSe, and CuxFe1.01-xSe at ambient pressure. Pressure quenching bismuth at 77 K and 4.2 K from pressures up to 26.6 GPa successfully produced metastable superconducting phases with varying Tcs from ∼ 5 K up to a new record of 9 K. By changing the pressure quenching parameters, different metastable phases could be targeted, namely Bi-III with a Tc around 7 K and Bi-V with a Tc > 8 K. Temporal stability testing and thermal cycling revealed a lower temperature limit below ∼ 60 K and an upper temperature limit of 120 K – 150 K in metastable bismuth. Pressure quenches performed on FeSe and CuxFe1.01-xSe near the superconducting dome resulted in metastable phases with maximum Tcs of 37 K and 25 K, respectively. Thermal cycling of FeSe and CuxFe1.01-xSe showed a similar lower temperature limit for temperatures up to ∼ 120 K and an upper temperature limit around 175 K for CuxFe1.01-xSe. Annealing metastable FeSe to room temperature produced Tcs from 15 K – 24 K. Notably, a non-superconducting hexagonal phase retained in FeSe was slowly annealed to room temperature for a few days resulting in a superconducting phase near the dome peak. Lastly, a temporal stability test of metastable CuxFe1.01-xSe was conducted which showed perfect phase stability for 7 days when kept below 50 K. Overall, these results demonstrate the potential this technique has in targeting desirable superconducting phases induced or enhanced by pressure and retaining them in a metastable state at ambient pressure.
  • Item
    Geophysical Reservoir Characterization and Monitoring of CO2 Sequestration in the North Sea and Assessment of Hydrogen Storage Sites in the Gulf of Mexico region
    (2023-12) Pelemo-Daniels, Oyedoyin O; Stewart, Robert R.; Castagna, John P.; Ebrom, Daniel A.; Sager, William W.
    This work aims to enhance reservoir characterization accuracy for improved gas storage, sequestration, and monitoring analysis. The first objective is to conduct a comprehensive reservoir characterization of the Volve Field in the North Sea utilizing petrophysical analysis and rock physics modeling. This involves integrating core data, well-log, and synthetic modeling to determine the key petrophysical properties of the subsurface reservoirs. By analyzing the porosity, permeability, lithology, and fluid saturation, the reservoir parameters vital for successful exploration and storage are determined along with associated potential risks. The research then predicts these petrophysical properties from seismic inversion attributes using machine learning techniques. Next, we investigate the impact of pressure and temperature on the acoustic properties of various gases. These properties are essential for understanding subsurface gas effects. Acoustic Resonance Spectroscopy is employed to measure gas velocities and assess the effects of pressure and temperature. Our findings reveal that gas velocities generally increase with increasing pressure and temperature. To understand the impact of pressure on gas behavior in sandstone reservoirs, ultrasonic measurements on gas-saturated sandstones were carried out. The results show that P-wave and S-wave velocities generally decrease with increasing pore pressure. Furthermore, we performed an in-depth study of CO2 injection monitoring in the Sleipner Field, North Sea, focusing on the Utsira Formation. The research leverages advanced time-lapse inversion techniques and 4D seismic data analysis to enhance the accuracy of volume estimations and provide a comprehensive understanding of the dynamic behavior of the injected CO2 plume. The analysis encompasses cross-correlation, time shift, predictability, and other key elements to yield robust insights into the reservoir's response to CO2 injection. Interpreted gas volumes from the seismic changes closely align with the injected volume, with a calculated-to-actual ratio ranging from 0.9 to 1.1. Lastly, we focused on hydrogen storage development in salt caverns exploring the suitability of salt formations for hydrogen storage and operations. Of the 144 evaluated domes, 26 were selected in the initial screening due to their favorable structural integrity and storage potential. In summary, this thesis provides geophysical techniques, measurements, and case histories to augment our understanding of gas reservoirs and their changes.
  • Item
    Computationally Efficient DNN Mapping Search Heuristic Using Deep Reinforcement Learning
    (2023-12) Bakshi, Suyash; Johnsson, Lennart; Vilalta, Ricardo; Subhlok, Jaspal; Mattson, Timothy G.
    In this dissertation, we present a computationally efficient Reinforcement Learning (RL) search heuristic for finding high quality mappings of N perfectly nested loops, such as loops in Convolutional Neural Networks (CNNs) for high dimensional data sets, to accelerators with multiple processing elements (PEs) each with a memory hierarchy and a shared-memory for all PEs. Our RL search uses maximum potential operand reuse to guide the search process. It is computationally inexpensive compared to RL reward functions used by state-of-the-art mapping search methods. The maximum potential operand reuse for mappings is also used for an effective mapping pruning strategy that significantly contributes to the overall computational effectiveness of our RL search method. We also present a search space state representation and a parsing strategy therefore that produces only valid mappings. Unlike supervised learning methods, our RL search does not require training datasets, thus is easily applicable to different loop-nests and accelerators. We show that our RL search heuristic evaluated for 19 3D convolution layers, ten initial states, three 256 PE accelerator configurations, and two different operand datatypes, required only on average 10% of Timeloop’s random search floating-point operations, yet found mappings with on average 13% lower Energy-Delay-Product (EDP) for the same number of valid mappings. Further, the lowest EDP mappings found using our method had on average a 6.5x higher EDP than simple lower bound EDP estimates, with the best case being only 1.6x higher.