12023 research outputs found
Sort by
Physics-Based and Data-Driven Computational Models of Inelastic Deformations
Get PDFCrystalline materials inevitably exhibit inelastic deformation when applied to large enough loads. The behavior in this inelastic regime is a coupling of physics across several length scales: from initiating as defects at the atomic scale, interacting with crystal defects, and finally spanning multiple grains and influencing macroscopic stress behavior. These length-scale interactions make predicting material response an open challenge and an avenue for leveraging microscale physics for material design. This thesis examines developing physics-based and data-driven computational models to capture complex inelastic behavior at appropriate length scales.
First, we present a mesoscale model for capturing deformation twinning physics at the polycrystal scale. Mechanical twinning is a form of inelastic deformation observed in low-symmetry crystals, such as magnesium and other hexagonal close-packed (hcp) metals. Twinning, unlike slip, forms as bands collectively across grains with complex local morphology propagating into bulk behavior, drastically affecting strength and ductility. We, thus, propose a model where twinning is treated using a phase-field approach, while dislocation slip is considered using crystal plasticity. Lattice reorientation, length-scale effects, interactions between dislocations and twin boundaries, and twin and slip interactions with grain boundaries are all carefully considered. We first outline the model and its implementation using a novel approach of accelerated computational micromechanics in a two-dimensional, single twin-slip system, polycrystal case to demonstrate its capabilities. Finally, we consider multiple twin-slip systems and conduct three-dimensional simulations of polycrystalline magnesium. We summarize the insights gained from these studies and the implications on the macroscale behavior of hcp materials.
The second part of the thesis focuses on data-driven models for capturing microscopic history-dependent phenomena for multiscale modeling applications. The multiscale modeling framework has seen increased usage over the last few decades for its ability to capture complex material behavior over a range of time/length scales by solving a macroscale problem directly with a constitutive relation defined implicitly by the solution of a microscale problem. However, this implementation is computationally expensive -- needing to solve a microscale problem at each point and time of the macroscopic calculation. In this study, we examine the use of machine learning by utilizing data generated through repeated solutions of a microscale problem to: (i) gain insights into the history dependent macroscopic internal variables that govern the response and (ii) create a computationally efficient surrogate. We do so by introducing a recurrent neural operator, which can provide accurate approximations of the stress response and insights into the physics of the macroscopic problem. We illustrate these capabilities on a laminate composite and polycrystal made of elasto-viscoplastic materials, summarize insights on the learned internal variables, and accuracy of stress predictions.</p
Computation Foundations of Spatial Transcriptomics
Get PDFSingle-cell and spatial transcriptomics have come of age in the past few years; datasets and data analysis software packages have proliferated. With the increasing sizes of datasets, proliferating new data collection technologies, and mainstreaming of high-throughput technologies, the software can be improved for better speed and memory efficiency, standardized and consistent user interface for multiple technologies, and in documentation to onboard new users. First, I collected a database of spatial transcriptomics literature and analyzed the data on trends and sociology in this field. Based on the database and data analyses, I wrote a comprehensive book both qualitatively and quantitatively documenting the history of the field since the 1960s and reviewing more recent developments, which informed the software and methods I later developed. Then, to address the challenges with the pre-processing large datasets, we developed \texttt{kallisto} \texttt{bustools} for fast and modular pseudoalignment of sequencing reads to the transcriptome in single-cell RNA-seq (scRNA-seq), giving consistent results with the established and much more computationally demanding alignment method Cell Ranger. Briefly summarized are my attempt to map dissociated cells in scRNA-seq to a spatial gene expression reference and to build a image processing pipeline for image based spatial transcriptomics data analysis. Finally, to address the challenges in downstream analyses of spatial -omics data, I first wrote the new \texttt{SpatialFeatureExperiment} (SFE) data structure to represent and operate on geometries in spatial transcriptomics data and to organize results from spatial analyses. Based on SFE, I wrote Voyager, which brings decades of research in geospatial data analysis to spatial transcriptomics, to better utilize the opportunities from spatial information to gain novel biological insights. To reduce user learning curve, Voyager conforms to SCE styles and conventions and has a comprehensive documentation website and consistent user interface to many geospatial methods.</p
Machine Learning and Data Assimilation for Blending Incomplete Models and Noisy Data
Get PDFThe prediction and inference of dynamical systems is of widespread interest across scientific and engineering disciplines. Data assimilation (DA) offers a well-established and successful paradigm for blending such models with noisy observational data. However, traditional DA-based inference often fails when available data are insufficiently informative. Chapter 2 copes with this challenge by introducing constraints into Ensemble Kalman Filtering, which is shown to improve forecasting of glucose dynamics in real patient-level clinical data. Chapter 3 addresses this identifiability challenge by instead developing a simplified, reduced-order stochastic model for glucose dynamics that is more easily identified from patient data. Despite these successes, the forecasting performance of the methods are fundamentally limited by the fidelity of the employed model, which is often not fully understood a priori.
Chapter 4 presents a general picture of how noisy, partially-observed time-series data can be used to learn flexible (e.g., neural network-based) corrections to a pre-specified mechanistic model. In Chapter 5, the proposed methodology is then validated in simulated settings for glucose-insulin models. Chapter 6 provides further perspective on learning flexible model corrections, comparing approaches that use i) gradient-based or gradient-free optimization, ii) temporal or time-averaged data, iii) different model parameterizations, iv) deterministic and stochastic corrections, and v) physical conservation laws to constrain inference.
Chapter 7 studies how these perspectives on machine learning and dynamical systems can help us understand the roles of biochemical networks. In particular, it considers protein dimerization networks from the lens of approximation theory and evaluates how the equilibria of these networks can be fine-tuned to perform a variety of biological computations.</p
Complex Charge Compensation Mechanisms in Lithium-Rich Chalcogenide Cathodes
Get PDFLithium-ion batteries have revolutionized the world by enabling long-lasting portable electronics, electrified transportation, and grid storage solutions for renewable energy implementation. However, current commercialized technologies are limited by the one electron transfer per transition metal paradigm utilized by cathode materials that operate with an intercalation-based charge storage mechanism. Finding ways to increase the charge storage capabilities of the cathode into the multielectron regime has long been a focus of research efforts, and involvement of structural anions in the redox has been demonstrated as a promising way to accomplish multielectron storage. Layered lithium-rich oxide materials have been shown to afford dramatic improvements to overall storage capacity but are plagued by complex mechanisms and unwanted side reactions that lead to poor cycling stability and characterization difficulties. This thesis expands upon previous understanding of oxide-based anion redox materials and extends the exploration into sulfide and selenide systems, which allow the study of anion redox without the side processes that affect oxides. First, a dynamic charge compensation mechanism of late group metal-poor, lithium-rich oxide, Li2Ru0.3Mn0.7O3, is uncovered and found to involve an irreversible anion oxidation that leads to involvement of redox states on transition metals previously thought to be unavailable. Second, active electrolyte additives are explored as a method of stabilizing the cathode-electrolyte interface of anion redox material, Li2RuO3. Third, reversible anion redox is demonstrated in alkali-rich sulfides, Li2FeS2 and LiNaFeS2, and proven to occur through oxidation of sulfides (S2-) to persulfides ([S2]2-). Understanding of the structural ramifications of anion oxidation in Li2FeS2 is further expanded through computational and experimental methods. Fourth, the role of metal-anion covalency is systematically investigated through anion substitution of Li2FeS2 with S2-, highlighting the importance of a holistic understanding of changes to the electronic and physical structure of anion redox materials to predict long-term performance. Finally, detailed perspectives and future outlooks on sulfur redox in lithium battery systems are offered with an exhaustive survey of thermodynamically stable binary and ternary persulfide materials.</p
Development of Selective Carbon–Carbon Bond-Forming Reactions of Vinyl Carbocations
Get PDFCarbocationic intermediates play an important role in the construction of complex molecules, from biosynthetic pathways in nature to the synthesis of natural products by organic chemists. Herein, a brief introduction that surveys C–C bond forming reactions of vinyl carbocations is highlighted. The discussion of experimental work commences with the development of a main group-catalyzed approach towards accessing vinylated esters through the trapping of vinyl carbocations with silyl ketene acetals. Next, a Claisen-type rearrangement is discussed, which is a result of trapping vinyl carbocations with allyl ethers to form an allyl vinyl oxonium intermediate in situ that can subsequently rearrange. Finally, the last method that is highlighted includes the development of an asymmetric C–H insertion reaction of vinyl carbocations to forge bicyclic products in a highly enantioselective fashion
Who’s Chinatown, Whose Chinatown? Visions of Urban Progress in Los Angeles Chinatown, 1970-2020
Get PDFThis thesis explores the dynamics of urban development in Los Angeles (LA) Chinatown since the 1970s until present day. The historical narrative is driven by broad demographic shifts across LA County, alongside municipal and community politics that shape the material and cultural demands behind neighborhood change. Through this narrative, I challenge the traditional framings of resident versus business interests in Chinatowns, and instead highlight the complicated and often competing visions of progress throughout the community. I argue that “the youths” and “the elders” serve as key figures in this history: first, as dynamic actors and activists directly engaged in the process of development, and second, as subjects of discourse that actors mobilize towards different goals of development. Finally, I illuminate tensions between organizing as a representative of a community and organizing in solidarity towards the tangible needs of a community
Applied Machine Learning for Prediction and Control of Fluid Flows
Get PDFModern aerodynamic technologies such as unmanned aerial systems and horizontal axis wind turbines must regularly contend with forces from highly stochastic and turbulent atmospheric gusts. Conventional methods for modeling and controlling fluid flows are limited in their ability to mitigate these aerodynamic forces in real-time. By applying modern machine learning techniques in an experimental setting, this thesis demonstrates the utility of machine learning in addressing these important problems. We follow two complementary approaches towards this goal.
First, we find an end-to-end solution for control in a gusty environment with model-free reinforcement learning. We deploy state-of-the-art reinforcement learning algorithms on a generalized aerodynamic test-bed consisting of an airfoil with motorized trailing edge flaps. The system features embedded flow sensors, enabling the inclusion of flow measurements in state observations. We place this system in a highly irregular wake behind a bluff-body, dynamically mounted on elastic bands and therefore free to oscillate, and train reinforcement learning agents to minimize the net lifting force on the system by controlling the position of the trailing edge flaps. We find that model-free reinforcement learning agents can outperform basic linear controllers in this gusty, turbulent environment. We also show that augmenting state observations with flow measurements can lead to more consistent learning of the system dynamics.
Next, we explore Fourier neural operators (FNOs) as a method for forecasting the time evolution of turbulent fluid flows. FNOs are capable of learning underlying operator solutions to families of partial differential equations and can be evaluated in just milliseconds. We specifically focus on training FNOs with experimentally measured velocity fields of bluff body wakes in the subcritical regime. To the best of our knowledge, this is the first application of operator learning for fluid mechanics that features experimental measurements. We find that FNOs can accurately predict the evolution of these turbulent wakes even when trained with imperfect measurements. We then show that FNOs can quickly adapt to unseen conditions with minimal data and training through transfer learning. Finally, we consider the performance of FNOs over longer prediction horizons. This approach could enable real-time gust prediction capabilities and monitoring for applied aerodynamic systems.</p
Application of Heterojunction Ni-Sb-SnO₂ Anodes for Electrochemical Water Treatment
Get PDFClean water supply and adequate sanitation services are critical for public health as well as for food production. Small-scale decentralized treatment represents an attractive alternative that can provide necessary water treatment in many parts of the developing world where centralized wastewater treatment facilities are not practical owing to financial, geographical, or political constraints. Electrochemical oxidation (EO) is a suitable technique for decentralized treatment settings since it does not require the addition of auxiliary chemicals and offers fast reaction kinetics and modular treatment capacity. EO is considered a versatile technology since it can degrade a wide array of contaminants and inactivate waterborne pathogens. The chemical composition of the anode, where EO takes place, is a key factor that controls reactive species production and thus treatment efficiency and energy consumption. Ideal anodes for wastewater treatment should have high overpotential for oxygen evolution (“nonactive” anodes) and favor complete organics oxidation through direct electron transfer and/or reactions with potent oxidants such as hydroxyl radical and ozone. Common nonactive anodes including antimony-doped tin oxide (Sb-SnO₂), lead oxide (PbO₂), and boron-doped diamond (BDD) have attracted wide research interests. The work presented in this thesis centered around a newly designed heterojunction Ni-Sb-SnO₂2-based anode (NAT/AT) and its various applications in decentralized water and wastewater treatment. Direct treatment using NAT/AT has proved to be efficient for chemical oxygen demand removal, trace organic compound degradation, and microbial disinfection. Detailed investigation into pharmaceutical degradation kinetics and transformation products further established NAT/AT as a potential treatment alternative for the control of pharmaceuticals and their metabolites in hospital wastewaters. NAT/AT is also capable of synthesizing ferrates (e.g., FeO₄²⁻) in circumneutral conditions, the high oxidation state iron species that represents another group of powerful oxidants well-suited for decentralized treatment purposes. In an additional effort to tackle high concentrations of ammonium often present in latrine wastewaters, functionalized metal-organic framework (MOF), a class of materials featuring high porosity, abundant active sites, and highly tunable physical and chemical properties, was used to recover the ammonium nitrogen. Various modifications of MOF-808, a highly water stable MOF, were designed and synthesized to achieve urea hydrolysis, ammonium capture, and real-time ammonium sensing in sequence. In combination, the described works provide a powerful toolkit that can be used in treating various waste streams before discharge and/or reuse
Enhancing the Ethylene and Propylene Selectivities in the Methanol-to-Olefins Reaction by Exploiting the Intricate Relationship between Framework Topology and Acidity
No full textThis thesis describes and presents results from several related projects within the theme of molecular sieve synthesis and catalysis. The early part of the thesis focuses on understanding the link between cage size/dimension and acidity (i.e., acid site density and strength) in the methanol-to-olefins (MTO) reaction. This relationship between cage size and acidity, once identified and investigated, is exploited in the latter parts of the thesis to rationally design materials that are able to steer the light olefins product distribution toward either more ethylene or propylene in a significant improvement over SAPO-34 (CHA), the commercial catalyst.
In Chapters 2, 44 zeolites and silicoaluminophosphates (SAPOs) belonging to five frameworks (AEI, CHE, LEV, SWY, and ERI) with a wide range of Si/Al=4-31 and Si/(Al+P)=0.04-0.3, are synthesized and characterized using a myriad of techniques. Their MTO behavior is then systematically investigated to rationalize the effect of cage dimensions on the olefins product distribution as a function of acid site density and strength. The results from this study show that changes in acid site density and strength play a secondary role to the dominating influence of cage architecture on product distribution in AEI- and CHA-type molecular sieves. Decreasing the cage size, in going from AEI and CHA to LEV, SWY, and ERI, however, results in substantial changes in the ethylene-to-propylene ratio (E/P) as a function of acidity. These changes are attributed to differences in the identity and concentration of the hydrocarbon-pool (HP) species that form, particularly in early stages of the reaction.
In Chapters 3 and 4, ERI-type molecular sieves (e.g., SSZ-98, UZM-12, ERI-type zeolites, and SAPO-17) are thoroughly investigated as promising methanol-to-ethylene materials due to their narrow cage size. Specifically, numerous ERI-type molecular sieves are synthesized using several organic structure-directing agents (OSDAs) with varied Si/Al or Si/T-atoms ratios. The list of ERI-related materials synthesized and tested in MTO included a new disordered SAPO, denoted as CIT-16P, which upon thermal treatment in air transforms to SAPO-17 (ERI). The reaction results show that decreasing the Si/Al (or increasing the Si/T) ratio, irrespective of other material properties, improves the E/P of ERI-type molecular sieves (E/P=1.1-1.9) over CHA-type molecular sieves (E/P=0.82-0.85) in MTO. Dissolution-extraction experiments reveal that the rapid formation of cyclic intermediates and the shift in their composition toward less-methylated methylbenzenes and methylnaphthalenes are found to be key to enhancing the ethylene selectivity in ERI-type molecular sieves.
In Chapter 5, several SAT-type molecular sieves are investigated as promising methanol-to-propylene catalysts. This effort entails the synthesis of CIT-17, an SAT SAPO-type molecular sieve, which is isostructural to STA-2 (MgAPO-SAT). Following the successful synthesis of CIT-17, the MTO behavior of several SAT-type molecular sieves (MgAPO, CoAPO, and SAPO) are investigated in MTO. The combination of low acidity of CIT-17 and unique structural features of the narrow SAT-cage lead to a catalytic pathway and mechanism that predominantly favors propylene (propylene-to-ethylene ratios (P/E) of 2-4.2; propylene selectivity of 40-50%). Indeed, CIT-17 achieves one of the highest P/E ratio values reported for this class of materials.</p