Search

Accelerating the simulation of turbulence to stationarity is a critical challenge in various engineering applications. This study presents an innovative equation-free multiscale approach combined with a machine learning technique to address this challenge in the context of the one-dimensional stochastic Burgers' equation, a widely used toy model for turbulence. We employ an encoder-decoder recurrent neural network to perform super-resolution reconstruction of the velocity field from lower-dimensional energy spectrum data, enabling seamless transitions between fine and coarse levels of description. The proposed multiscale-machine learning framework significantly accelerates the computation of the statistically stationary turbulent Burgers' velocity field, achieving up to 442 times faster wall clock time compared to direct numerical simulation, while maintaining three-digit accuracy in the velocity field. Our findings demonstrate the potential of integrating equation-free multiscale methods with machine learning methods to efficiently simulate stochastic partial differential equations and highlight the possibility of using this approach to simulate stochastic systems in other engineering domains. Master of Science In many practical engineering problems, simulating turbulence can be computationally expensive and time-consuming. This research explores an innovative method to accelerate these simulations using a combination of equation-free multiscale techniques and deep learning. Multiscale methods allow researchers to simulate the behavior of a system at a coarser scale, even when the specific equations describing its evolution are only available for a finer scale. This can be particularly helpful when there is a notable difference in the time scales between the coarser and finer scales of a system. The ``equation-free approach multiscale method coarse projective integration" can then be used to speed up the simulations of the system's evolution. Turbulence is an ideal candidate for this approach since it can be argued that it evolves to a statistically steady state on two different time scales. Over the course of evolution, the shape of the energy spectrum (the coarse scale) changes slowly, while the velocity field (the fine scale) fluctuates rapidly. However, applying this multiscale framework to turbulence simulations has been challenging due to the lack of a method for reconstructing the velocity field from the lower-dimensional energy spectrum data. This is necessary for moving between the two levels of description in the multiscale simulation framework. In this study, we tackled this challenge by employing a deep neural network model called an encoder-decoder sequence-to-sequence architecture. The model was used to capture and learn the conversions between the structure of the velocity field and the energy spectrum for the one-dimensional stochastic Burgers' equation, a simplified model of turbulence. By combining multiscale techniques with deep learning, we were able to achieve a much faster and more efficient simulation of the turbulent Burgers' velocity field. The findings of this study demonstrated that this novel approach could recover the final steady-state turbulent Burgers' velocity field up to 442 times faster than the traditional direct numerical simulations, while maintaining a high level of accuracy. This breakthrough has the potential to significantly improve the efficiency of turbulence simulations in a variety of engineering applications, making it easier to study and understand these complex phenomena.

Over the past decade, offshore wind has positioned itself as one of the most promising renewable energy markets. While this field is currently dominated by fixed-bottom wind turbines located within a limited depth range, floating turbines are showing promise as a way to capture the more developed wind profiles available in deeper waters. Currently, the main challenge with floating offshore wind is that the systems experience larger ultimate loads compared to fixed bottom turbines. These larger loads are caused by the increased motion inherent with floating structures. This study looks to analyze the effects that traditional and inerter based structural control methods can have on vibration reduction of floating offshore wind turbines. Models are developed adding tuned mass dampers (TMD) and tuned inerter dampers (TID) into the three main columns of a semisubmersible platform. Results showed that for free decay tests, heave and pitch root mean square (RMS) values were reduced significantly by the addition of passive structural control. The inerter based structural control consistently outperformed traditional TMD and also allowed for similar performance with significantly reduced physical mass values. For regular wave excitation both control methods resulted in significant reductions to the heave and pitch RMS values compared to the baseline, with the TID outperforming the TMD . And for an irregular wave analysis, it was found that both control configurations were still able to provide meaningful reductions to the baseline model. Master of Science Over the past decade, offshore wind has positioned itself as one of the most promising renewable energy markets. While this field is currently dominated by fixed-bottom wind turbines located within a limited depth range, floating turbines are showing promise as a way to capture the more developed wind profiles available in deeper waters. Currently, the main challenge with floating offshore wind is that the systems experience larger ultimate loads compared to fixed bottom turbines. These larger loads are caused by the increased motion inherent with floating structures. This study looks to analyze the effects that traditional and enhanced motion reduction technology can have on floating offshore wind turbines. Models are developed adding the traditional and enhanced motion reduction technology into the three main columns of a semisubmersible platform. Results showed that for several different tests, the motion reduction technology has a positive effect on the turbines. For test dropping the system from a set height, the motion reduction technology allowed the system to come to rest much faster. Moreover, the enhanced technology performed better than the traditional technology. The same results were found when the system was excited by simulated waves.

Ocean wave energy has the potential to play a crucial role in the shift to renewable energy. In order to improve wave energy conversion techniques, a recognition of the sub-optimal nature of traditional sequential design processes due to the interconnectedness of subsystems such as the geometry, power take-off, and controls is necessary. A codesign optimization in this paper seeks to include effects of all subsystems within one optimization loop in order to reach a fully optimal design for an oscillating-surge wave energy converter. A width and height sweep serves as a brute force geometry optimization while optimizing the power take-off components and controls using a pseudo-spectral method for each geometry. An investigation of electrical power and mechanical power maximization also outlines the contrasting nature of the two objectives to illustrate electrical power maximization's importance for identifying optimality. The codesign optimization leads to an optimal design with a width of 12 m and a height of 10 m. The power take-off and controls systems are also examined more in depth to identify important areas for increased focus during detailed design. Ultimately, the codesign optimization leads to a 61.4% increase in the objective function over the optimal design from a sequential design process while also requiring about half the power take-off torque. Master of Science Ocean wave energy has the potential to play a crucial role in the shift to renewable energy sources. The Earth's vast oceans have immense energy potentials throughout the world, which often follow the seasonal trends of electricity demand in temperate climates. Wave energy harvesting is a technology which has been studied significantly, but has not yet experienced commercial success, partially due to the lack of convergence on a type of wave energy converter. In order to improve wave energy conversion techniques and support the convergence on a particular type, a recognition of the sub-optimal nature of traditional sequential design processes due to the interconnectedness of subsystems is necessary. A codesign optimization in this paper seeks to include effects of all subsystems within one optimization loop in order to reach a fully optimal design for an oscillating-surge wave energy converter. A width and height sweep serves as a brute force geometry optimization while optimizing the power take-off and control components for each geometry. The codesign optimization leads to an optimal design with a width of 12 m and a height of 10 m. Ultimately, the codesign optimization leads to a 62% increase in performance over the result from a sequential design process.

The point absorber is one of the most popular types of ocean wave energy converter (WEC) that harvests energy from the ocean. Often such a WEC is deployed in an ocean location with tidal currents or ocean streams, or serves as a mobile platform to power the blue economy. The shape of the floating body, or buoy, of the point absorber type WEC is important for the wave energy capture ratio and for the current drag force. In this work, a new approach to optimize the shape of the point absorber buoy is developed to reduce the ocean current drag force on the buoy while capturing more energy from ocean waves. A specific parametric modeling is constructed to define the shape of the buoy with 12 parameters. The implementation of neural networks significantly reduces the computational time compared to solving hydrodynamics equations for each iteration. And the optimal shape of the buoy is solved using a genetic algorithm with multiple self-defined functions. The final optimal shape of the buoy in a case study reduces 68.7% of current drag force compared to a cylinder-shaped buoy, while maintaining the same level of energy capture ratio from ocean waves. The method presented in this work has the capability to define and optimize a complex buoy shape, and solve for a multi-objective optimization problem. Master of Science The marine kinetic energy includes ocean waves power, tidal power, ocean current power, ocean thermal power and river power. The total potential marine kinetic energy in 2021 is 2300 TWh/year, where 1400 TWh/year is from the ocean wave power. To discover and harvest the huge potential power from the marine, researchers have been developed for different types of WECs for several decades. One of the most successful concepts is the point absorber typed WEC, which can extract waver energy from the heaving vibration motion of a floating body and convert the kinetic energy into electrical energy. This thesis presents an optimization strategy to optimize the shape of the floating body to improve power extraction and reduce the installation cost by implementing the machine learning tool and genetic algorithm. Compared with the state-of-the-art optimization strategies, the proposed optimization method allows the floating body to have more parameters in shape changes and reduces the computational cost from minutes to milliseconds. The final optimized floating body shape performs extraordinarily compared to the other two state-of-the-art floating body shapes.

The accuracy of plenoptic and tomographic particle image velocimetry (PIV) experimental methods is measured by simulating three-dimensional flows and measuring the errors in the estimated versus true velocity fields. Parametric studies investigate the accuracy of these methods by simulating a range of camera numbers, camera angles, calibration errors, and particle densities. The plenoptic simulations combine lightfield imaging techniques with standard tomographic techniques and are shown to produce higher fidelity measurements than either technique alone. The tomographic PIV simulations are centered around testing software developed for processing large quantities of data that were produced during an experimental investigation of the flow field about a 3D printed model of the flying snake Chrysopelea paradisi. A description of this tomographic PIV experiment is given along with basic results and recommendations for future investigation. Doctor of Philosophy Two different experimental measurement techniques that can be used to measure three-dimensional fluid flow fields are discussed. The first measurement technique that is investigated in simulations uses cameras with arrays of lenses to simultaneously capture images of a flow field from multiple different angles. A method of combining the data from multiple cameras is discussed and shown to yield more accurate estimates of the three-dimensional flow fields than from a single camera alone. An additional measurement technique that uses a group of standard cameras to measure three-dimensional flow fields is also discussed with respect to software that was developed for processing a large volume dataset. This software was developed for processing data collected during an experimental investigation of the flow field about a 3D printed model of the flying snake Chrysopelea paradisi. A description of this experiment is given along with basic results and recommendations for future investigation.

In recent years, the large amount of labeled data available has helped tend machine learning (ML) research toward using purely data driven end-to-end pipelines, e.g., in deep neural network research. However, in many situations, data is limited and of poor quality. Traditional ML pipelines are known to be susceptible to various issues when trained on low volumes of non-representative, noisy datasets. We investigate the question of whether prior domain knowledge about the problem being modeled can be employed within the ML pipeline to improve model performance under data paucity and in noisy contexts? This report presents recent developments as well as details, novel contributions in the context of incorporating prior domain knowledge in various data-driven modeling (i.e., machine learning - ML) pipelines particularly geared towards scientific applications. Such domain knowledge exists in various forms and can be incorporated into the machine learning pipeline using different implicit and explicit methods (termed: science-guided machine learning (SGML)). All the novel techniques proposed in this report have been presented in the context of developing SGML to model fluid dynamics applications, but can be easily generalized to other applications. Specifically, we present SGML pipelines to (i) incorporate prior domain knowledge into the ML model architecture (ii) incorporate knowledge about the distribution of the target process as statistical priors for improved prediction performance (iii) leverage prior knowledge to quantify consistency of ML decisions with scientific principles (iv) explicitly incorporate known mathematical relationships of scientific phenomena to influence the ML pipeline (v) develop science-guided transfer learning to improve performance under data paucity. Each technique that is presented, has been designed with a focus on simplicity and minimal cost of implementation with a goal of yielding significant improvements in model performance especially under low data volumes or under noisy data conditions. In each application, we demonstrate through rigorous qualitative and quantitative experiments that our SGML pipelines achieve significant improvements in performance and interpretability over corresponding models that are purely data-driven and agnostic to scientific knowledge. Doctor of Philosophy In this work, we present techniques for incorporating scientific knowledge into machine learning (ML) pipelines. We demonstrate these techniques with ML models trained with low data volumes as well as with non-representative, noisy datasets. In both these cases, we demonstrate through rigorous experimentation that incorporating scientific domain knowledge into the ML pipeline using our proposed science guided machine learning (SGML) techniques, leads to significant performance improvement.

Master of Science Aviation gas turbine engine particle ingestion is known to reduce engine lifespans and even pose a threat to safe operation in the worst case. Particles being ingested into an engine can be modeled using multiphase flow techniques. Devices called inertial particle separators are designed to remove particles from the flow into the engine. One challenge with designing such a separator is figuring out how to efficiently expel the small particles from the flow while not unnecessarily increasing pressure loss with excessive twists and turns in the geometry. Designers usually have to develop such geometries using multiphase flow computational fluid dynamics (CFD) that solve the fluid and particle dynamics. The abundance of data associated with CFD, and especially multiphase flows make it an ideal application to study with machine learning (ML). Because such multiphase simulations are very computationally expensive, it is desirable to develop "cheaper" methods. This is the long term goal of this work; we want to create ML surrogates that decrease the computational cost of simulating the particle and fluid flow in particle separator geometries such that designs can be iterated more quickly. In this work we introduce how artificial neural networks (ANNs), which are a tool used in ML, can be used to predict particle acceleration in fluid flow. The ANNs are shown to learn the acceleration predictions with acceptable accuracy for the training data generated with canonical flow cases. However, the ML model struggles to become generalizable to actual CFD simulations.

M.S. Electronics, wearable devices are important in nowadays informationalized lifestyle. One prominent problem with those electronic devices is that almost all of them depend on batteries as power sources, which has become a bottleneck due to the limited life span. Constantly replacing or recharging batteries is inconvenient, burdensome, and sometimes even impossible. This problem is more intractable when the power cannot be accessed conveniently (such as during fieldwork, hiking, and military missions). What’s more, no matter how much energy the battery stores, it will drain eventually. In addition, large battery will add extra weight and occupy space. Substitute power supply that conquer these aforementioned dilemmas are thus highly desirable. Energy harvesting by its nature could be an inexhaustible replacement for batteries. This insight inspires so many energy harvesting researchers tirelessly working and trying to make it happen. Suspended backpack is an effective way to harvest energy from human motions.This study evaluates different energy harvesting backpack configurations and comprehensively assessed the power generation, dynamic interaction and human comfort. Dynamic modelling considering the dynamic interaction between human body and backpack is established to optimize the design. Bench tests and treadmill tests are carried out to evaluate the real performance. Experimental results show that the harvesting energy from human motion via a suspended energy-harvesting backpack could incessantly generate considerable electricity applicable for charging carry-on electronic devices. The potential application scenarios of this technology include solders, field-workers as well as outdoor adventure.

With the aim to understand the synergistic roles played by kinematics, aerodynamics, and inertia in flapping wing maneuvers, this thesis first investigates the plunging motion of a simple flat plate as it is a fundamental motion in the kinematics of many flying animals. A wide range of frequency (k) and amplitude (h) is investigated to account for a robust kinematic characterization in the form of plunge velocity (kh). Leading Edge Vortices (LEVs) are found to be responsible for producing thrust while Trailing Edge Vortices (TEVs) produce drag. The vortex dynamics becomes nonlinear for higher kh and three main vortex-vortex interactions (VVI) are identified in the flow-field. To estimate the sole effect of LEVs on thrust coefficient, TEVs are eliminated by introducing a splitter plate. This resulted in reduced non-linearity in VVI and facilitated a parametrization of aerodynamic thrust coefficient with key kinematic features, frequency (k) and amplitude (h) [C_T= A.k^1.4 h-B where A and B are constants]. This is followed by investigating the more direct problem of bio-inspired MAV research – the interplay of kinematics, aerodynamics, and inertia on maneuvering bat flights. At first, an ascending right turn of a H. pratti bat is investigated to elucidate on the kinematic features and aerodynamic mechanisms used to effectuate the maneuver. Deceleration in flight speed, an increase in flapping frequency, shortening of the upstroke, and thrust generation at the end of the upstroke is observed during this maneuver. The turn is initiated by the synergisytic implementation of roll and yaw rotation where the turning moments are generated by drawing the inside wing closer to the body, by introducing phase lags in force generation between the two wings and by redirecting force production to the outer part of the wing outside of the turn. Upon comparison with a similar maneuver by a H. armiger bat, some commonalities as well as differences were observed. This analysis was followed by a comparative study among different maneuvering flights (a straight flight, two ascending right turns, and a U-turn) in order to establish the complete motion dynamics of a maneuver in action. The individual effects of aerodynamics and wing inertia for maneuvering flights of a H. armiger and H. pratti are investigated. It is found that for both, translation and rotation the overall trajectory trend is mostly driven by the aerodynamic forces and moments, whereas inertial effects drive the intricate intra-cycle fluctuations as well as the vertical velocity and altitude gain during ascent. Additionally, inertial moments play a dominant role for effecting yaw rotations where the importance of the Coriolis and centrifugal moments increase with increasing acuteness of the maneuver, with the largest effect of centrifugal moments being evidenced in the U-turn. Doctor of Philosophy The study of flapping wing is of paramount interest in the field of small aerial and aquatic vehicle propulsion. The intricate mechanisms acting behind a flapping wing maneuver can be explained by the synergistic roles played by 3 main components; details of the wing motion or the kinematics, how the air reacts to the wing motion or the aerodynamics, and the effort or force required to move the wings or wing inertia. This dissertation systematically reports the contribution of these components to a flapping flight maneuver. At first, the plunging motion of a simple flat plate is investigated as it is a fundamental motion in the flapping flight of many flying animals. A wide range of frequency and amplitude is investigated and their effect is characterized by a single parameter called "plunge velocity". It is found that, the resultant flow field becomes disorderly for higher plunge velocities which can be characterized by three different types of vortex interactions. The observed results facilitated a robust parametrization of aerodynamic thrust production with key kinematic features, frequency and amplitude. After this, the dissertation focuses on the bio-inspiration aspect of flapping flight by investigating the interplay of kinematics, aerodynamics, and inertia of maneuvering bat flights. At first, an ascending right turn of one species (H. pratti) is investigated to elucidate on the kinematic features and aerodynamic mechanisms used to effectuate the maneuver. Some characteristic features observed are – lowering of flight speed, increase in flapping rate, shortening of upstrokes, and generation of a forward force at the end of the upstroke. It is observed, that the bat turns by using synergistic body rotations in multiple directions which are effected by various techniques such as - drawing the wing inside the turn closer to the body, and changing the timing and location of the forces produced between the two wings. Upon comparison with a similar maneuver by a H. armiger bat, some commonalities as well as differences were observed in the maneuver mechanisms. This analysis was followed by a comparative study among different maneuvering flights (a straight flight, two ascending right turns, and a U-turn) to establish the complete motion dynamics of a maneuver. The individual contributions of aerodynamics and wing inertia for maneuvering flights of a H. armiger and H. pratti are investigated. It is found that for both, translation and rotation the overall trajectory is mostly influenced by the aerodynamic forces and moments, whereas inertial effects are responsible for trajectory fluctuations during a flapping cycle as well contributing to altitude gain during ascent for the H. armiger bat.

One of the principal impediments to the broad success of conventional chemotherapy is poor delivery to and transport within the tumor microenvironment (TME), caused by irregular and leaky vasculature, the lack of functional lymphatics, and underscored by the overproduction of extracellular matrix (ECM) proteins such as collagen. Coupled with limited specificity, the high chemotherapeutic doses needed to effectively treat tumors often lead to unacceptable levels of damage to healthy tissues. Bacteria-based cancer therapy (BBCT) is an innovative alternative. Attenuated strains of species such as Salmonella Typhimurium have been shown to preferentially replicate in the TME, competing for cellular resources and imparting intrinsic and immune-mediated cytotoxic effects on cancer cells. Nevertheless, the immense successes observed in in vitro and immunocompromised murine models have not translated to the clinic, attributable to the lack of sufficient tumor colonization. Synthetic biology today enables the precision engineering of microbes with traits for improved survival, penetration, and replication in the TME, rationally optimizable through computational modeling. In this dissertation, we explore several facets of rationally engineering of bacteria toward augmenting bacterial penetration and retention in the TME. Namely, we (1) develop a novel assay to interrogate the neutrophil migratory response to pathogens and characterize the effects of modifying the molecular structure of the outer membrane (OM) of S. Typhimurium, (2) develop a mathematical model of bacterial intratumoral transport and growth and explore the effects of nutrient availability and the tumor ECM on colonization, (3) engineer bacteria that constitutively secrete collagenase and show significantly augmented transport in collagen hydrogels and collagen-rich tumor spheroids, and (4) develop computational models to explore control schemes for the programmed behavior of bacteria-based biohybrid systems, which will leverage the engineered bacteria to deliver therapeutics to the TME. Our work serves as the foundation for the logical and efficient design of the next generation of BBCTs. Doctor of Philosophy Cancer is one of the deadliest diseases facing our world today not because of a lack of effective medications in most cases, but because of our inability to target the diseased sites with those treatments. Many tumors lie in deep and sensitive regions that render them untouchable by direct physical means. Poor vascularization leads to only small fractions of toxic, systemically injected drugs being deposited in tumors. State-of-the-art treatments such as so-called "nano-medicines" that can target features of the diseased tissues and immunotherapies that train the immune system to attack tumor cells have made tremendous strides, but for many types of cancer, the underlying challenge of reaching cells far from blood vessels and targeting immunologically cold tumors remains. Bacteria-based cancer therapy (BBCT) presents an exciting opportunity to address these challenges. Based on microorganisms that can self-propel, proliferate, and display a preference for diseased tissues, their potential not only to carry chemotherapeutic payloads but also to elicit directly toxic or immunotherapeutic effects on cancer cells is clear from experimental work. Nevertheless, the same delivery and transport barriers facing other treatments, as well as immune-mediated clearance, have limited BBCTs' clinical success. Advances in synthetic biology and computational modeling today make the precision engineering of BBCT for traits that favor targeted cancer therapy a reality. The central hypothesis of this dissertation is that endowing tumor-targeting bacteria with a tissue-degrading enzyme has the potential to enhance tumor penetration and colonization. This dissertation work has led to development of computational and experimental frameworks for the design, testing, and optimization of BBCTs through direct quantitative assessment of the immune response, simulations to both optimize nutrient consumption for optimal growth and for programming genetic control strategies, and characterization of transport in tissue. Our work serves as a foundation for engineering "intelligent" BBCT.

Point absorber style marine energy harvesters have been investigated based on their structure, energy harvesting efficiency, and reliability along with costs. However, due to the continuously varying ocean conditions and climates, the system usually suffers low power output and reliability from low input and high Peak to Average Ratio (PAR). Therefore, a Mechanical Motion Rectifier (MMR) based point absorber is introduced in this thesis to promote the harvesting efficiency and reduce the PAR by unifying the input rotation, and allow disengagement inside the gearbox during low power output phase. A 1:20 scale full system was then designed, prototyped, and tested based on the MMR. The bench test results show that the proposed MMR based point absorber could improve the energy conversion efficiency by 10 percent, which brings feasibility to the implementation. Traditional Wave Energy Converter(WEC) can only harvest ocean waves while ocean current is also one of the significant energy sources widely existing in ocean. In order to further increase the energy harvesting efficiency, one individual energy input source shows its limits. A vast majority of places around the world tends to co-exist both marine waves and current, and extracting energy from both sources could potentially increase the electric power output. Therefore, the Hybrid Wave and Current Energy Harvester (HWCEC) is introduced along with the hybrid gearbox. It is capable of harvesting energy from both ocean waves and current simultaneously so that the electric power output is significantly higher from a combined system. Tank test data shows 38-79 percent of electric power output promotion of an HWCEC compared to a regular WEC, and 70 percent reduced PAR in irregular wave condition. After that, system electric damping has been thoroughly investigated on both electrical side and mechanical side. The best power output corresponding electrical resistance is identical to the generator internal resistance while the best gear ratio of 3.5 is determined via both simulation and tank test. Furthermore, the system's PAR has been investigated by analyzing the trend of the peak occurrence. Tank test data shows the HWCEC's output power peak occurrence is at roughly 20 percent located at its PAR average. Therefore, the HWCEC system can promote energy harvesting efficiency to the combined system design, and improve its reliability from a significantly reduced peak to average ratio. It also gives HWCEC a large variety of deployable locations compared to a regular WEC under more marine environment. Furthermore, a new design of the Hybrid model, Hybrid LITE, is then developed, which not only features the HWCEC features, but also a lightweight, immersive and inflatable design for fast deployment and transportation. Since the system is built with an open water chassis, the overall system robustness is significantly improved since no water sealing is required on the powertrain compared to the HWCEC. Master of Science Ocean contains enormous amount of Marine Hydrokinetic (MHK) energy including ocean waves, tidal streams, and ocean current. Marine energy was investigated due to its continuous, massive and high-density hydrokinetic power output. In order to better serve the needs for ocean surface applications and take advantage of high energy density compared to other renewable energy sources, Wave Energy Converters (WEC) has been investigated, which harvests energy from the ocean wave. In the past years of study, it came to our attention that places such as the west coast of the U.S., northern Europe, and the Mediterranean area tend to have both abundant marine wave and current energy. Therefore, a new design of the Hybrid Wave and Current Energy Converter (HWCEC) is investigated for higher power output. In order to combine the energy sources from waves and current, a Hybrid Gearbox was selected to joint the power and unifies the motion from the wave for a higher efficiency. Simulations and 1:10 ratio co-existing wave and current basin test have been conducted for the HWCEC. By using the same system, single wave or current input are used as the baselines and the dual input HWCEC has demonstrated great benefit and potential. The electric damping and the gearbox ratio of the HWCEC are studied for the best power output in both simulation and tank test. The result shows that the HWCEC could promote up to 38-71 percent of electricity output in a regular wave condition, and 79 percent in irregular wave condition. The Peak to Average Ratio (PAR) is a key factor for system's mechanical reliability. The testing shows that the HWCEC can reduce 70 percent of the peak motion and contribute to the average, which is an indirect indicator of the system's better reliability. Furthermore, to align the needs of the design for real-life applications, The Hybrid LITE Converter idea was then developed for special deployment requirements for the future application of the Hybrid system. It has a novel open-system design with the implementation of a newly designed hybrid gearbox. This converter has the potential of promoting the reliability, deployability and weight reduction for easy transportation from its open system design compared to HWCEC. The system modeling could be done as future work varies from the changing deployment locations for higher electric power output.

The detonation-based engine has the key advantage of increased thermodynamic efficiency over the traditional constant pressure combustor. These detonation-based engines are also known as Pressure Gain Combustion systems (PGC) and Rotating Detonation Combustor (RDC) is a form of PGC, in which the detonation wave propagates azimuthally around an annular combustor. Prior researchers have performed a high fidelity 3-D numerical simulation of a rotating detonation combustor (RDC) to understand the flow physics such as detonation wave velocity, pressure profile, wave structure; however, performing these 3-D simulations is computationally expensive. 2-D simulations are a potential alternative to reduce computational cost. In most RDCs, fuel and oxidizer are injected discretely from separate plenums, and this discrete fuel/air injection results in inhomogeneous mixing within the domain. Due to the discrete fuel injection locations, fuel/oxidizer will stratify to form localized pockets of rich and lean mixtures. The motivation of the present study is to investigate the impact of unmixedness and stratification length scales on the performance of an RDC using a 2-D numerical approach. Unmixedness, which is defined as the standard deviation of equivalence ratio normalized by the mean global equivalence ratio, is a measure of the degree of fuel-oxidizer inhomogeneity. To model the effect of unmixedness in a 2-D domain, a lognormal distribution of the fuel mass fraction is generated with a mean equivalence ratio of 1 and varying standard deviations at the inlet boundary as a numerical source term. Moreover, to model the effects of stratification length scales, fuel mass fraction at the inlet boundary cells is bundled for a given length scale, and the mass fractions for these bundles are updated based on the lognormal distribution after every three-time steps. Using this methodology, 2-D numerical analyses are carried out to investigate the performance of an RDC for an H2-air mixture with varying unmixedness and stratification length scales. Results show that mean detonation velocity decreases and wave speed variation increases with an increase in unmixedness. However, with an increase in stratification length scale mean velocity remain relatively unchanged but variation in local velocity increases. The detonation wave front corrugation also increases with an increase in mixture inhomogeneity. The mean detonation cell size increases with an increase in unmixedness. The cell shape becomes more distorted and irregular with an increase in stratification length scale and unmixedness. The combined effect of unmixedness and stratification length scale leads to a decrease in pressure gain. Overall, this concept is able to elucidate the effects of varying unmixedness and stratification length scales on the performance of an RDC. Master of Science Pressure Gain Combustion (PGC) system has gained significant focus in recent years due to its increased thermodynamic efficiency over a constant pressure Brayton Cycle. Rotating Detonation Combustor (RDC) is a type of PGC system, which is thermodynamically more efficient than the conventional gas turbine combustor. One of the main aspects of the detonation process is the rapid burning of the fuel-oxidizer mixture, which occurs so fast that there is not enough time for pressure to equilibrate. Therefore, the process is thermodynamically closer to a constant volume process rather than a constant pressure process. A constant volume cycle is thermodynamically more efficient than a constant pressure Brayton cycle. In an RDC, a mixture of fuel and air is injected axially, and a detonation wave propagates continuously through the circumferential section. Numerical simulation of an RDC provides additional flexibility over experiments in understanding the flow physics, detonation wave structure, and analyzing the physical and chemical processes involved in the detonation cycle. Prior researchers have utilized a full-scale 3-D numerical simulation for understanding the performance of an RDC. However, the major challenge with 3-D analyses is the computational expense. Thus, to overcome this, an inexpensive 2-D simulation is used to model the flow physics of an RDC. In most RDCs, the fuel and oxidizer are injected discretely from separate plenums. Due to the discrete fuel injection, the fuel/air mixture is never perfectly premixed and results in a stratified flow field. The objective of the current work is to develop a novel approach to independently investigate the effects of varying unmixedness and stratification length scales on RDC performance using a 2-D simulation.

Investigation of momentum and heat transfer between the fluid and solid phase is critical to the study of fluid-particle systems. Dense suspensions are characterized by the solid fraction (ratio of solid volume to total volume), the particle Reynolds number, and the shape of the particle. The behavior of non-spherical particles deviates considerably from spherical particle shapes which have been studied extensively in the literature. Momentum transfer, to first-order, is driven by drag forces experienced by the particles in suspension, followed by lift and lateral forces, and also through the transmission of fluid torque to the particles. The subject of this thesis is a family of prolate ellipsoidal particle geometries of aspect ratios (AR) 2.5, 5.0 and 10.0 at nominal solid fractions (φ) between 0.1 and 0.3, and suspensions of cylinders of AR=0.25. The nominal particle Reynolds number (Re) is varied between 10 to 200, representative of fluidized beds. Fluid forces and heat transfer coefficients are obtained numerically by Particle Resolved Simulations (PRS) using the Immersed Boundary Method (IBM). The method enables the calculation of the interstitial flow and pressure field surrounding each particle in suspension leading to the direct integration of fluid forces acting on each particle in the suspension. A substantial outcome of the research is the development of a new drag force correlation for random suspensions of prolate ellipsoids over the full range of geometries and conditioned studied. In many practical applications, especially as the deviation from the spherical shape increases, particles are not oriented randomly to the flow direction, resulting in suspensions which have a mean preferential orientation. It is shown that the mean suspension drag varies linearly with the orientation parameter, which varies from -2.0 for particles oriented parallel to the flow direction to 1.0 for particles normal to the flow direction. This result is significant as it allows easy calculation of drag force for suspension with any preferential orientation. The heat transfer coefficient or Nusselt number is investigated for prolate ellipsoid suspensions. Significantly, two methods of calculating the heat transfer coefficient in the literature are reconciled and it is established that one asymptotes to the other. It is also established that unlike the drag force, at low Reynolds number the suspension mean heat transfer coefficient is very sensitive to the spatial distribution of particles or local-to-particle solid fractions. For the same mean solid fraction, suspensions dominated by particle clusters or high local solid fractions can exhibit Nusselt numbers which are lower than the minimum Nusselt number imposed by pure conduction on a single particle in isolation. This results from the dominant effect of thermal wakes at low Reynolds numbers. As the Reynolds number increases, the effect of particle clusters on heat transfer becomes less consequential. For the 0.25 aspect ratio cylinder, it was found that while existing correlations under predicted the drag forces, a sinusoidal function F_(d,θ)=F_(d,θ=0°)+(F_(d,θ=90°)-F_(d,θ=0°) )sin⁡(θ) captured the variation of normalized drag with respect to inclination angle over the range 10≤Re≤300 and 0≤φ≤0.3. Further the mean ensemble drag followed F_d=F_(d,θ=0°)+1/2(F_(d,θ=90°)-F_(d,θ=0°)). It was shown that lift forces were between 20% to 80% of drag forces and could not be neglected in models of fluid-particle interaction forces. Comparing the pitching fluid torque to collision torque during an elastic collision showed that as the particle equivalent diameter, density, and collision velocities decreased, fluid torque could be of the same order of magnitude as collisional torque and it too could not be neglected from models of particle transport in suspensions. Doctor of Philosophy Momentum and heat exchange between the fluids (air, water…) and suspensions of solid particles plays a critical role in power generation, chemical processing plants, pharmaceuticals, in the environment, and many other applications. One of the key components in momentum exchange are the forces felt by the particles in the suspension due to the flow of the fluid around them and the amount of heat the fluid can transfer to or from the particles. The fluid forces and heat transfer depend on many factors, chief among them being the properties of the fluid (density, viscosity, thermal properties) and the properties of the particles in the suspension (size, shape, density, thermal properties, concentration). This introduces a wide range of parameters that have the potential to affect the way the fluid and particles behave and move. Experimental measurements are very difficult and expensive to conduct in these systems and computational modeling can play a key role in characterization. For accuracy, computational models have to have the correct physical laws encoded in the software. The objective of this thesis is to use very high-fidelity computer models to characterize the forces and heat transfer under different conditions to develop general formulas or correlations which can then be used in less expensive computer models. Three basic particle shapes are considered in this study, a sphere, a disk like cylindrical particles, and particles of ellipsoidal shapes. More specifically, Particle Resolved Simulations of flow through suspensions of ellipsoids with aspect ratio of 2.5, 5, 10 and cylinders with aspect ratio of 0.25 are performed. The Reynolds number range covered is [10, 200] for ellipsoids and [10, 300] for cylinders with solid fraction range of [0.1, 0.3]. New fluid drag force correlations are proposed for the ellipsoid and cylinder suspensions, respectively, and heat transfer behavior is also investigated.

In this thesis, a continuum sensitivity analysis method is presented for calculation of shape sensitivities of an elastic bar. The governing differential equations and boundary conditions for the elastic bar are differentiated with respect to the shape design parameter to derive the continuum sensitivity equations. The continuum sensitivity equations are linear ordinary differential equations in terms of local or material shape design derivatives, otherwise known as shape sensitivities. One of the novelties of this work is the derivation of three variational formulations for obtaining shape sensitivities, one in terms of the local sensitivity and two in terms of the material sensitivity. These derivations involve evaluating (a) the variational form of the continuum sensitivity equations, or (b) the sensitivity of the variational form of the analysis equations. We demonstrate their implementation for various combinations of design velocity and global basis functions. These variational formulations are further solved using finite element analysis. The order of convergence of each variational formulation is determined by comparing the sensitivity solutions with the exact solutions for analytical test cases. This research focusses on 1-D structural equations. In future work, the three variational formulations can be derived for 2-D and 3-D structural and fluid domains. Master of Science When solving an optimization problem, the extreme value of the performance metric of interest is calculated by tuning the values of the design variables. Some optimization problems involve shape change as one of the design variables. Change in shape leads to change in the boundary locations. This leads to a change in the domain definition and the boundary conditions. We consider a 1-D structural element, an elastic bar, for this study. Subsequently, we demonstrate a method for calculating the sensitivity of solution (e.g. displacement at a point) to change in the shape (length for 1-D case) of the elastic bar. These sensitivities, known as shape sensitivities, are critical for design optimization problems. We make use of continuum analytical shape sensitivity analysis to derive three variational formulations to compute these shape sensitivities. The accuracy and convergence of solutions is verified using a finite element analysis code. In future, the approach can be extended to multi-dimensional structural and fluid domain problems.

Ocean wave as a renewable energy source possesses great potential for solving the world energy crisis and benefit human beings. The total theoretical potential wave power on the ocean-facing coastlines of the world is around 30,000 TWh, although cannot all be adopted for generating electricity, the amount of the power can be absorbed still can occupy a large portion of the world's total energy consumption. However, multiple reasons have stopped the ocean wave energy from being widely adopted, and among those reasons, the most important one is immature of the Power Take-off (PTO) technology. In this dissertation, a self-reactive two-body wave energy point absorber that is embedded with a novel PTO using the unique mechanism of Mechanical Motion Rectifier (MMR) is investigated through design, analysis and testing to improve the energy harvesting efficiency and the reliability of the PTO. The MMR mechanism can transfer the reciprocated bi-directional movement of the ocean wave into unidirectional rotation of the generator. As a result, this mechanism brings in two advantages towards the PTO. The first advantage it possess is that the alternating stress of the PTO is changed into normal stress, hence the reliability of the components are expected to be improved significantly. The other advantage it brings in is a unique phenomenon of engagement and disengagement during the operation, which lead to a piecewise nonlinear dynamic property of the PTO. This nonlinearity of the PTO can contribute to an expanded frequency domain bandwidth and better efficiency, which are verified through both numerical simulation and in-lab experiment. During the in-lab test, the prototyped PTO achieved energy transfer efficiency as high as 81.2%, and over 40% of efficiency improvement compared with the traditional non-MMR PTO under low-speed condition, proving the previously proposed advantage. Through a more comprehensive study, the MMR PTO is further characterized and a refined dynamic model. The refined model can accurately predict the dynamic response of the PTO. The major factors that can influence the performance of the MMR PTO, which are the inertia of the PTO, the damping coefficient, and the excitation frequency, are explored through analysis and experiment comprehensively. The results show that the increase on the inertia of the PTO and excitation frequency, and decrease on the damping coefficient can lead to a longer disengagement of the PTO and can be expressed analytically. Besides the research on the PTO, the body structure of the point absorber is analyzed. Due to the low-frequency of the ocean wave excitation, usually a very large body dimension for the floating buoy of the point absorber is desired to match with that frequency. To solve this issue, a self-reactive two-body structure is designed where an additional frequency between the two interactive bodies are added to match the ocean wave frequency by adopting an additional reactive submerged body. The self-reactive two-body structure is tested in a wave to compare with the single body design. The results show that the two-body structure can successfully achieve the frequency matching function, and it can improve more than 50% of total power absorption compared with the single body design. Doctor of Philosophy Ocean wave as a renewable energy source possesses great potential for solving the world energy crisis and benefit human beings. The total theoretical potential wave power on the ocean-facing coastlines of the world is around 30,000 TWh, although impossible to be all transferred into electricity, the amount of the power can be absorbed still can cover a large portion of the world's total energy consumption. However, multiple reasons have stopped the ocean wave energy from being widely adopted, and among those reasons, the most important one is immature of the Power Take-off (PTO) technology. In this dissertation, a novel two body wave energy converter with a PTO using the unique mechanism of Mechanical Motion Rectifier (MMR) is investigated through design, analysis, and testing. To improve the energy harvesting efficiency and the reliability of the PTO, the dissertation induced a mechanical PTO that uses MMR mechanism which can transfer the reciprocated bi-directional movement of the ocean wave into unidirectional rotation of the generator. This mechanism brings in a unique phenomenon of engagement and disengagement and a piecewise nonlinear dynamic property into the PTO. Through a comprehensive study, the MMR PTO is further characterized and a refined dynamic model that can accurately predict the dynamic response of the PTO is established. The major factors that can influence the performance of the MMR PTO are explored and discussed both analytically and experimentally. Moreover, as it has been theoretically hypothesis that using a two-body structure for designing the point absorbers can help it to achieve a frequency tuning effect for it to better match with the excitation frequency of the ocean wave, it lacks experimental verification. In this dissertation, a scaled two-body point absorber prototype is developed and put into a wave tank to compare with the single body structure. The test results show that through the use of two-body structure and by designing the mass ratio between the two bodies properly, the point absorber can successfully match the excitation frequency of the wave. The highest power capture width ratio (CWR) achieved during the test is 58.7%, which exceeds the results of similar prototypes, proving the advantage of the proposed design.

Bats have many impressive flight characteristics such as the ability to rapidly change direction, carry substantial loads, and maintain good flight efficiency. For several years, researchers have been working towards an understanding of the specific aerodynamic phenomena which relate the unique wing structure of bats to their flight abilities. Computational fluid dynamics, a powerful tool used extensively across aerospace research, has led to substantial progress in the understanding of insect flight. However, due to technical challenges, numerical simulation has seen limited use in bat flight research. For this research, we develop, validate, and apply computational modeling techniques to three modes of bat flight: straight flight, sweeping turn, and U-turn maneuver. 3D kinematic data collection was achieved using a 28 camera multi-perspective optical motion capture system. The calibration of the cameras was conducted using a multi-camera self-calibration method. Point correspondences between cameras and frames was achieved using a human-supervised software package developed for this project. After the collection of kinematic data, we carried out aerodynamic flow simulations using the incompressible Navier-Stokes solver, GenIDLEST. The immersed boundary method (IBM) was used to impose moving boundary conditions representing the wing kinematics. Validation of the computational model was preformed through a grid independence study as well as careful evaluation of other relevant simulation parameters. Verification of the model was performed by comparing simulated aerodynamic loads to the expected loads based on the observed flight trajectories. Additionally, we established that we had a sufficient resolution of the wing kinematics, by calculating the sensitivity of the simulation results to the number of kinematic markers used during motion capture. For this study, three particular flights are analyzed—a straight and level flight, a sweeping turn, and a sharp 180 degree turn. During straight flight, typical flight velocities observed in the flight tunnel were 2-3 m/s resulting in a Reynolds number of about 12,000. Lift generation occurred almost exclusively during the downstroke, and peaks mid-downstroke. At the beginning of each downstroke, the effective angle of attack of the wings transitions from negative to positive and a leading edge vortex (LEV) quickly forms. LEVs are known to augment lift generation in flapping flight and allow lift to remain high at large angles of attack. During the end of each downstroke, the LEVs break up and lift drops substantially. As the wingbeat cycle transitions from downstroke to upstroke, the wings rotate such that the wing chordline is vertical as the wing moves upward. This wing rotation is critical for mitigating negative lift during the upstroke. Many of the basic flight mechanisms used for straight flight—i.e. LEV formation, wing rotation during upstrokes—were also observed during the sweeping turn. In addition, asymmetries in the wing kinematics and consequently the aerodynamics were observed. Early in the turn, the bank angle was low and elevated levels of thrust were generated by the outer wing during both the upstroke and downstroke causing a yaw moment. As the bat moved towards the middle of the turn, the bank angle increased to 20-25 degrees. Although the bank angle remained nominally constant during the middle and later portion of the turn, there was variation within each wingbeat cycle. Specifically, the bank angle dropped during each upstroke and subsequently was recovered during each downstroke as a consequence of elevated lift on the outer wing. Banking served to redirect the net force vector laterally causing a radial, centripetal force. Considering the mass of the bat, the nominal flight velocity, and the radius of curvature, the magnitude of the radial force fully explained the expected centripetal acceleration during the middle and later portion of the turn. Over the entire turn, yaw was found to be important in initiating the turn while banking was more important during the middle part of the turn. Over the course of 5 wingbeat cycles, the change in bearing angle (direction of flight) was about 45 degrees. Analysis of the U-turn flight showed many of the same characteristics as were observed during the sweeping turn, as well as a few key differences. The bat's ability to rotate its body rapidly appears to be more limited than its ability to change its trajectory. For this reason, the yaw rotation began about one to two cycles before the rapid bearing angle change and was stretched out over several wingbeat cycles. At the apex of the U-turn, the bat combined a high roll angle with a low flight velocity magnitude to very rapidly redirect its bearing direction and negotiate a low radius of curvature flight trajectory. Increases in roll angle occurred almost exclusively during the downstrokes, while both the upstroke and downstroke were active in generating yaw. Elevated thrust on the left outer wing during the end of the upstroke was observed throughout the flight, and elevated drag on the right inside wing did not appear to have an impact on the turn. We hope that this project motivates and facilitates further computational analysis into bat flight aerodynamics. Additionally, the data and findings will be useful for applications such as the design of bioinspired MAVs or flexible membrane energy harvesting technology. Doctor of Philosophy Bats have many impressive flight characteristics such as the ability to rapidly change direction, carry substantial loads, and maintain good flight efficiency. A better understanding of the physics of how bats fly can help scientists and engineers build more maneuverable, quieter, and more efficient bioinspired micro air vehicles. This engineering approach leverages the incredible capabilities observed in nature, but requires detailed knowledge of the animal as a prerequisite. Computational fluid dynamics, a powerful tool used extensively across aerospace research, has led to substantial progress in the understanding of animal flight broadly. However, due to technical challenges, numerical simulation has seen limited use in bat flight research. For this research, we develop, validate, and apply computer modeling techniques to the investigation of bat flight aerodynamics. Three particular modes of flight were analyzed—a straight and level flight, a sweeping turn, and a sharp 180 degree turn. During straight flight, typical flight velocities observed in the flight tunnel were 2-3 m/s. Lift generation, the force keeping the bat aloft, occurred almost exclusively during the downstroke, and peaks mid-downstroke. As the wing flap transitions from downstroke to upstroke, the wings rotate such that the wing is vertical as it moves upward. This wing rotation is critical for maximizing lift force during flight. During the sweeping turn, asymmetries in the wing kinematics and consequently the aerodynamics were observed. Early in the turn, the bank angle was low and elevated levels of thrust were generated by the outer wing during both the upstroke and downstroke causing rotation of the bat. As the bat moved towards the middle of the turn, the bank angle increased to 20-25 degrees. Banking served to redirect the net force vector laterally causing a turning force. Over the course of 5 wingbeat cycles, the change in direction of flight was about 45 degrees. Analysis of the U-turn flight showed many of the same characteristics as were observed during the sweeping turn, as well as a few key differences. At the apex of the U-turn, the bat combined a high roll angle with a low flight velocity magnitude to very rapidly redirect its bearing direction and negotiate a low radius of curvature flight trajectory. We hope that this project motivates and facilitates further computer simulations studying bat flight aerodynamics. Additionally, the data and findings will be useful for applications such as the design of bioinspired MAVs or flexible membrane energy harvesting technology.

Today, we see a renewed interest in aircraft with multiple propellers. To support conceptual design of these vehicles, one of the major needs is a fast and accurate method for estimating wing aerodynamic characteristics in the presence of multiple propellers. For the method to be effective, it must be easy to use, have rapid turnaround time and should be able to capture major wing–propeller interaction effects with sufficient accuracy. This research is primarily motivated by the need to assess the effectiveness of computational fluid dynamics (CFD) for simulating aerodynamic characteristics of wings with multiple propellers. The scope of the present research is limited to investigating the interaction between a single tractor propeller and a wing. This research aims to compare computational results from a Reynolds-Averaged Navier-Stokes (RANS) method, StarCCM+, and a vortex lattice method (VLM), VSP Aero. Two configurations that are analysed are 1) WIPP Configuration (Workshop for Integrated Propeller Prediction) 2) APROPOS Configuration. For WIPP, computational results are compared with measured lift and drag data for several angles of attack and Mach numbers. StarCCM+ results of wake flow field are compared with WIPP's wake survey data. For APROPOS, computed data for lift-to-drag ratio of the wing are compared with test data for multiple vertical and spanwise locations of the propeller. The results of the simulations are used to assess the effectiveness of the two CFD methods used in this research. Master of Science Today, we see a renewed interest in aircraft with multiple propellers due to an increasing demand for vehicles which fly short distances at low altitudes, be it flying taxis, delivery drones or small passenger aircrafts. To support conceptual design of vehicles, one of the major needs is a fast and accurate method for estimating wing aerodynamic characteristics in the presence of multiple propellers. For the method to be effective, it must be easy to use, have rapid turnaround time and should be able to capture major wing–propeller inter- action effects with sufficient accuracy. This research is primarily motivated by the need to assess the effectiveness of computational fluid dynamics (CFD) for simulating aerodynamic characteristics of wings with multiple propellers. Then only can we can take full advantage of the capabilities of the CFD methods and support design of emerging propeller driven air vehicles with an appropriate level of confidence. This research aims to compare high level methods with increasingly complex geometries and realistic models of physics like Reynolds Averaged Navier Stokes (RANS) and low level methods that rely on simplified geometry and simplified physics models like Vortex Lattice Methods (VLM). We will analyse multiple configurations and validate them against experi- mental data and thus assessing the effectiveness of the CFD models. This research investigates two configurations, 1) WIPP configuration 2) APROPOS configuration, for which experimental data is available. The results of the simulations are used to assess the effectiveness of the two CFD methods used in this research.

Transport phenomena in heterogeneous media play a crucial role in numerous engineering applications such as hydrocarbon recovery from shales and material processing. Understanding and predicting these phenomena is critical for the success of these applications. In this dissertation, nanoscale transport phenomena in porous media are studied through physics-based simulations, and the effective solution of forward and inverse transport phenomena problems in heterogeneous media is tackled using data-driven, deep learning approaches. For nanoscale transport in porous media, the storage and recovery of gas from ultra-tight shale formations are investigated at the single-pore scale using molecular dynamics simulations. In the single-component gas recovery, a super-diffusive scaling law was found for the gas production due to the strong gas adsorption-desorption effects. For binary gas (methane/ethane) mixtures, surface adsorption contributes greatly to the storage of both gas in nanopores, with ethane enriched compared to methane. Ethane is produced from nanopores as effectively as the lighter methane despite its slower self-diffusion than the methane, and this phenomenon is traced to the strong couplings between the transport of the two species in the nanopore. The dying of solvent-loaded nanoporous filtration cakes by a purge gas flowing through them is next studied. The novelty and challenge of this problem lie in the fact that the drainage and evaporation can occur simultaneously. Using pore-network modeling, three distinct drying stages are identified. While drainage contributes less and less as drying proceeds through the first two stages, it can still contribute considerably to the net drying rate because of the strong coupling between the drainage and evaporation processes in the filtration cake. For the solution of transport phenomena problems using deep learning, first, convolutional neural networks with various architectures are trained to predict the effective diffusivity of two-dimensional (2D) porous media with complex and realistic structures from their images. Next, the inverse problem of reconstructing the structure of 2D heterogeneous composites featuring high-conductivity, circular fillers from the composites' temperature field is studied. This problem is challenging because of the high dimensionality of the temperature and conductivity fields. A deep-learning model based on convolutional neural networks with a U-shape architecture and the encoding-decoding processes is developed. The trained model can predict the distribution of fillers with good accuracy even when coarse-grained temperature data (less than 1% of the full data) are used as an input. Incorporating the temperature measurements in regions where the deep learning model has low prediction confidence can improve the model's prediction accuracy. Doctor of Philosophy Multiphysics transport phenomena inside structures with non-uniform pores or properties are common in engineering applications, e.g., gas recovery from shale reservoirs and drying of porous materials. Research on these transport phenomena can help improve related applications. In this dissertation, multiphysics transport in several types of structures is studied using physics-based simulations and data-driven deep learning models. In physics-based simulations, the multicomponent and multiphase transport phenomena in porous media are solved at the pore scale. The recovery of methane and methane-ethane mixtures from nanopores is studied using simulations to track motions and interactions of methane and ethane molecules inside the nanopores. The strong gas-pore wall interactions lead to significant adsorption of gas near the pore wall and contribute greatly to the gas storage in these pores. Because of strong gas adsorption and couplings between the transport of different gas species, several interesting and practically important observations have been found during the gas recovery process. For example, lighter methane and heavier ethane are recovered at similar rates. Pore-scale modeling are applied to study the drying of nanoporous filtration cakes, during which drainage and evaporation can occur concurrently. The drying is found to proceed in three distinct stages and the drainage-evaporation coupling greatly affects the drying rate. In deep learning modeling, convolutional neural networks are trained to predict the diffusivity of two-dimensional porous media by taking the image of their structures as input. The model can predict the diffusivity of the porous media accurately with computational cost orders of magnitude lower than physics-based simulations. A deep learning model is also developed to reconstruct the structure of fillers inside a two-dimensional matrix from its temperature field. The trained model can predict the structure of fillers accurately using full-scale and coarse-grained temperature input data. The predictions of the deep learning model can be improved by adding additional true temperature data in regions where the model has low prediction confidence.

Heat transfer augmentation is of paramount importance in energy transfer and storage systems and the idea of using the inherent vibrations in a system to enhance heat transfer needs to be thoroughly researched upon. The current study numerically investigates an infinitesimally thin plate-fin undergoing forced oscillations over a range of amplitudes and frequencies in the presence of an approach flow. Reduced frequencies of 0.25 ≤ k ≤16 and plunge amplitudes of 0.03125 ≤ h ≤ 8 are investigated at Re=100 and Pr = 0.71. It is shown that the combined effect of frequency and amplitude on heat transfer enhancement can be accounted for as a single parameter “plunge velocity” (0.25 ≤ kh ≤ 4) instead of the individual frequency and amplitude values. For kh > 0.5 a significant increase in Nusselt number ( is observed compared to a stationary plate. With increasing kh or more vigorous oscillations, the increase in becomes more prominent and similar trends and comparable magnitudes were observed for a constant value. Unlike the hydrodynamic counterpart of the study, both Leading Edge Vortices (LEVs) and Trailing Edge Vortices (TEVs) are found to act positively to induce enhanced heat transfer on the plate. Finally, the dependence of heat transfer augmentation on the frequency and amplitude of vibration is quantified with a simple parameterization for a plate-fin in a fluid medium. M.S. Heat transfer enhancement is of paramount importance in energy transfer and storage systems. The idea of using the inherent mechanical vibrations in a heat producing system to enhance transfer of unwanted heat from that system needs to be thoroughly researched upon. To investigate this idea, we numerically study an infinitesimally thin plate-fin undergoing forced oscillations over a range of amplitudes and frequencies in the presence of an incoming air flow. It is shown that the combined effect of frequency and amplitude on heat transfer enhancement can be accounted for as a single parameter called “plunge velocity” instead of the individual frequency and amplitude values. For a significant plunge velocity, a significant increase in Nusselt number ( is observed compared to a stationary plate representing an increase in the extent of heat transferred. With more vigorous oscillations, the increase in becomes more prominent and similar trends and comparable magnitudes were observed for a constant value. Finally, the dependence of heat transfer augmentation on the frequency and amplitude of vibration is quantified with a simple parameterization for a plate-fin in a fluid medium.

Electric charges are often found in naturally or artificially formed droplets, such as raindrops and those generated by Kelvin's water dropper. In contrast to the impact of neutral droplets on a flat solid surface upon which a thin convex lens shape layer of the gas film is typically formed, I show that the delicate gas thin film can be fundamentally altered for even weakly charged droplets both experimentally and numerically. As the charge level is raised above a critical level of about 1% of the Rayleigh limit for representative impact conditions, the Maxwell stress overcomes the gas pressure buildup to deform the droplet bottom surface. A conical liquid tip forms and pierces Through the gas film, leading to a circular contact line moving outwards that does not trap any gas. The critical charge level only depends on the capillary number based on the gas viscosity. The deformation applies to common liquids and molten alloy droplets. Even dielectric surfaces can also induce conical deformation. The charged droplets can also deform upon hydrophobic surfaces, and increase the contact time on hydrophobic surfaces or even avoid bouncing. Doctor of Philosophy Electric charges are often found in naturally or artificially formed droplets, such as raindrops, waterfall, and inkjet printer. Neutral droplets impact on flat surfaces will usually trap a bubble inside because of the viscosity of air. The air bubble entrapped can be ignored if the droplet is water because the air bubble will eventually pinch-off. However, if the droplet is metal or some other viscous liquid, the air bubble will stay inside the liquid. This entrapped air bubble is undesired under some circumstances. For example, the existence of air bubble during metal 3D printing can influence the physical property. I show that the delicate gas thin film can be fundamentally altered for even weakly charged droplets both experimentally and numerically. As the charge level is raised above a critical level of about 1% of the maximum charges a droplet can carry for representative impact conditions, the electric stress will dominate the deformation of droplet. A conical liquid tip forms at the droplet bottom, avoiding the entrapment of air bubble. The critical charge level is experimentally proved to be only dependent on the gas viscosity and impact velocity. The deformation applies to common liquids and molten alloy droplets. Even dielectric surfaces can also induce conical deformation. The charged droplets can also deform upon hydrophobic surfaces, and increase the contact time on hydrophobic surfaces or even avoid bouncing.

Thermoacoustic instabilities arise and sustain due to the coupling of unsteady heat release from the flame and the acoustic field. One potential driving mechanism for these instabilities arise when velocity fluctuations (u') at the fuel injection location causes perturbations in the local equivalence ratio and is convected to the flame location generating an unsteady heat release (q') at a particular convection time delay, τ. Physically, τ is the time for the fuel to convect from injection to the flame. The n-τ Flame Transfer Function (FTF) is commonly used to model this relationship assuming an infinitesimally thin flame with a fixed τ. In practical systems, complex swirling flows, multiple fuel injections points, and recirculation zones create a distribution of τ, which can vary widely making a statistical description more representative. Furthermore, increased flame lengths and higher frequency instabilities with short acoustic wavelengths challenge the 'thin-flame' approximation. The present study outlines a methodology of using distributed convective fuel time delays and heat release rates in a one-dimensional (1-D) linear stability model based on the transfer matrix approach. CFD analyses, with the Flamelet Generated Manifold (FGM) combustion model are performed and probability density functions (PDFs) of the convective time delay and local heat release rates are extracted. These are then used as inputs to the 1-D Thermoacoustic model. Results are compared with the experimental results, and the proposed methodology improves the accuracy of stability predictions of 1-D Thermoacoustic modeling. Master of Science Gas turbines that operate with lean, premixed air-fuel mixtures are highly efficient and produce significantly lesser emission of pollutants. However, they are highly susceptible to self-induced thermoacoustic oscillations which can excite larger pressure fluctuation which can damage critical components or catastrophic engine failure. Such a combustion system is considered to be unstable since the oscillation amplitude increases with time. Understanding the non-linear feedback mechanisms driving the system unstable and their cause are naturally of high interest to the industry. Highly resolved, but computationally demanding simulations can predict the stability of the system accurately, but become bottlenecks delaying iterative design improvements. Low order numerical models counter this with quick solutions but use simplified representations of the flame and feedback mechanisms, resulting in unreliable stability predictions. The current study bridges the gap between these methods by modifying the numerical model, allowing it to incorporate a better representation of fluid flow fields and flame structures that are obtained through computationally cheaper simulations. Experiments are conducted to verify the predictions and a technique that can be used to identify regions of the flame that contribute to amplitude growth is introduced. The improved model shows notable improvement in its prediction capabilities compared to existing models.

Particulate systems in practical applications such as biomass combustion, blood cellular systems and granular particles in fluidized beds, have often been computationally represented using spherical surfaces, even though the majority of particles in archetypal fluid-solid systems are non-spherical. While spherical particles are more cost-effective to simulate, notable deficiencies of these implementations are their substantial inaccuracies in predicting the dynamics of particle mixtures. Alternatively, modeling dense fluid-particulate systems using non-spherical particles involves increased complexity, with computational cost manifesting as the biggest bottleneck. However, with recent advancements in computer hardware, simulations of three-dimensional particulate systems using irregular shaped particles have garnered significant interest. In this research, a novel Discrete Element Method (DEM) model that incorporates geometry definition, collision detection, and post-collision kinematics has been developed to accurately simulate non-spherical particulate systems. Superellipsoids, which account for 80% of particles commonly found in nature, are used to represent non-spherical shapes. Collisions between these particles are processed using a distance function computation carried out with respect to their surfaces. An event - driven model and a time-driven model have been employed in the current framework to resolve collisions. The collision model's influence on non–spherical particle dynamics is verified by observing the conservation of momentum and total kinetic energy. Furthermore, the non-spherical DEM model is coupled with an in-house fluid flow solver (GenIDLEST). The combined CFD-DEM model's results are validated by comparing to experimental measurements in a fluidized bed. The parallel scalability of the non-spherical DEM model is evaluated in terms of its efficiency and speedup. Major factors affecting wall clock time of simulations are analyzed and an estimate of the model's dependency on these factors is documented. The developed framework allows for a wide range of particle geometries to be simulated in GenIDLEST. Master of Science CFD – DEM (Discrete Element Method) is a technique of coupling fluid flow solvers with granular solid particles. CFD – DEM simulations are beneficial in recreating pragmatic applications such as blood cellular flows, fluidized beds and pharmaceutics. Up until recently, particles in these flows have been modeled as spheres as the generation of particle geometry and collision detection algorithms are straightforward. However, in real – life occurrences, most particles are irregular in shape, and approximating them as spheres in computational works leads to a substantial loss of accuracy. On the other hand, non – spherical particles are more complex to generate. When these particles are in motion, they collide and exhibit complex trajectories. Majority of the wall clock time is spent in resolving collisions between these non – spherical particles. Hence, generic algorithms to detect and resolve collisions have to be incorporated. This primary focus of this research work is to develop collision detection and resolution algorithms for non – spherical particles. Collisions are detected using inherent geometrical properties of the class of particles used. Two popular models (event-driven and time-driven) are implemented and utilized to update the trajectories of particles. These models are coupled with an in – house fluid solver (GenIDLEST) and the functioning of the DEM model is validated with experimental results from previous research works. Also, since the computational effort required is higher in the case of non – spherical particulate simulations, an estimate of the scalability of the problem and factors influencing time to simulations are presented.

Detonation cycles are identified as an efficient alternative to the Brayton cycles used in power and propulsion applications. Rotating Detonation Engine (RDE) operating on a detonation cycle works by compressing the working fluid across a detonation wave, thereby reducing the number of compressor stages required in the thermodynamic cycle. Numerical analyses of RDEs are flexible in understanding the flow field within the RDE, however, three-dimensional analyses are expensive due to the differences in time-scale required to resolve the combustion process and flow-field. The alternate two-dimensional analyses are generally modeled with perfectly premixed fuel injection and do not capture the effects of improper mixing arising due to discrete injection of fuel and oxidizer into the chamber. To model realistic injection in a 2-D analysis, the current work uses an approach in which, a Probability Density Function (PDF) of the fuel mass fraction at the chamber inlet is extracted from a 3-D, cold-flow simulation and is used as an inlet boundary condition for fuel mass fraction in the 2-D analysis. The 2-D simulation requires only 0.4% of the CPU hours for one revolution of the detonation compared to an equivalent 3-D simulation. Using this method, a perfectly premixed RDE is comparing with a non-premixed case. The performance is found to vary between the two cases. The mean detonation velocities, time-averaged static pressure profiles are found to be similar between the two cases, while the local detonation velocities and peak pressure values vary in the non-premixed case due to local pockets fuel rich/lean mixtures. The mean detonation cell sizes are similar, but the distribution in the non-premixed case is closer due to stronger shock structures. An analytical method is used to check the effects of fuel-product stratification and heat loss from the RDE and these effects adversely affect the local detonation velocity. Overall, this method of modeling captures the complex physics in an RDE with the advantage of reduced computational cost and therefore can be used for design and diagnostic purposes. Master of Science The conventional Brayton cycle used in power and propulsion applications is highly optimized, at cycle and component levels. In pursuit of higher thermodynamic efficiency, detonation cycles are identified as an efficient alternative and gained increased attention in the scientific community. In a Rotating Detonation Engine (RDE), which is based on the detonation cycle, the compression of gases occurs across a shock wave. This method of achieving high compression ratios reduces the number of compressor stages required for operation. In an RDE (where combustion occurs between two coaxial cylinders), the fuel and oxidizer are injected axially into the combustion chamber where the detonation is initiated. The resultant detonation wave spins continuously in the azimuthal direction, consuming fresh fuel mixture. The combustion products expand and exhaust axially providing thrust/mechanical energy when coupled with a turbine. Numerical analyses of RDEs are flexible over experimental analysis, in terms of understanding the flow physics and the physical/chemical processes occurring within the engine. However, three-dimensional numerical analyses are computationally expansive, and therefore demanding an equivalent, efficient two-dimensional analysis. In most RDEs, fuel and oxidizer are injected from separate plenums into the chamber. This type of injection leads to inhomogeneity of the fuel-air mixture within the RDE which adversely affects the performance of the engine. The current study uses a novel method to effectively capture these physics in a 2-D numerical analysis. Furthermore, the performance of the combustor is compared between perfectly premixed injection and discrete, non-premixed injection. The method used in this work can be used for any injector design and is a powerful/efficient way to numerically analyze a Rotating Detonation Engine.

According to the United Nations, 55% of the world's population currently lives in urban areas and which is projected to increase to 67% by 2050. Thus, it is imperative that effective strategies are developed to mitigate urban pollution. Complementing field experiments, computational fluid dynamics (CFD) analyses are becoming an effective strategy for identifying critical factors that influence urban pollution and its mitigation. This thesis focuses on two scales of the urban micro-climate environment: (i) evaluation of LES simulations with a simplified grid for modeling pollution dispersion in a street canyon and (ii) investigation of the effects of leaf surface micro-characteristics, wind speed, and particle sizes on the dry deposition of fine particulate matter (PM2.5). The first of these studies focuses on reproducing the pollution dispersion in a street canyon measured in a wind tunnel at Karlsruhe Institute of Technology (KIT), Germany. A simplified grid with the Large Eddy Simulations (LES) approach for canyon ratio W/H = 1 is proposed with the goal to reduce the computational cost by eliminating the need to model the entire canyon while striving to preserve the mixing induced by individual jets used to model vehicle emission in the experiment. LES is also capable of providing transient flow field and pollution concentration data not available with widely-used steady approaches such as RANS. The time-dependent information is crucial for pollution mitigation since pedestrians are usually exposed to pollution on a short-time basis. The predictions are in satisfactory agreement with the experiment for W/H = 1, yielding the Pearson correlation coefficient R = 0.81, with better performance near the leeward wall. Due to the small span modeled, three-dimensional instabilities fail to develop which could probably explain the overprediction of pollution concentration near ground level. However, other LES investigations where the full canyon was modeled also observed over-predictions. The use of a discrete emission source was not observed to provide benefits. The current model could be further improved by using a larger spanwise domain with a continuous line source to allow large wavelength instabilities to develop and increase turbulent diffusion. The second part of this thesis investigates the impact of trichome morphology and wind speed on the deposition of 0.3 μm and 1.0 μm particles on leaves. Using the one-way coupling approach to predict the fluid-particle interactions with the assumption that all particles that impact the leaf or trichome surface deposit, trichomes of 5 μm and 20 μm in diameter are modeled as equally spaced and uniform cylinders on an infinitely large plane. The results show that trichome diameter, density, and wind speed have a favorable impact on deposition velocity. Comparing to the smooth leaf, the presence of the thicker 20 μm hairs increases the deposition velocity by 1.5 – 4 times, whereas, the presence of short 5um trichomes reduces the deposition by 15 - 45%. Increasing trichome height from H/D = 20 to 30 shows benefits for the thinner trichomes but lowers the deposition for the densely packed thicker trichomes. Less aerosol deposition is also observed when the particle diameter increases from 0.3 μm to 1.0 μm. Due to the non-uniform contributions of these various traits, a non-dimensional ratio Rhp is proposed to model the aerosol deposition on leaf surface at wind speed of 1 m/s which yields a satisfactory linear correlation coefficient of 0.89 for 0 < R_hp < 0.3. Comparing to other published field and wind tunnel experiments conducted on a much larger scale, the deposition velocities predicted are at the lower end (U_dep^* = 0.002 to 0.012 cm/s) because of the idealized conditions. Nonetheless, the results still offer valuable insight into the effects of trichome morphology on pollutant deposition in isolation from other macro-factors. Master of Science According to the United Nations, 55% of the world’s population currently lives in urban areas and which is projected to increase to 67% by 2050. Thus, it is imperative that effective strategies are developed to mitigate urban pollution. Complementing field experiments, computational fluid dynamics (CFD) analyses are becoming an effective strategy for identifying critical factors that influence urban pollution and its mitigation. This thesis focuses on two scales of the urban micro-climate environment: (i) evaluation of Large Eddy Simulation (LES) with a simplified method for modeling pollution dispersion in a street canyon and (ii) investigation of the effects of leaf surface micro-characteristics, wind speed, and particle sizes on the dry deposition of fine particulate matter (PM2.5). The first of these studies focuses on reproducing the pollution dispersion in a street canyon measured in a wind tunnel at Karlsruhe Institute of Technology (KIT), Germany. A simplified grid with the LES approach for canyon ratio W/H = 1 is proposed. The goal of this study is to reduce the computational cost by modelling the canyon with a very thin span instead of the entire canyon while providing time-dependent information which is crucial for pollution mitigation since pedestrians are usually exposed to pollution on a short-time basis. The predictions are in satisfactory agreement with the experiment for W/H = 1 with better performance near the leeward wall (i.e. the left wall) and overprediction of pollution concentration near ground level – as observed by other LES investigations. The current model could be further improved by using a larger spanwise domain with a continuous line source to allow instabilities to develop, thus improve prediction accuracy. The second part of this thesis investigates the impact of trichome (i.e. a hair or an outgrowth from leaf surface) morphology and wind speed on the deposition of 0.3 mm and 1.0 mm particles on leaves. The results show that trichome diameter, density, and wind speed have a favorable impact on deposition velocity. Less aerosol deposition is also observed when the particle diameter increases from 0.3 mm to 1.0 mm. No clear effects is observed by altering the trichome height. Due to the non-uniform contributions of these various traits, a non-dimensional ratio D∗ �D∗ �2 Rhp = hair hair is proposed to model the aerosol deposition on leaf surface at wind speed of D∗ H∗ S∗ p hair hair 1 m/s which yields a satisfactory linear correlation coefficient of 0.89 for 0 < Rhp < 0.3. This ratio includes trichome diameter (D∗ ), height (H∗ ), spacing (S∗ ) as well as the ratio of hair hair hair trichome diameter to particle diameter (D∗ /D∗ ). The results offer valuable insight into the hair p effects of trichome morphology on pollutant deposition in isolation from other macro-factors.

The present thesis focuses on evaluating a conjugate heat transfer (CHT) simulation in a ribbed cooling passage with a fully developed flow assumption using LES with the immersed boundary method (IBM-LES-CHT). The IBM with the LES model (IBM-LES) and the IBM with CHT boundary condition (IBM-CHT) frameworks are validated prior to the main simulations by simulating purely convective heat transfer (iso-flux) in the ribbed duct, and a developing laminar boundary layer flow over a two-dimensional flat plate with heat conduction, respectively. For the main conjugate simulations, a ribbed duct geometry with a blockage ratio of 0.3 is simulated at a bulk Reynolds number of 10,000 with a conjugate boundary condition applied to the rib surface. The nominal Biot number is kept at 1, which is similar to the comparative experiment. As a means to overcome a large time scale disparity between the fluid and the solid regions, the use of a high artificial solid thermal diffusivity is compared to the physical diffusivity. It is shown that while the diffusivity impacts the instantaneous fluctuations in temperature, heat transfer and Nusselt numbers, it has an insignificantly small effect on the mean Nusselt number. The comparison between the IBM-LES-CHT and iso-flux simulations shows that the iso-flux case predicts higher local Nusselt numbers at the back face of the rib. Furthermore, the local Nusselt number augmentation ratio (EF) predicted by IBM-LES-CHT is compared to the body fitted grid (BFG) simulation, experiment and another LES conjugate simulation. Even though there is a mismatch between IBM-LES-CHT prediction and other studies at the front face of the rib, the area-averaged EF compares reasonably well in other regions between IBM-LES-CHT prediction and the comparative studies. Master of Science The present thesis focuses on the computational study of the conjugate heat transfer (CHT) investigation on the turbine internal ribbed cooling channel. Plenty of prior research on turbine internal cooling channel have been conducted by considering only the convective heat transfer at the wall, which assumes an iso-flux (constant heat flux) boundary condition at the surface. However, applying an iso-flux condition on the surface is far from the realistic heat transfer mechanism occurring in internal cooling systems. In this work, a conjugate heat transfer analysis of the cooling channel, which considers both the conduction within the solid wall and the convection at the ribbed inner wall surface, is conducted for more realistic heat transfer coefficient prediction at the inner ribbed wall. For the simulation, the computational mesh is generated by the immersed boundary method (IBM), which can ease the mesh generation by simply immersing the CAD geometry into the background volume grid. The IBM is combined with the conjugate boundary condition to simulate the internal ribbed cooling channel. The conjugate simulation is compared with the experimental data and another computational study for the validation. Even though there are some discrepancy between the IBM simulation and other comparative studies, overall results are in good agreement. From the thermal prediction comparison between the iso-flux case and the conjugate case v using the IBM, it is found that the heat transfer predicted by the conjugate case is different from the iso-flux case by more than 40 percent at the rib back face. The present study shows the potential of the IBM framework with the conjugate boundary condition for more complicated geometry, such as full turbine blade model with external and internal cooling system.

Exploring unconventional hydrocarbon reservoirs and enhancing the recovery of hydrocarbon from conventional reservoirs are necessary for meeting the society's ever-increasing energy demand and requires a thorough understanding of the multiphase interfacial and transport phenomena in these reservoirs. This dissertation performs pore-scale studies of interfacial thermodynamics and multiphase hydrodynamics in shale reservoirs and conventional oil-brine-rock (OBR) systems. In shale gas reservoirs, the imbibition of water through surface hydration into gas-filled mica pores was found to follow the diffusive scaling law, but with an effective diffusivity much larger than the self-diffusivity of water molecules. The invasion of gas into water-filled pores with width down to 2nm occurs at a critical invasion pressure similar to that predicted by the classical capillary theories if effects of disjoining pressure and diffusiveness of water-gas interfaces are considered. The invasion of oil droplets into water-filled pores can face a free energy barrier if the pressure difference along pore is small. The computed free energy profiles are quantitatively captured by continuum theories if capillary and disjoining pressure effects are considered. Small droplets can invade a pore through thermal activation even if an energy barrier exists for its invasion. In conventional oil reservoirs, low-salinity waterflooding is an enhanced oil recovery method that relies on the modification of thin brine films in OBR systems by salinity change. A systematic study of the structure, disjoining pressure, and dynamic properties of these thin brine films was performed. As brine films are squeezed down to sub-nanometer scale, the structure of water-rock and water-oil interfaces changes marginally, but that of the electrical double layers in the films changes greatly. The disjoining pressure in the film and its response to salinity change follow the trend predicted by the DLVO theory, although the hydration and double layer forces are not simple additive as commonly assumed. A notable slip between the brine film and the oil phase can occur. The role of thin liquid films in multiphase transport in hydrocarbon reservoirs revealed here helps lay foundation for manipulating and leveraging these films to enhance hydrocarbon production and to minimize environmental damage during such extraction. Doctor of Philosophy Meeting the ever-increasing energy demand requires efficient extraction of hydrocarbons from unconventional reservoirs and enhanced recovery from conventional reservoirs, which necessitate a thorough understanding of the interfacial and transport phenomena involved in the extraction process. Abundant water is found in both conventional oil reservoirs and emerging hydrocarbon reservoirs such as shales. The interfacial behavior and transport of water and hydrocarbon in these systems can largely affect the oil and gas recovery process, but are not well understood, especially at pore scale. To fill in the knowledge gap on these important problems, this dissertation focuses on the pore-scale multiphase interfacial and transport phenomena in hydrocarbon reservoirs. In shales, water is found to imbibe into strongly hydrophilic nanopores even though the pore is filled with highly pressurized methane. Methane gas can invade into water-filled nanopores if its pressure exceeds a threshold value, and the thin residual water films on the pore walls significantly affect the threshold pressure. Oil droplet can invade pores narrower than their diameter, and the energy cost for their invasion can only be computed accurately if the surface forces in the thin film formed between the droplet and pore surface are considered. In conventional reservoirs, thin brine films between oil droplet and rock greatly affect the wettability of oil droplets on the rock surface and thus the effectiveness of low-salinity waterflooding. In brine films with sub-nanometer thickness, the ion distribution differs from that near isolated rock surfaces but the structure of water-brine/rock interfaces is similar to their unconfined counterparts. The disjoining pressure in thin brine films and its response to the salinity change follow the trend predicted by classical theories, but new features are also found. A notable slip between the brine film and the oil phase can occur, which can facilitate the recovery of oil from reservoirs.

Effective locomotion and maneuvering in aquatic environments is important for survival for marine fauna. The ability to move quickly, change direction, and tune energy consumption for long migrations is critical for escape from predators and pursuit of prey. This controlled propulsion in terms of varying speed, turning rates, and actuation effort is of interest for the next generation of underwater vehicle design. Integration of biological functional simplicity, robustness, and superior performance enables robotic vehicles to successfully complete difficult and dynamic operational goals. Gelatinous animals known as Cnidarians employ a wide variety of propulsive methods, ranging from the simple but efficient propulsion of large jellyfish to the rapid and highly maneuverable multi-jet propulsion of colonial animals known as siphonophores. This dissertation studies how these two extremes of underwater soft body propulsion are able to achieve simple yet effective control and locomotion, and thus inform the design of effective vehicle propulsion control and actuation. Two large single bell jellyfish robots, Cyro 2 and Cyro 3, were designed and constructed to implement the simple body form and propulsive methods of large jellyfish to study the unique locomotive characteristics and fluid interactions that generate straight swimming and turning maneuvers. The other extreme of small soft-body colonies moving by multi-jet propulsion was subsequently investigated in-depth, starting with a characterization of the biological fluid jetting actions and gaits. The results of these performance capabilities were then applied to an experimental robotic model with bio-inspired construction and controls to verify an elegant but highly functional neurological control scheme and the kinematic capabilities from varying jetting gait patterns. PHD The ability to move rapidly in any direction is a primary characteristic of successful animal species. Evasion of predators, as well as pursuit of prey, is paramount for survival. Jellyfish are excellent examples of animals that have thrived for millions of years with varied methods of moving in their diverse environments. However, the propulsion methods of large jellyfish in straight swimming and turning have not been well understood until recent years. This dissertation focuses on the fundamental understanding of the locomotion and fluid interaction that jellyfish use for propulsion. A large jellyfish robot, named Cyro 2 (“Cy” for the species Cyanea, “ro” for robot, and the second generation of the design), was constructed to explore the role of various structural and fluidic parameters on the locomotion characteristics of the largest jellyfish species Cyanea. The successor Cyro 3 was designed to mimic the complex motions of large jellyfish during maneuvering. Motion tracking and fluid analysis of the robot during turning was utilized to explain how jellyfish dynamically control their orientation. These results inspired further study of a unique relative of jellyfish, the siphonophore, which can swim with a modular chain of soft pumping bodies that coordinate without a central nervous system. This unique control strategy and method of movement underwater was studied by analyzing specimens of the siphonophore Nanomia. Development and modeling of elegant control techniques inspired by this species is presented and implemented on an experimental model that uses this unique propulsion method to validate and expand upon observations of live specimens. Combined, the results obtained in this dissertation open the possibility of designing advanced underwater vehicles.

Fire dynamics is a complex process involving multi-mode heat transfer, reacting fluid flow, and the reaction of combustible materials. High-fidelity predictions of fire behavior using computational fluid dynamics (CFD) models come at a significant computational cost where simulation times are often measured in hours, days, or even weeks. A new simulation method is to use a machine learning approach which uses artificial neural networks (ANNs) to represent underlying connections between data to make predictions of new inputs. The field of image analysis has seen significant advancements in ANN performance by using feature based layers in the network architecture. Inspired by these advancements, a generalized procedure to design ANNs to make spatially resolved predictions in multi-physics applications is presented and applied to different fire applications. A deep convolutional inverse graphics network (DCIGN) was developed to predict the two-dimensional spatially resolved spread of a wildland fire. The network uses an image stack corresponding to the spatially resolved landscape, weather, and current fire perimeter (which can be obtained from measurements) to predict the fire perimeter six hours in the future. A transpose convolutional neural network (TCNN) was developed to predict the spatially resolved thermal flow field in a compartment fire from coarse zone fire model predictions. The network uses thirty-five parameters describing the geometry of the room and the ventilation conditions to predict the full-field temperature and velocity throughout the room. The data for use in training and testing both networks was generated using high-fidelity CFD fire simulations. Overall, the ANN predictions in each network agree with simulation predictions for validation scenarios. The computational time to evaluate the ANNs is 10,000x faster than the high-fidelity fire simulations. This work represents a first step in developing super real-time full-field fire predictions for different applications. Ph. D. The National Fire Protection Agency estimates the total cost of fire in the United States at $300 billion annually. In 2017 alone, there were 3,400 civilian fire fatalities, 14,670 civilian fire injuries, and an estimated $23 billion direct property loss in the United States. Large scale fires in the wildland urban interface (WUI) and in large buildings still represent a significant hazard to life, property, and the environment. Researchers and fire safety engineers often use computer simulations to predict the behavior of a fire to assist in reducing the hazard of fire. Unfortunately, typical simulations of fire scenarios may take hours, days, or even weeks to run which limits their use to small areas or sections of buildings. A new method is to use a machine learning approach which uses artificial neural networks (ANNs) to represent underlying connections between data to make new predictions of fire behavior. Inspired by advancements in the field of image processing, this research developed a procedure to use machine learning to make rapid high resolution predictions of fire behavior. An ANN was developed to predict the perimeter of a wildland fire six hours in the future based on a set of images corresponding to the landscape, weather, and current fire perimeter, all of which can be obtained directly from measurements (US Geological Survey, Automated Surface Observation System, and satellites). In addition, an ANN was developed to predict high-resolution temperature and velocity fields within a floor of a building based on predictions from a coarse model. The data for use in training and testing these networks was generated using high-resolution fire simulations. Overall, the network predictions agree well with simulation predictions for new scenarios. In addition, the time to run the model is 10,000x faster than the typical simulations. The work presented herein represents a first step in developing high resolution computer simulations for different fire scenarios that run very quickly.

An accurate measurement of bubbly flow has a significant value for understanding the bubble behavior, heat and energy transfer pattern in different engineering systems. It also helps to advance the theoretical model development in two-phase flow study. Due to the interaction between the gas and liquid phase, the flow patterns are complicated in recorded image data. The segmentation and reconstruction of overlapping bubbles in these images is a challenging task. This dissertation provides a complete set of image processing algorithms for bubbly flow measurement. The developed algorithm can deal with bubble overlapping issues and reconstruct bubble outline in 2D high speed images under a wide void fraction range. Key bubbly flow parameters such as void fraction, interfacial area concentration, bubble number density and velocity can be computed automatically after bubble segmentation. The time-averaged bubbly flow distributions are generated based on the extracted parameters for flow characteristic study. A 3D imaging system is developed for 3D bubble reconstruction. The proposed 3D reconstruction algorithm can restore the bubble shape in a time sequence for accurate flow visualization with minimum assumptions. The 3D reconstruction algorithm shows an error of less than 2% in volume measurement compared to the syringe reading. Finally, a new image synthesis framework called Bubble Generative Adversarial Networks (BubGAN) is proposed by combining the conventional image processing algorithm and deep learning technique. This framework aims to provide a generic benchmark tool for assessing the performance of the existed image processing algorithms with significant quality improvement in synthetic bubbly flow image generation. Ph. D. Bubbly flow phenomenon exists in a wide variety of systems, for example, nuclear reactor, heat exchanger, chemical bubble column and biological system. The accurate measurement of the bubble distribution can be helpful to understand the behaviors of these systems. Due to the complexity of the bubbly flow images, it is not practical to manually process and label these data for analysis. This dissertation developed a complete suite of image processing algorithms to process bubbly flow images. The proposed algorithms have the capability of segmenting 2D dense bubble images and reconstructing 3D bubble shape in coordinate with multiple camera systems. The bubbly flow patterns and characteristics are analyzed in this dissertation. Finally, a generic image processing benchmark tool called Bubble Generative Adversarial Networks (BubGAN) is proposed by combining the conventional image processing and deep learning techniques together. The BubGAN framework aims to bridge the gap between real bubbly images and synthetic images used for algorithm benchmark and algorithm.

Fuel burnup calculations provide material concentrations and intrinsic neutron and gamma source strengths as a function of irradiation and cooling time. Detailed, full-core 3D burnup calculations are critical for nuclear fuel management studies, including core design and spent fuel storage safety and safeguards analysis. For core design, specifically during refueling, full- core pin-wise, axially-dependent burnup distributions are necessary to determine assembly positioning to efficiently utilize fuel resources. In spent fuel storage criticality safety analysis, detailed burnup distributions enable best-estimate analysis which allows for more effective utilization of storage space. Additionally, detailed knowledge of neutron and gamma source distributions provide the ability to ensure nuclear material safeguards. The need for accurate and efficient burnup calculations has become more urgent for the simulation of advanced reactors and monitoring and safeguards of spent fuel pools. To this end, the Virginia Tech Transport Theory Group (VT3G) has been working on advanced computational tools for accurate modeling and simulation of nuclear systems in real-time. These tools are based on the Multi-stage Response-function Transport (MRT) methodology. For monitoring and safety evaluation of spent fuel pools and casks, the RAPID (Real-time Analysis for Particle transport and In-situ Detection) code system has been developed. This dissertation presents a novel methodology and algorithm for performing 3D fuel bur- nup calculations, referred to as bRAPID- Burnup with RAPID . bRAPID utilizes the existing RAPID code system for accurate calculation of 3D fission source distributions as the trans- port calculation tool to drive the 3D burnup calculation. bRAPID is capable of accurately and efficiently calculating assembly-wise axially-dependent fission source and burnup dis- tributions, and irradiated-fuel properties including material compositions, neutron source, gamma source, spontaneous fission source, and activities. bRAPID performs 3D burnup calculations in a fraction of the time required by state-of-the-art methodologies because it utilizes a pre-calculated database of response functions. The bRAPID database pre-calculation procedure, and its automation, is presented. The ex- isting RAPID code is then benchmarked against the MCNP and Serpent Monte Carlo codes for a spent fuel pool and the U.S. Naval Academy Subcritical Reactor facility. RAPID is shown to accurately calculate eigenvalue, subcritical multiplication, and 3D fission source dis- tributions. Finally, bRAPID is compared to traditional, state-of-the art Serpent Monte Carlo burnup calculations and its performance will be evaluated. It is important to note that the automated pre-calculation proceedure is required for evaluating the performance of bRAPID. Additionally, benchmarking of the RAPID code is necessary to understand RAPID's ability to solve problems with variable burnups distributions and to asses its accuracy. Ph. D. In a nuclear reactor, the energy released from a fission reaction, the splitting of an atomic nucleus into smaller parts, is harnessed to generate electricity. Nuclear reactors rely on fuel, typically comprised of uranium oxide (UO₂). While the reactor is operating and the fuel is being used, or “burned”, for power production it is undergoing numerous nuclear reactions, including fission, and radioactive decays which alter the material composition. Knowing the time evolution of fuel as it is burned in the reactor, i.e., concentration of isotopes and sources of radiation, is critical. Nuclear reactor designers and operators use this information to optimize power production and perform safety analysis of used nuclear fuel. By performing fuel burnup calculations, material concentrations and radiation source strengths are obtained as a function of time in an operating nuclear reactor. Using traditional computational techniques, these calculations are extremely time consuming and, for certain problems, can be difficult to obtain an accurate solution. Ideally, a reactor designer would like to know the three-dimensional (3D) distribution of material compositions and sources; however this level of detail would require an excessive amount of calculation time, therefore simplified models and assumptions are used. For the design of the new generation of nuclear reactors, and monitoring and safeguards analysis, this level of detail will be required in lieu of the availability of experimental facilities which do not currently exist. This dissertation presents a novel methodology and algorithm for performing accurate 3D fuel burnup calculations in real-time, referred to as bRAPID (Burnup with RAPID). bRAPID utilizes an existing nuclear software, RAPID (Real-time Analysis for Particle transport and In-situ Detection), developed in the Virginia Tech Transport Theory Group (VT3G), which has been shown to accurately solve time-independent nuclear calculations in significantly less time than traditional approaches. bRAPID is capable of accurately calculating 3D material and source distributions as a function of time in an operating nuclear reactor, and requires significantly less time and computational resources than traditional approaches. To ensure that bRAPID is relatively easy to use, a number of automated routines have been developed and are presented. RAPID is benchmarked against the traditional code systems MCNP (Monte Carlo N-Particle) and Serpent, both of which are widely used in the nuclear community, for a spent fuel storage pool and the U.S. Naval Academy subcritical nuclear reactor facility. RAPID is shown to accurately calculate system parameters (eigenvalue and subcritical multiplication factor) and 3D fission source distributions. Finally, bRAPID is compared to the traditional burnup approach, using the Serpent code system. bRAPID is shown to accurately calculate system parameters and 3D material and source distributions in significantly less time than the traditional approach.

This thesis presents a computational framework that models the coupled behavior between sand deposits and their surrounding fluid. Particle deposits that form in gas turbine engines and industrial burners, can change flow dynamics and heat transfer, leading to performance degradation and impacting durability. The proposed coupled framework allows insight into the coupled behavior of sand deposits at high temperatures with the flow, which has not been available previously. The coupling is done by using a CFD-DEM framework in which a physics based collision model is used to predict the post-collision state-of-the-sand-particle. The collision model is sensitive to temperature dependent material properties of sand. Particle deposition is determined by the particle's softening temperature and the calculated coefficient of restitution of the collision. The multiphase treatment facilitates conduction through the porous deposit and the coupling between the deposit and the fluid field. The coupled framework was first used to model the behavior of softened sand particles in a laminar impinging jet flow field. The temperature of the jet and the impact surface were varied(T^* = 1000 – 1600 K), to observe particle behavior under different temperature conditions. The Reynolds number(Rejet = 20, 75, 100) and particle Stokes numbers (Stp = 0.53, 0.85, 2.66, 3.19) were also varied to observe any effects the particles' responsiveness had on deposition and the flow field. The coupled framework was found to increase or decrease capture efficiency, when compared to an uncoupled simulation, by as much as 10% depending on the temperature field. Deposits that formed on the impact surface, using the coupled framework, altered the velocity field by as much as 130% but had a limited effect on the temperature field. Simulations were also done that looked at the formation of an equilibrium deposit when a cold jet impinged on a relatively hotter surface, under continuous particle injection. An equilibrium deposit was found to form as deposited particles created a heat barrier on the high temperature surface, limiting more particle deposition. However, due to the transient nature of the system, the deposit temperature increased once deposition was halted. Further particle injection was not performed, but it can be predicted that the formed deposit would begin to grow again. Additionally, a Large-Eddy Simulation (LES) simulation, with the inclusion of the Smagorinsky subgrid model, was performed to observe particle deposition in a turbulent flow field. Deposition of sand particles was observed as a turbulent jet (Re jet=23000,T_jet^*= 1200 K) impinged on a hotter surface(T_surf^*= 1600 K). Differences between the simulated flow field and relevant experiments were attributed to differing jet exit conditions and impact surface thermal conditions. The deposit was not substantive enough to have a significant effect on the flow field. With no difference in the flow field, no difference was found in the capture efficiency between the coupled and decoupled frameworks. Master of Science

Computational Fluid Dynamics Coupled to Discrete Element Method (CFD-DEM) is widely used in simulating a large variety of particulate flow system. This Eulerian-Lagrangian technique tracks all the particles included in the system by the application of point mass models in their equation of motion. CFD-DEM is a more accurate (and more expensive) technique compared to an Eulerian-Eulerian representation. Compared to Particle Resolved Simulations (PRS), CFD-DEM is less expensive since it does not require resolving the flow around each particles and thus can be applied to larger scale systems. Nevertheless, simulating industrial and natural scale systems is a challenge for this numerical technique. This is because the cost of CFD-DEM is proportional to the number of particles in the system under consideration. Thus, massively parallel codes are used to tackle these problems with the help of supercomputers. In this thesis, the CFD-DEM capability in the in-house code Generalized Incompressible Direct and Large Eddy Simulation of Turbulence (GenIDLEST) is used to investigate large scale dense particulate flow systems. Central to the contributions made by this work are developments to reduce the computational cost of CFD-DEM. This includes the development and validation of reduced order history force model for use in large scale systems and validation of the representative particle model, which lumps multiple particles into one, thus reducing the number of particles that need to be tracked in the system. Numerical difficulties in the form of long integration times and instabilities encountered in fully coupling the fluid and particle phases in highly energetic systems are alleviated by proposing a partial coupling scheme which maintains the accuracy of full-coupling to a large extent but at a reduced computational cost. The proposed partial-coupling is found to have a better convergence behavior compared to the full coupling in large systems and can be used in cases where full coupling is not feasible or impractical to use. Alternative modeling approaches for the tangential treatment of the soft-sphere impact model to avoid storing individual impact deformation are proposed and tested. A time advancement technique is developed and proposed for use in dense particulate systems with a hard-sphere impact model. The new advancement technique allows for the use of larger time steps which can speed-up the time to solution by as much as an order of magnitude. PHD

Bats are some of the most agile flyers in nature. Their wings are highly articulated which affords them very fine control over shape and form. This thesis investigates the flight of Hipposideros Pratti. The flight pattern studied is nominally level and straight. Measured wing kinematics are used to describe the wing motion. It is shown that Proper Orthogonal Decomposition (POD) can be used to effectively to filter the measured kinematics to eliminate outliers which usually manifest as low energy higher POD modes, but which can impact the stability of aerodynamic simulations. Through aerodynamic simulations it is established that the first two modes from the POD analysis recover 62% of the lift, and reflect a drag force instead of thrust, whereas the first three modes recover 77% of the thrust and even more lift than the native kinematics. This demonstrates that mode 2, which features a combination of spanwise twisting (pitching) and chordwise cambering, is critical for the generation of lift, and more so for thrust. Based on these inferences, it is concluded that the first 7 modes are sufficient to represent the full native kinematics. The aerodynamic simulations are conducted using the immersed boundary method on 128 processors. They utilize a grid of 31 million cells and the bat wing is represented by about 50000 surface elements. The movement of the immersed wing surface is defined by piecewise cubic splines that describe the time evolution of each control point on the wing. The major contribution of this work is the decomposition of the native kinematics into canonical flapping wing physical descriptors comprising of the flapping motion, stroke-plane deviation, pitching motion, chordwise, and spanwise cambering. It is shown that the pitching mode harvests a Leading Edge Vortex (LEV) during the upstroke to produce thrust. It also stabilizes the LEV during downstroke, as a result, larger lift and thrust production is observed. Chordwise cambering mode allows the LEV to glide over and cover a large portion of the wing thus contributing to more lift while the spanwise cambering mode mitigates the intensification of LEV during the upstroke by relative rotation of outer part of the wing ( hand wing ) with respect to the inner part of the wing ( arm wing). While this thesis concerns itself with near straight-level flight, the proposed decomposition can be applied to any complex flight maneuver and provide a basis for unified comparison not only over different bat flight regimes but also across other flying insects and birds. MS

Gas-solid flow is fundamental to many industrial processes. Extensive experimental and numerical studies have been devoted to understand the interphase momentum and heat transfer in these systems. Most of the studies have focused on spherical particle shapes, however, in most natural and industrial processes, the particle shape is seldom spherical. In fact, particle shape is one of the important parameters that can have a significant impact on momentum, heat and mass transfer, which are fundamental to all processes. In this study particle-resolved simulations are performed to study momentum and heat transfer in flow through a fixed random assembly of ellipsoidal particles with sphericity of 0.887. The incompressible Navier-Stokes equations are solved using the Immersed Boundary Method (IBM). A Framework for generating particle assembly is developed using physics engine PhysX. High-order boundary conditions are developed for immersed boundary method to resolve the heat transfer in the vicinity of fluid/particle boundary with better accuracy. A complete framework using particle-resolved simulation study assembly of particles with any shape is developed. The drag force of spherical particles and ellipsoid particles are investigated. Available correlations are evaluated based on simulation results and recommendations are made regarding the best combinations. The heat transfer in assembly of ellipsoidal particle is investigated, and a correlation is proposed for the particle shape studied. The lift force, lateral force and torque of ellipsoid particles in assembly and their variations are quantitatively presented and it is shown that under certain conditions these forces and torques cannot be neglected as is done in the larger literature. Ph. D.

As worldwide energy consumption continues to increase, so does the demand for renewable energy sources. The total available wave energy resource for the United States alone is 2,640 TWh/yr; nearly two thirds of the 4,000 TWh of electricity used in the United States each year. It is estimated that nearly half of that available energy is recoverable through wave energy conversion techniques. In this thesis, a two-body 'point absorber' type wave energy converter with a mechanical power-takeoff is investigated. The two-body wave energy converter extracts energy through the relative motion of a floating buoy and a neutrally buoyant submerged body. Using a linear frequency-domain model, analytical solutions of the optimal power and the corresponding power-takeoff components are derived for the two-body wave energy converter. Using these solutions, a case study is conducted to investigate the influence of the submerged body size on the absorbed power of the device in regular and irregular waves. Here it is found that an optimal mass ratio between the submerged body and floating buoy exists where the device will achieve resonance. Furthermore, a case study to investigate the influence of the submerged body shape on the absorbed power is conducted using a time-domain numerical model. Here it is found that the submerged body should be designed to reduce the effects of drag, but to maintain relatively large hydrodynamic added mass and excitation force. To validate the analytical and numerical models, a 1/30th scale model of a two-body wave energy converter is tested in a wave tank. The results of the wave tank tests show that the two-body wave energy converter can absorb nearly twice the energy of a single-body 'point absorber' type wave energy converter. Master of Science

According to the World Health Organization (WHO), Cardiovascular Disease (CVD) is the leading cause of death in the world. Biomechanics and fluid dynamics of blood flow play an important role in CVD mediation. Shear stress plays a major role in platelet-substrate interactions and thrombus formation and growth in blood flow, where under both pathological and physiological conditions platelet adhesion and accumulation occur. In this study, a three-dimensional dynamic model of platelet-rich thrombus growth in stenosed vessels using computational fluid dynamics (CFD) methods is introduced. Platelet adhesion, aggregation and activation kinetics are modeled by solving mass transport equations for blood components involved in thrombosis. The model was first verified under three different shear conditions and at two heparin levels. Three-dimensional simulations were then carried out to evaluate the performance of the model for severely damaged (stripped) aortas with mild and severe stenosis degrees. For these cases, linear shear-dependent functions were developed for platelet-surface and platelet-platelet adhesion rates. It was confirmed that the platelet adhesion rate is not only a function of Reynolds number (or wall shear rate) but also the stenosis severity of the vessel. General correlations for adhesion rates of platelets as functions of stenosis and Reynolds number were obtained based on these cases. The model was applied to different experimental systems and shown to agree well with measured platelet deposition. Then, the Arbitrary Lagrangian Eulerian (ALE) formulation was used to model dynamic growth by including geometry change in the simulation procedure. The wall boundaries were discretely moved based on the amount of platelet deposition that occurs on the vessel wall. To emulate the dynamic behavior of platelet adhesion kinetics during thrombus growth, the validated model for platelet adhesion, which calculates platelet-surface adhesion rates as a function of stenosis severity and Reynolds number, was applied to the model. The model successfully predicts the nonlinear growth of thrombi in the stenosed area. These simulations provide a useful guide to understand the effect of growing thrombus on platelet deposition rate, platelet activation kinetics and occurrence of thromboembolism (TE) in highly stenosed arteries. Ph. D.

There are still many open questions regarding spatiotemporal chaos although many well developed theories exist for chaos in time. Rayleigh-B'enard convection is a paradigmatic example of spatiotemporal chaos that is also experimentally accessible. Discoveries uncovered using numerics can often be compared with experiments which can provide new physical insights. Lyapunov diagnostics can provide important information about the dynamics of small perturbations for chaotic systems. Covariant Lyapunov vectors reveal the true direction of perturbation growth and decay. The degree of hyperbolicity can also be quantified by the covariant Lyapunov vectors. To know whether a dynamical system is hyperbolic is important for the development of a theoretical understanding. In this thesis, the degree of hyperbolicity is calculated for chaotic Rayleigh-B'enard convection. For the values of the Rayleigh number explored, it is shown that the dynamics are non-hyperbolic. The spatial distribution of the covariant Lyapunov vectors is different for the different Lyapunov vectors. Localization is used to quantify this variation. The spatial localization of the covariant Lyapunov vectors has a decreasing trend as the order of the Lyapunov vector increases. The spatial localization of the covariant Lyapunov vectors are found to be related to the instantaneous Lyapunov exponents. The correlation is stronger as the order of the Lyapunov vector decreases. The covariant Lyapunov vectors are also computed using a spectral element approach. This allows an exploration of the covariant Lyapunov vectors in larger domains and for experimental conditions. The finite conductivity and finite thickness of the lateral boundaries of an experimental convection domain is also studied. Results are presented for the variation of the Nusselt number and fractal dimension for different boundary conditions. The fractal dimension changes dramatically with the variation of the finite conductivity. Ph. D.

Systemic chemotherapy is a major therapeutic approach for nearly all types and stages of cancer. Success of this treatment depends not only on the efficacy of the therapeutics but also on the transport of the drug to all tumor cells in sufficient concentrations. Intratumoral drug transport is limited by characteristics of the tumor microenvironment such as elevated interstitial pressure and sparse, irregular vascularization. Moreover, poor tumor selectivity, leads to systemic toxicity. Bacteria possess a host of characteristics that address the aforementioned challenges in conventional drug delivery approaches including tumor selectivity, preferential tumor colonization, effective tumor penetration, which can be augmented via genetic engineering. However, in clinical trials conducted to date, bacteria have rarely been able to inhibit tumor growth solely by their presence in the tumor. The overall goal of this doctoral dissertation is to develop a novel tumor treatment system based on Salmonella Typhimurium VNP20009 (genetically modified for preferential tumor colonization and attenuation) coupled with biodegradable copolymer, poly(lactic-co-glycolic acid) nanoparticles, hereafter referred to as NanoBEADS (Nanoscale Bacteria Enabled Autonomous Drug Delivery System). To this end, a NanoBEADS fabrication procedure that is robust and repeatable was established and a microfluidic chemotaxis-based sorting platform for the separation NanoBEADS from unattached nanoparticles was developed. The transport efficacy of NanoBEADS compared to the commonly used passively-diffusing nanoparticle was investigated in vitro and in vivo and the intratumoral penetration of the therapeutic vectors was quantified using a custom image processing algorithm. The mechanism of intratumoral penetration was elucidated through 2D and 3D invasion assays. Lastly, we developed a biophysical model of intratumoral transport of NanoBEADS based on the intratumoral penetration experimental results towards the theoretical evaluation of the drug transport profile following the administration of NanoBEADS. PHD

Higher turbine inlet temperatures (TIT) are required for higher overall efficiency of gas turbine engines. Due to the constant push towards achieving high TIT, the heat load on high pressure turbine components has been increasing with time. Gas turbine airfoils are equipped with several sophisticated cooling technologies which protect them from harsh external environment and increase their operating life and reduce the maintenance cost. The turbine airfoils are coated with thermal barrier coatings (TBCs) and the external surface is protected by film cooling. The internals of gas turbine blades are cooled by relatively colder air bled off from the compressor discharge. Gas turbine internals can be divided into three broad segments – Leading edge section, (2) mid-chord section and (3) trailing edge section. The leading edge of the airfoil is subjected to extreme heat loads due to hot main gas stagnation and high turbulence intensity of the combustor exit gases. The leading edge is typically cooled by jet impingement which cross-over the rib turbulators in the feed chamber. The mid-chord section of the turbine airfoils have serpentine passages connected via. 180° bends, and they feature turbulence promotors which enhance the heat exchange rates between the coolant and the internal walls of the airfoil. The trailing edge section is typically cooled by array of pin fins. On one hand, the coolant routed through the internal passages of turbine airfoil help maintain the airfoil temperatures within safe limits of operation, the cooled air comes at a cost of loss of high pressure air from the compressor section. The aim of this study is to develop internal cooling concepts which have high thermal hydraulic performance, i.e. to gain high levels heat transfer enhancement due to cooling concepts at lower pumping power requirements. Experimental and numerical studies have been carried out and new rib turbulator designs such as Criss-Cross pattern, compound channels featuring uniquely organized ribs and dimples, novel jet impingement hole shapes have been developed which have high thermal-hydraulic performance. Further, gas turbine blades rotate at high rotational speeds. The internal flow routed thought the serpentine passages are subjected to Coriolis and centrifugal buoyancy forces. The combined effects of these forces results in enhancement and reduction in heat transfer on the pressure side and suction side internal walls. This leads to non-uniformity in the heat transfer enhancement which leads to non-uniform cooling and increase in the sites of high and low internal wall temperatures. Development of cooling concepts which have high thermal hydraulic performance under non-rotating conditions is important, however, under rotation, the heat transfer characteristics of the internal passages is significantly different in an unfavorable way. So the aim of the turbine cooling research is to have concepts which provide highly efficient and uniform cooling. The negative effects of rotation has been addressed in this study and new orientation of two-pass cooling channels has been presented which utilizes the rotational energy in favor of heat transfer enhancement on both pressure and suction side internal walls. Present study has led to several new cooling concepts which are efficient under both stationary and rotating conditions. Ph. D.

While the gas turbine research community is continuously pursuing development of higher cyclic efficiency designs by increasing the combustor firing temperatures and thermally resistant turbine vane / blade materials, a simultaneous effort to reduce the emission levels of high temperature driven thermal NOX also needs to be addressed. Lean premixed combustion has been found as one of the solutions to these objectives. However, since less amount of air is available for backside cooling of liner walls, it becomes very important to characterize the convective heat transfer that occurs on the inside wall of the combustor liners. These studies were explored using laboratory scale experiments as well as numerical approaches for several inlet flow conditions under both non-reacting and reacting flows. These studies may be expected to provide valuable insights for the industrial design communities towards identifying thermal hot spot locations as well as in quantifying the heat transfer magnitude, thus aiding in effective designs of the liner walls. Lean premixed gas turbine combustor flows involve strongly coupled interactions between several aspects of physics such as the degree of swirl imparted by the inlet fuel nozzle, premixing of the fuel and incoming air, lean premixed combustion within the combustor domain, the interaction of swirling flow with combustion driven heat release resulting in flow dilation, the resulting pressure fluctuations leading to thermo-acoustic instabilities there by creating a feedback loop with incoming reactants resulting in flow instabilities leading to flame lift off, flame extinction etc. Hence understanding combustion driven swirling flow in combustors continues to be a topic of intense research. In the present study, numerical predictions of swirl driven combustor flows were analyzed for a specific swirl number of an industrial fuel nozzle (swirler) using a commercial computational fluid dynamics tool and compared against in-house experimental data. The latter data was obtained from a newly developed test rig at Applied Propulsion and Power Laboratory (APPL) at Virginia Tech. The simulations were performed and investigated for several flow Reynolds numbers under non-reacting condition using various two equation turbulence models as well as a scale resolving model. The work was also extended to reacting flow modeling (using a partially premixed model) for a specific Reynolds number. These efforts were carried out in order investigate the flow behavior and also characterize convective heat transfer along the combustor wall (liner). Additionally, several parametric studies were performed towards investigating the effect of combustor geometry on swirling flow and liner hear transfer; and also to investigate the effect of inlet swirl on the jet impingement location along the liner wall under both non-reacting as well as reacting conditions. The numerical results show detailed comparison against experiments for swirling flow profiles within the combustor under reacting conditions indicating a good reliability of steady state modeling approaches for reacting conditions; however, the limitations of steady state RANS turbulence models were observed for non-reacting swirling flow conditions, where the flow profiles deviate from experimental observations in the central recirculation region. Also, the numerical comparison of liner wall heat transfer characteristics against experiments showed a sensitivity to Reynolds numbers. These studies offer to provide preliminary insights of RANS predictions based on commercial CFD tools in predicting swirling, non-reacting and reacting flow and heat transfer. They can be extended to reacting flow heat transfer studies in future and also may be upgraded to unsteady LES predictions to complement future experimental observations conducted at the in-house test facility. Ph. D.

The performance of a vapor compression system is known to be affected by the ingestion of liquid droplets in the compressor. In these multiphase flows, the liquid and the vapor phase are tightly coupled. Therefore the interphase heat, mass and momentum transfer as well as droplet dynamics including droplet breakup and droplet-wall interactions play a vital role in governing these flows. Only thermodynamic analyses or two-dimensional mean-line calculations are not sufficient to gain an in-depth understanding of the complex multiphase flow field within the compressor. The objective of this research was to extend the current understanding of the operation of a multistage centrifugal compressor under two-phase flow conditions, by performing three-dimensional computational analysis. In this work, two-phase flow of a single constituent (refrigerant R134a) through a two-stage, in-line centrifugal compressor was analyzed using CFD. The CFD model accounted for real gas behavior of the vapor phase. Novel user defined routines were implemented to ensure accurate calculations of interphase heat, mass and momentum transfer terms and to model droplet impact on the compressor surfaces. An erosion model was developed and implemented to locate the erosion "hot spots" and to estimate the amount of material eroded. To understand the effects of increasing liquid carryover, the mass flow rate of the liquid phase was increased from 1% to 5% of the vapor mass flow rate. The influence of droplet size on the compressor performance was assessed by varying the droplet diameter at the inlet from 100 microns to 400 microns. The results of the two-phase flow simulations were compared with the simulation involving only the vapor phase. Liquid carryover altered the flow field within the compressor, and as a result, both impellers were observed to operate at off-design conditions. This effect was more pronounced for the second impeller. The overall effects of liquid carryover were detrimental to the compressor performance. The erosion calculations showed maximum erosion potential on the blade and shroud of the first impeller. The results from this investigation provided new and useful information that can be used to support improved design solutions. Ph. D.

The delivery of drugs into solid tumors is not trivial due to obstructions in the tumor microenvironment. Innovative drug delivery vehicles are currently being designed to overcome this challenge. In this research, computational fluid dynamics (CFD) simulations were used to evaluate the behavior of several drug delivery vectors in tumor capillaries—specifically motile bacteria, non-motile bacteria, and nanoparticles. Red blood cells, bacteria, and nanoparticles were imposed in the flow using the immersed boundary method. A human capillary model was developed using a novel method of handling deformable red blood cells (RBC). The capillary model was validated with experimental data from the literature. A stochastic model of bacteria motility was defined based on experimentally observed run and tumble behavior. The capillary and bacteria models were combined to simulate the intracapillary transport of bacteria. Non-motile bacteria and nanoparticles of 200 nm, 300 nm, and 405 nm were also simulated in capillary flow for comparison to motile bacteria. Motile bacteria tended to swim into the plasma layer near the capillary wall, while non-motile bacteria tended to get caught in the bolus flow between the RBCs. The nanoparticles were more impacted by Brownian motion and small scale fluid fluctuations, so they did not trend toward a single region of the flow. Motile bacteria were found to have the longest residence time in a 1 mm long capillary as well as the highest average radial velocity. This suggests motile bacteria may enter the interstitium at a higher rate than non-motile bacteria or nanoparticles of diameters between 200–405 nm. Master of Science

The thermal efficiency of a gas turbine directly depends on the rotor inlet temperature. The ever increasing demand for more power and advances in the field of engineering enabled this temperature to be pushed higher. But the material strength of the blades and vanes can often impose restrictions on the thermal load it can bear. This is where gas turbine cooling becomes very critical and a better cooling design has the potential to extend the blade life span, enables higher rotor inlet temperatures, conserves compressor bleed air. Among various kinds of cooling involved in gas turbines, film cooling will be the subject of this study. A novel concept for film cooling holes referred to as anti-vortex design proposed in 2007 is explored in this study. Coolant exits through two bifurcated cylindrical holes that branched out on either side of the central hole resulting in a tripod-like arrangement. Coolant from the side holes interacted with the mainstream and produced vortices that countered the main central rotating vortex pairs, weakening it and pushing the coolant jet towards the surface. In order to understand the performance of this anti-vortex tripod film cooling, a flat plate test setup and a low speed subsonic wind tunnel linear cascade were built. Transient heat transfer experiments were carried out in the flat plate test setup using Infrared thermography. Film cooling performance was quantified by measuring adiabatic effectiveness and heat transfer coefficient ratio. In order to gauge the performance, other standard hole geometries were also tested and compared with. Following the results from the flat plate test rig, film cooling performance was also evaluated on the surface of an airfoil. Adiabatic effectiveness was measured at different coolant mass flow rates. The tripod hole consistently provided better cooling compared to the standard cylindrical hole in both the flat plate and cascade experiments. In order to understand the anti-vortex concept which is one of the primary reason behind better performance of the tripod film cooling hole geometry, numerical simulations (CFD) were carried out at steady state using RANS turbulence models. The interaction of the coolant from the side holes with the mainstream forms vortices that tries to suppress the vortex formed by the central hole. This causes the coolant jet from the central to stay close to the surface and increases its coverage. Additionally, the coolant getting distributed into three individual units reduces the exit momentum ratio. Tripod holes were found to be capable of providing better effectiveness even while consuming almost half the coolant used by the standard cylindrical holes. Ph. D.

In this dissertation, various VAT parameters are investigated to determine the effect of the overall efficiency of the turbine at a high Reynolds number. To increase the efficiency of the vertical axis turbines, 2D CFD simulations are completed in an effort to better understand the physics behind the operation of these turbines. Specifically, the effect of advance ratio, solidity, and wake interactions were investigated. Simulations were completed in OpenFOAM using the k-ω SST turbulence model at a nominal Reynolds number of 500,000 using a NACA 0015 airfoil. To simulate the motion of the turbine, Arbitrary Mesh Interfacing (AMI) was used. For all of the parameters tested, it was found that the geometric effective angle of attack seen by the turbine blades had a significant impact on the power extracted from the flow. The range of effective angles of attack was found to decrease as the advance ratio increased. In spite of this, a severe loss in the power coefficient occurred at an advance ratio of 2.5 during which the blade experienced dynamic stall. This effect was also seen when the number of turbine blades was changed to four, at a solidity of 1.08. This negative impact on performance was found to be due to the increase in the drag component of the tangential force when dynamic stall occurs. Results indicate that wake interactions between subsequent blades have a large impact on performance especially when the wake interaction alters the flow direction sufficiently to create conditions for dynamic stall. To improve the performance of the VAT in the presence of dynamic stall, calculations were completed of a static twisted blade profile using GenIDLEST and OpenFOAM. There was found to be no improvement in the lift coefficient when comparing the twisted blade profile with a 2D blade at the same median angle of attack as the twisted blade. To further see the effects of the twisted blade, an effective VAT pitching motion was given to the blade and again compared to a 2D blade with the same motion. In this case there was significant improvement seen in the performance of the twisted blade. Master of Science

Energy conservation, efficiency, and environmental responsibility are priorities for modern energy technologies. The ever increasing demands for lower pollutants and higher performance have driven the development of low-emission gas turbine engines, operating at lean equivalence ratios and at increasingly higher turbine inlet temperatures. This has placed new constraints on gas turbine combustor design, particularly in regards to the cooling technologies available for the combustor liner walls. To optimize combustor thermal management, and in turn optimize overall engine performance, detailed measurements of the flame side heat transfer are required. However, given the challenging environment at which gas turbine combustors operate, there are currently only limited studies that quantify flame side combustor heat transfer; in particular at reacting conditions. The objective of the present work was to develop methodologies to measure heat transfer within a reacting gas turbine combustor. To accomplish this, an optically accessible research combustor system was designed and constructed at Virginia Tech, capable of operating at 650 K inlet temperature, maximum air mass flow rates of 1.3 kg/s, and flame temperatures over 1800 K. Flow and heat transfer measurements at non-reacting and reacting conditions were carried out for Reynolds numbers (Re) with respect to the combustor diameter ranging from ~11 500 to ~140 000 (depending on the condition). Particle Image Velocimetry (PIV) was used to measure the non-reacting flow field within the burner, leading to the identification of coherent structures in the flow that accounted for over 30% of the flow fluctuation kinetic energy along the swirling jet shear layers. The capability of infrared (IR) thermography to image surface temperatures through a fused silica (quartz) glass was demonstrated at non-reacting conditions. IR thermography was then used to measure the non-reacting steady state heat transfer along the combustor liner. A peak in heat transfer was identified at ~1 nozzle diameter downstream of the combustor dome plate. The peak Nusselt number along the liner was over 18 times higher than that predicted from fully developed turbulent pipe flow correlations, which have traditionally been used to estimate flame side combustor heat transfer. For the reacting measurements, a novel time-dependent heat transfer methodology was developed that allowed for the investigation of transient heat loads, including those occurring during engine ignition and shutdown. The methodology was validated at non-reacting conditions, by comparing results from an experiment with changing flow temperature, to the results obtained at steady state. The difference between the time-dependent and the steady state measurements were between 3% and 17.3% for different mass flow conditions. The time-dependent methodology was applied to reacting conditions for combustor Reynolds numbers of ~12 000 and ~24 000. At an equivalence ratio of ~0.5 and a combustor Reynolds number of ~12 000, the peak heat load location in reaction was shifted downstream by 0.2 nozzle diameters compared to the non-reacting cases. At higher equivalence ratios, and more visibly at a Reynolds number of ~24 000, the heat transfer distribution along the combustor liner exhibited two peaks, upstream and downstream of the impingement location (X/DN=0.8-1.0 and X/DN=2.5). Reacting PIV was performed at Re=12 000 showing the presence of a strong corner recirculation, which could potentially convect reactants upstream of the impingement point, leading to the double peak structure observed. The methodologies developed have provided insight into heat transfer within gas turbine combustors. The methods can be used to explore additional conditions and expand the dataset beyond what is presented, to fully characterize reacting combustor heat transfer. Ph. D.

In this thesis, the ribbed ducts of the internal cooling passage in turbine blading are investigated to demonstrate the effects of high speed rotation. Rotation coupled with high temperature operating conditions alters the mean flow, turbulence, and heat transfer augmentation due to Coriolis and centrifugal buoyancy forces that arises from density stratification in the domain. Gas turbine engines operate in particle laden environments (sand, volcanic ash), and particulate matter ingested by the engine can make their way into the blade internal cooling passages over thousands of operating hours. These particulates can deposit on the walls of these cooling passages and degrade performance of the turbine blade. Large-Eddy Simulations (LES) with temperature dependent properties is used for turbulent flow and heat transfer in the ribbed cooling passages and Lagrangian tracking is used to calculate the particle trajectories together with a wall deposition model. The conditions used are Re=100,000, Rotation number, Ro = 0.0 and 0.2, and centrifugal Buoyancy parameters of Bo=0, 0.5, and 1.0. First, the independent effects of Coriolis and centrifugal buoyancy forces are investigated, with a focus on the additional augmentation obtained in heat transfer with the addition of centrifugal buoyancy. Coriolis forces are known to augment heat transfer at the trailing wall and attenuate the same at the leading wall. Phenomenological arguments stated that centrifugal buoyancy augments the effects of Coriolis forces in outward flow in the first pass while opposing the effect of Coriolis forces during inward flow in the second pass. In this study, it was found that in the first pass, centrifugal buoyancy had a greater effect in augmenting heat transfer at the trailing wall than in attenuating heat transfer at the leading wall. On the contrary, it aided heat transfer in the second half of the first pass at the leading wall by energizing the flow near the wall. Also, contrary to phenomenological arguments, inclusion of centrifugal buoyancy augmented heat transfer over Coriolis forces alone on both the leading and trailing walls of the second pass. Sand ingestion is then investigated, by injecting 200,000 particles in the size range of 0.5-175μm with 65% of the particles below 10 μm. Three duct wall temperatures are investigated, 950, 1000 and 1050 °C with an inlet temperature of flow and particles at 527 °C . The impingement, deposition levels, and impact characteristics are recorded as the particles move through the domain. It was found that the Coriolis force greatly increases deposition. This was made prevalent in the first pass, as 84% of the deposits in the domain occurred in the first pass for the rotating case, whereas only 27% of deposits occurred in the first pass for the stationary case with the majority of deposits occurring in the bend region. This was due to an increased interaction with the trailing wall in the rotating case whereas particles in the stationary case were allowed to remain in the mean flow and gain momentum, making rebounding from a wall during collision more likely than deposition. In contrast, the variation of wall temperatures caused little to no change in deposition levels. This was concluded to be a result of the high Reynolds number used in the flow. At high Reynolds numbers, the particles have a short residence times in the internal cooling circuit not allowing the flow and particles to heat up to the wall temperature. Overall, 87% of the injected particles deposited in the rotating duct whereas 58% deposited in the stationary duct. Master of Science

This dissertation presents an experimentally supported multi-physics model of a dielectric barrier discharge boundary layer flow control actuator. The model is independent of empirical data about the specific behavior of the system. This model contributes to the understanding of the specific mechanisms that enable the actuator to induce flow control. The multi-physics numerical model couples a fluid model, a chemistry model, and an electrostatics model. The chemistry model has been experimentally validated against known spectroscopic techniques, and the fluid model has been experimentally validated against the time-resolved shadowgraphy. The model demonstrates the capability to replicate emergent flow structures near a wall. These structures contribute to momentum transport that enhance the boundary layer’s wall attachment and provide for better flow control. An experiment was designed to validate the model predictions. The spectroscopic results confirmed the model predictions of an electron temperature of 0.282eV and an electron number density of 65.5 × 10⁻¹²kmol/m³ matching to within a relative error of 12.4% and 14.8%, respectively. The shadowgraphic results also confirmed the model predicted velocities of flow structures of 3.75m/s with a relative error of 10.9%. The distribution of results from both experimental and model velocity calculations strongly overlap each other. This validated model provides new and useful information on the effect of Dielectric Barrier Discharge actuators on flow control and performance. This work was supported in part by NSF grant CNS-0960081 and the HokieSpeed supercomputer at Virginia Tech. Ph. D.

This dissertation presents a formulation of the Continuous Sensitivity Equation Method (CSEM) applied to the Computational Fluid Dynamics (CFD) simulation of thermoacoustic instability problems. The proposed sensitivity analysis approach only requires a single run of the CFD simulation. Moreover, the sensitivities of field variables, pressure, velocity and temperature to boundary-condition parameters are directly obtained from the solution to sensitivity equations. Thermoacoustic instability is predicted by the Rayleigh criterion. The sensitivity of the Rayleigh index is computed utilizing the sensitivities of field variables. The application of the CSEM to thermoacoustic instability problems is demonstrated by two classic examples. The first example explores the effects of the heated wall temperature on the one-dimensional thermoacoustic convection. The sensitivity of the Rayleigh index, which is the indicator of thermoacoustic instabilities, is computed by the sensitivity of field variables. As the heat wall temperature increases, the sensitivity of the Rayleigh index decreases. The evolution from positive to negative sensitivity values suggests the transition from a destabilizing trend to stabilizing trend of the thermoacoustic system. Thermoacoustic instabilities in a self-excited Rijke tube are investigated following the relatively simple thermoacoustic convection problem. The complexity of simulating the Rijke tube increases in both dimensions and mechanisms which incorporate the species transport process and chemical reactions. As a representative model of the large lean premixed combustor, Rijke tube has been extensively studied. Quantitative sensitivity analysis sets the present work apart from previous research on the prediction and control of thermoacoustic instabilities. The effects of two boundary-condition parameters, i.e. the inlet mass flow rate and the equivalence ratio, are tested respectively. Small variations in both parameters predict a rapid change in sensitivities of field variables in the early stage of the total time length of 1.2s. The sensitivity of the Rayleigh index "blows up" at a specific time point of the early stage. In addition, variations in the inlet mass flow rate and the equivalence ratio lead to opposite effects on the sensitivity of the Rayleigh index. There exist some common findings on the application of the CSEM. For both thermoacoustic problems, the sensitivities of field variables and the Rayleigh index exhibit oscillatory nature, confirming that thermoacoustic instability is an overall effect of the coupling process between fluctuations of pressure and heat release rate. All the sensitivities of the Rayleigh index show rapid changes and "blow up" in the early stage. Although the numerical errors could influence the fidelity of computational results, it is believed that the rapid changes reflect the susceptibility to thermoacoustic instabilities in the studied systems. It should also be noted that the sensitivities are obtained for small variations in influential parameters. Therefore, the resulting sensitivities do not predict the occurrence of thermoacoustic instabilities under a condition that is far from the reference state determined by either CFD simulation results (employed in this dissertation) or experimental data. The sensitivity solver developed for the present research has the feature of flexibility. Additional mechanisms and more complicated instability criteria could be easily incorporated into the solver. Moreover, the sensitivity equations formulated in this dissertation are derived from the full set of nonlinear governing equations. Therefore, it is possible to extend the use of the sensitivity solver to other CFD problems. The developed sensitivity solver needs to be optimized to gain better performance, which is considered to be the primary future work of this research. Ph. D.

The need for energy efficient, clean and quiet, energy conversion devices for mobile and stationary applications has presented proton exchange membrane (PEM) fuel cells as a potential energy source. The use of PEM fuel cells for automotive and other transient applications, where there are rapid changes in load, presents a need for better understanding of transient behavior. In particular at low humidity operations; one of the factors critical to the performance and durability of fuel cell systems is water transport in various fuel cell layers, including water absorption in membrane. An essential aspect to optimization of transient behavior of fuel cells is a fundamental understanding of response of fuel cell system to dynamic changes in load and operating parameters. This forms the first objective of the dissertation. An insight in to the time scales associated with various transport phenomena will be discussed in detail. In the second component on the study, the effects of membrane properties on the dynamic behavior of the fuel cells are analyzed with focus on membrane dry-out for low humidity operations. The mechanical behavior of the membrane is directly related to the changes in humidity levels in membrane and is explored as a part third objective of the dissertation. Numerical studies addressing this objective will be presented. Finally, porous media undergoing physical deposition (or erosion) are common in many applications, including electrochemical systems such as fuel cells (for example, electrodes, catalyst layer s, etc.) and batteries. The transport properties of these porous media are a function of the deposition and the change in the porous structures with time. A dynamic fractal model is introduced to describe such structures undergoing deposition and, in turn, to evaluate the changes in their physical properties as a function of the deposition. Ph. D.