Browsing by Author "Xiao, Heng"

Now showing 1 - 20 of 40
  • Thumbnail Image
    4D combustion and flow diagnostics based on tomographic chemiluminescence (TC) and volumetric laser-induced fluorescence (VLIF)
    Wu, Yue (Virginia Tech, 2016-12-02)
    Optical diagnostics have become indispensable tools for the study of turbulent flows and flames. However, optical diagnostics developed in the past have been primarily limited to measurements at a point, along a line, or across a two-dimensional (2D) plane; while turbulent flows and flames are inherently four-dimensional (three-dimensional in space and transient in time). As a result, diagnostic techniques which can provide 4D measurement have been long desired. The purpose of this dissertation is to investigate two of such 4D diagnostics both for the fundamental study of turbulent flow and combustion processes and also for the applied research of practical devices. These two diagnostics are respectively code named tomographic chemiluminescence (TC) and volumetric laser induced fluorescence (VLIF). For the TC technique, the emission of light as the result of combustion (i.e. chemiluminescence) is firstly recorded by multiple cameras placed at different orientations. A numerical algorithm is then applied on the data recorded to reconstruct the 4D flame structure. For the VLIF technique, a laser is used to excite a specific species in the flow or flame. The excited species then de-excite to emit light at a wavelength longer than the laser wavelength. The emitted light is then captured by optical sensors and again, the numerical algorithm is applied to reconstruct the flow or flame structure. This dissertation describes the numerical and experimental validation of these two techniques, and explores their capabilities and limitations. It is expected that the results obtained in this dissertation lay the groundwork for further development and expanded application of 4D diagnostics for the study of turbulent flows and combustion processes.
  • Thumbnail Image
    Accelerating a Coupled SPH-FEM Solver through Heterogeneous Computing for use in Fluid-Structure Interaction Problems
    Gilbert, John Nicholas (Virginia Tech, 2015-06-08)
    This work presents a partitioned approach to simulating free-surface flow interaction with hyper-elastic structures in which a smoothed particle hydrodynamics (SPH) solver is coupled with a finite-element (FEM) solver. SPH is a mesh-free, Lagrangian numerical technique frequently employed to study physical phenomena involving large deformations, such as fragmentation or breaking waves. As a mesh-free Lagrangian method, SPH makes an attractive alternative to traditional grid-based methods for modeling free-surface flows and/or problems with rapid deformations where frequent re-meshing and additional free-surface tracking algorithms are non-trivial. This work continues and extends the earlier coupled 2D SPH-FEM approach of Yang et al. [1,2] by linking a double-precision GPU implementation of a 3D weakly compressible SPH formulation [3] with the open source finite element software Code_Aster [4]. Using this approach, the fluid domain is evolved on the GPU, while the CPU updates the structural domain. Finally, the partitioned solutions are coupled using a traditional staggered algorithm.
  • Thumbnail Image
    Acoustic Inversion for Uncertainty Reduction in Reynolds-Averaged Navier-Stokes-Based Jet Noise Prediction
    Zhang, Xin-Lei; Xiao, Heng; Wu, Ting; He, Guowei (American Institute of Aeronautics and Astronautics, 2021-12-13)
    The Reynolds-averaged Navier–Stokes (RANS)-based method is a practical tool to provide rapid assessment of jet noise-reduction concepts. However, the RANS-based method requires modeling assumptions to represent noise generation and propagation, which often reduces the predictive accuracy due to the model-form uncertainties. In this work, the ensemble Kalman filter-based acoustic inversion method is introduced to reduce uncertainties in the turbulent kinetic energy and dissipation rate based on the far-field noise and the axial centerline velocity data. The results show that jet noise data are more effective from which to infer turbulent kinetic energy and dissipation rate compared to velocity data. Moreover, the inferred noise source is able to improve the estimation of the turbulent flowfield and the far-field noise at unobserved locations. Further, the noise model parameters are also considered uncertain quantities, demonstrating the ability of the proposed framework to reduce uncertainties in both the RANS and noise models. Finally, one realistic case with experimental data is investigated to show the practicality of the proposed framework. The method opens up the possibility for the inverse modeling of jet noise sources by incorporating far-field noise data that are relatively straightforward to be measured compared to the velocity field.
  • Thumbnail Image
    Application of Defocusing Technique to Bubble Depth Measurement
    Mugikura, Yuki (Virginia Tech, 2017)
    The thesis presents a defocusing technique to extract bubble depth information. Typically, when a bubble is out of focus in an image, the bubble is ignored by applying a filter or thresholding. However, it is known that a bubble image becomes blurred as the bubble moves away from the focal plane. Then, this technique is applied to determine the bubble distance along the optical path based on the blurriness or intensity gradient information of the bubble. Using the image processing algorithm, images captured in three different experiments are analyzed to develop a correlation between the bubble distance and its intensity gradient. The suggested models to predict the bubble depth are also developed based on the measurement data and evaluated with the measured data. When the intensity gradient of the bubble is lower or when a bubble is located farther from the focal plane, the model can predict the distance more accurately. However, the models show larger absolute and relative error when the bubble is near the focal plane. To improve the prediction in that region, another model should be considered. Also, depth of field analysis is introduced in order to compare three experimental results with different imaging setups. The applicability of the approach is analyzed and evaluated.
  • Thumbnail Image
    Application of r-Adaptation Techniques for Discretization Error Improvement in CFD
    Tyson, William Conrad (Virginia Tech, 2015-12-08)
    Computational fluid dynamics (CFD) has proven to be an invaluable tool for both engineering design and analysis. As the performance of engineering devices become more reliant upon the accuracy of CFD simulations, it is necessary to not only quantify and but also to reduce the numerical error present in a solution. Discretization error is often the primary source of numerical error. Discretization error is introduced locally into the solution by truncation error. Truncation error represents the higher order terms in an infinite series which are truncated during the discretization of the continuous governing equations of a model. Discretization error can be reduced through uniform grid refinement but is often impractical for typical engineering problems. Grid adaptation provides an efficient means for improving solution accuracy without the exponential increase in computational time associated with uniform grid refinement. Solution accuracy can be improved through local grid refinement, often referred to as h-adaptation, or by node relocation in the computational domain, often referred to as r-adaptation. The goal of this work is to examine the effectiveness of several r-adaptation techniques for reducing discretization error. A framework for geometry preservation is presented, and truncation error is used to drive adaptation. Sample problems include both subsonic and supersonic inviscid flows. Discretization error reductions of up to an order of magnitude are achieved on adapted grids.
  • Thumbnail Image
    Assessment of Model Validation, Calibration, and Prediction Approaches in the Presence of Uncertainty
    Whiting, Nolan Wagner (Virginia Tech, 2019-07-19)
    Model validation is the process of determining the degree to which a model is an accurate representation of the true value in the real world. The results of a model validation study can be used to either quantify the model form uncertainty or to improve/calibrate the model. However, the model validation process can become complicated if there is uncertainty in the simulation and/or experimental outcomes. These uncertainties can be in the form of aleatory uncertainties due to randomness or epistemic uncertainties due to lack of knowledge. Four different approaches are used for addressing model validation and calibration: 1) the area validation metric (AVM), 2) a modified area validation metric (MAVM) with confidence intervals, 3) the standard validation uncertainty from ASME VandV 20, and 4) Bayesian updating of a model discrepancy term. Details are given for the application of the MAVM for accounting for small experimental sample sizes. To provide an unambiguous assessment of these different approaches, synthetic experimental values were generated from computational fluid dynamics simulations of a multi-element airfoil. A simplified model was then developed using thin airfoil theory. This simplified model was then assessed using the synthetic experimental data. The quantities examined include the two dimensional lift and moment coefficients for the airfoil with varying angles of attack and flap deflection angles. Each of these validation/calibration approaches will be assessed for their ability to tightly encapsulate the true value in nature at locations both where experimental results are provided and prediction locations where no experimental data are available. Generally it was seen that the MAVM performed the best in cases where there is a sparse amount of data and/or large extrapolations and Bayesian calibration outperformed the others where there is an extensive amount of experimental data that covers the application domain.
  • Thumbnail Image
    Augmented Neural Network Surrogate Models for Polynomial Chaos Expansions and Reduced Order Modeling
    Cooper, Rachel Gray (Virginia Tech, 2021-05-20)
    Mathematical models describing real world processes are becoming increasingly complex to better match the dynamics of the true system. While this is a positive step towards more complete knowledge of our world, numerical evaluations of these models become increasingly computationally inefficient, requiring increased resources or time to evaluate. This has led to the need for simplified surrogates to these complex mathematical models. A growing surrogate modeling solution is with the usage of neural networks. Neural networks (NN) are known to generalize an approximation across a diverse dataset and minimize the solution along complex nonlinear boundaries. Additionally, these surrogate models can be found using only incomplete knowledge of the true dynamics. However, NN surrogates often suffer from a lack of interpretability, where the decisions made in the training process are not fully understood, and the roles of individual neurons are not well defined. We present two solutions towards this lack of interpretability. The first focuses on mimicking polynomial chaos (PC) modeling techniques, modifying the structure of a NN to produce polynomial approximations of the underlying dynamics. This methodology allows for an extractable meaning from the network and results in improvement in accuracy over traditional PC methods. Secondly, we examine the construction of a reduced order modeling scheme using NN autoencoders, guiding the decisions of the training process to better match the real dynamics. This guiding process is performed via a physics-informed (PI) penalty, resulting in a speed-up in training convergence, but still results in poor performance compared to traditional schemes.
  • Thumbnail Image
    Combined Experimental and Numerical Study of Active Thermal Control of Battery Modules
    He, Fan (Virginia Tech, 2015-04-16)
    Lithium ion (Li-ion) batteries have been identified as a promising solution to meet the increasing demands for alternative energy in electric vehicles (EVs) and hybrid electric vehicle (HEVs). This work describes experimental and numerical study of thermal management of battery module consisting of cylindrical Li-ion cells, with an emphasis on the use of active control to achieve optimal cooling performance with minimal parasitic power consumption. The major contribution from this work is the first experimental demonstration (based on our review of archival journal and conference literature) and the corresponding analysis of active thermal control of battery modules. The results suggest that the active control strategy, when combined with reciprocating cooling flow, can reduce the parasitic energy consumption and cooling flow amount substantially. Compared with results using passive control with unidirectional cooling flow, the parasitic energy consumption was reduced by about 80%. This contribution was achieved in three steps, which was detailed in this dissertation in chapters 2, 3, and 4, respectively. In the first step, an experimental facility and a corresponding CFD model were developed to capture the thermal behavior of multiple battery cells. Based on the experimental and CFD results, a reduced-order model (ROM) was then developed for active monitoring and control purposes. In the second step, the ROM was parameterized and an observer-based control strategy was developed to control the core temperature of battery cells. Finally, based on the experimental facility and the ROM model, the active control of a battery module was demonstrated. Each of these steps represents an important facet of the thermal management problem, and it is expected that the results and specifics documented in this dissertation lay the groundwork to facilitate further study.
  • Thumbnail Image
    Development and Use of a Spatially Accurate Polynomial Chaos Method for Aerospace Applications
    Schaefer, John Anthony (Virginia Tech, 2023-01-24)
    Uncertainty is prevalent throughout the design, analysis, and optimization of aerospace products. When scientific computing is used to support these tasks, sources of uncertainty may include the freestream flight conditions of a vehicle, physical modeling parameters, geometric fidelity, numerical error, and model-form uncertainty, among others. Moreover, while some uncertainties may be treated as probabilistic, aleatory sources, other uncertainties are non-probabilistic and epistemic due to a lack of knowledge, and cannot be rigorously treated using classical statistics or Bayesian approaches. An additional complication for propagating uncertainty is that many aerospace scientific computing tools may be computationally expensive; for example, a single high-fidelity computational fluid dynamics solution may require several days or even weeks to complete. It is therefore necessary to employ uncertainty propagation strategies that require as few solutions as possible. The Non-Intrusive Polynomial Chaos (NIPC) method has grown in popularity in recent decades due to its ability to propagate both aleatory and epistemic parametric sources of uncertainty in a computationally efficient manner. While traditional Monte Carlo methods might require thousands to millions of function evaluations to achieve statistical convergence, NIPC typically requires tens to hundreds for problems with similar numbers of uncertain dimensions. Despite this efficiency, NIPC is limited in one important aspect: it can only propagate uncertainty at a particular point in a design space or flight envelope. For optimization or aerodynamic database problems that require uncertainty estimates at many more than one point, the use of NIPC quickly becomes computationally intractable. This dissertation introduces a new method entitled Spatially Accurate Polynomial Chaos (SAPC) that extends the original NIPC approach for the spatial regression of aleatory and epistemic parametric sources of uncertainty. Throughout the dissertation, the SAPC method is applied to various aerospace problems of interest. These include the regression of aerodynamic force and moment uncertainties throughout the flight envelope of a commercial aircraft, the design under uncertainty of a two-stream propulsive mixer device, and the robust design of a low-boom supersonic demonstrator aircraft. Collectively the results suggest that SAPC may be useful for a large variety of engineering applications.
  • Thumbnail Image
    Development and Validation of Reconstruction Algorithms for 3D Tomography Diagnostics
    Lei, Qingchun (Virginia Tech, 2017-01-10)
    This work reports three reconstruction algorithms developed to address the practical issues encountered in 3D tomography diagnostics, such as the limited view angles available in many practical applications, the large scale and nonlinearity of the problems when they are in 3D, and the measurement uncertainty. These algorithms are: an algebraic reconstruction technique (ART) screening algorithm, a nonlinear iterative reconstruction technique (NIRT), and an iterative reconstruction technique integrating view registration optimization (IRT-VRO) algorithm. The ART screening algorithm was developed to enhance the performance of the traditional ART algorithm to solve linear tomography problems, the NIRT was to solve nonlinear tomography problems, and the IRT-VRO was to address the issue of view registration uncertainty in both linear and nonlinear problems. This dissertation describes the mathematical formulations, and the experimental and numerical validations for these algorithms. It is expected that the results obtained in this dissertation to lay the groundwork for their further development and expanded adaption in the deployment of tomography diagnostics in various practical applications.
  • Thumbnail Image
    Development of a Fast X-ray Line Detector System for Two-Phase Flow Measurement
    Song, Kyle (Virginia Tech, 2016-12-08)
    Measuring void fraction distribution in two-phase flow has been a challenging task for many decades because of its complex and fast-changing interfacial structure. In this study, a non-intrusive X-ray measurement system is developed and calibrated to mitigate this challenge. This approach has several advantages over the conventional methods such as the multi-sensor conductivity probe, wire-mesh sensor, impedance void meter, or direct optical imaging. The X-ray densitometry technique is non-intrusive, insensitive to flow regime changes, capable of measuring high temperature or high-pressure flows, and has reasonable penetration depth. With the advancement of detector technology, the system developed in this work can further achieve high spatial resolution (100 micron per pixel) and high temporal resolution (1000 frames per second). This work mainly focuses on the following aspects of the system development: establishing a geometrical model for the line detector system, conducting spectral analysis for X-ray attenuation in two-phase flow, and performing calibration tests. The geometrical model has considered the measurement plane, geometry of the test-section wall and flow channel, relative position of the X-ray source and detector pixels. By assuming axisymmetry, an algorithm has been developed to convert void fraction distribution along the detector pixels to the radial void profile in a circular pipe. The X-ray spectral analysis yielded a novel prediction model for non-chromatic X-rays and non-uniform structure materials such as the internal two-phase flow which contains gas, liquid and solid wall materials. A calibration experiment has been carried out to optimize the detector conversion factor for each detector pixels. Finally, the data measured by the developed X-ray system are compared with the double-sensor conductivity probe and gas flow meter for sample bubbly flow and slug flow conditions. The results show reasonable agreement between these different measuring techniques.
  • Thumbnail Image
    Development of Advanced Image Processing Algorithms for Bubbly Flow Measurement
    Fu, Yucheng (Virginia Tech, 2018-10-16)
    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.
  • Thumbnail Image
    Development of Surrogate Model for FEM Error Prediction using Deep Learning
    Jain, Siddharth (Virginia Tech, 2022-07-07)
    This research is a proof-of-concept study to develop a surrogate model, using deep learning (DL), to predict solution error for a given model with a given mesh. For this research, we have taken the von Mises stress contours and have predicted two different types of error indicators contours, namely (i) von Mises error indicator (MISESERI), and (ii) energy density error indicator (ENDENERI). Error indicators are designed to identify the solution domain areas where the gradient has not been properly captured. It uses the spatial gradient distribution of the existing solution for a given mesh to estimate the error. Due to poor meshing and nature of the finite element method, these error indicators are leveraged to study and reduce errors in the finite element solution using an adaptive remeshing scheme. Adaptive re-meshing is an iterative and computationally expensive process to reduce the error computed during the post-processing step. To overcome this limitation we propose an approach to replace it using data-driven techniques. We have introduced an image processing-based surrogate model designed to solve an image-to-image regression problem using convolutional neural networks (CNN) that takes a 256 × 256 colored image of von mises stress contour and outputs the required error indicator. To train this model with good generalization performance we have developed four different geometries for each of the three case studies: (i) quarter plate with a hole, (b) simply supported plate with multiple holes, and (c) simply supported stiffened plate. The entire research is implemented in a three phase approach, phase I involves the design and development of a CNN to perform training on stress contour images with their corresponding von Mises stress values volume-averaged over the entire domain. Phase II involves developing a surrogate model to perform image-to-image regression and the final phase III involves extending the capabilities of phase II and making the surrogate model more generalized and robust. The final surrogate model used to train the global dataset of 12,000 images consists of three auto encoders, one encoder-decoder assembly, and two multi-output regression neural networks. With the error of less than 1% in the neural network training shows good memorization and generalization performance. Our final surrogate model takes 15.5 hours to train and less than a minute to predict the error indicators on testing datasets. Thus, this present study can be considered a good first step toward developing an adaptive remeshing scheme using deep neural networks.
  • Thumbnail Image
    Ensemble Gradient for Learning Turbulence Models from Indirect Observations
    Strofer, Carlos A. Michelen; Zhang, Xin-Lei; Xiao, Heng (Global Science Press, 2021-11-01)
    Training data-driven turbulence models with high fidelity Reynolds stress can be impractical and recently such models have been trained with velocity and pressure measurements. For gradient-based optimization, such as training deep learning models, this requires evaluating the sensitivities of the RANS equations. This paper explores the use of an ensemble approximation of the sensitivities of the RANS equations in training data-driven turbulence models with indirect observations. A deep neural network representing the turbulence model is trained using the network’s gradients obtained by backpropagation and the ensemble approximation of the RANS sensitivities. Different ensemble approximations are explored and a method based on explicit projection onto the sample space is presented. As validation, the gradient approximations from the different methods are compared to that from the continuous adjoint equations. The ensemble approximation is then used to learn different turbulence models from velocity observations. In all cases, the learned model predicts improved velocities. However, it was observed that once the sensitivity of the velocity to the underlying model becomes small, the approximate nature of the ensemble gradient hinders further optimization of the underlying model. The benefits and limitations of the ensemble gradient approximation are discussed, in particular as compared to the adjoint equations.
  • Thumbnail Image
    Experimental and Modeling Study of the Thermal Management of Li-ion Battery Packs
    Wang, Haoting (Virginia Tech, 2017-10-13)
    This work reports the experimental and numerical study of the thermal management of Li-ion battery packs under the context of electric vehicle (EV) or hybrid EV (HEV) applications. Li-ion batteries have been extensively demonstrated as an important power source for EVs or HEVs. However, thermal management is a critical challenge for their widespread deployment, due to their highly dynamic operation and the wide range of environments under which they operate. To address these challenges, this work developed several experimental platforms to study adaptive thermal management strategies. Parallel to the experimental effort, multi-disciplinary models integrating heat transfer, fluid mechanics, and electro-thermal dynamics have been developed and validated, including detailed CFD models and lumped parameter models. The major contributions are twofold. First, this work developed actively controlled strategies and experimentally demonstrated their effectiveness on a practical sized battery pack and dynamic thermal loads. The results show that these strategies effectively reduced both the parasitic energy consumption and the temperature non-uniformity while maintaining the maximum temperature rise in the pack. Second, this work established a new two dimensional lumped parameter thermal model to overcome the limitations of existing thermal models and extend their applicable range. This new model provides accurate surface and core temperatures simulations comparable to detailed CFD models with a fraction of the computational cost.
  • Thumbnail Image
    Frame-independent vector-cloud neural network for nonlocal constitutive modeling on arbitrary grids
    Zhou, Xu-Hui; Han, Jiequn; Xiao, Heng (Elsevier, 2022-01-01)
    Constitutive models are widely used for modeling complex systems in science and engineering, where first-principle-based, well-resolved simulations are often prohibitively expensive. For example, in fluid dynamics, constitutive models are required to describe nonlocal, unresolved physics such as turbulence and laminar–turbulent transition. However, traditional constitutive models based on partial differential equations (PDEs) often lack robustness and are too rigid to accommodate diverse calibration datasets. We propose a frame-independent, nonlocal constitutive model based on a vector-cloud neural network that can be learned with data. The model predicts the closure variable at a point based on the flow information in its neighborhood. Such nonlocal information is represented by a group of points, each having a feature vector attached to it, and thus the input is referred to as vector cloud. The cloud is mapped to the closure variable through a frame-independent neural network, invariant both to coordinate translation and rotation and to the ordering of points in the cloud. As such, the network can deal with any number of arbitrarily arranged grid points and thus is suitable for unstructured meshes in fluid simulations. The merits of the proposed network are demonstrated for scalar transport PDEs on a family of parameterized periodic hill geometries. The vector-cloud neural network is a promising tool not only as nonlocal constitutive models and but also as general surrogate models for PDEs on irregular domains.
  • Thumbnail Image
    High-Fidelity Numerical Simulation of Shallow Water Waves
    Zainali, Amir (Virginia Tech, 2016-12-09)
    Tsunamis impose significant threat to human life and coastal infrastructure. The goal of my dissertation is to develop a robust, accurate, and computationally efficient numerical model for quantitative hazard assessment of tsunamis. The length scale of the physical domain of interest ranges from hundreds of kilometers, in the case of landslide-generated tsunamis, to thousands of kilometers, in the case of far-field tsunamis, while the water depth varies from couple of kilometers, in deep ocean, to few centimeters, in the vicinity of shoreline. The large multi-scale computational domain leads to challenging and expensive numerical simulations. I present and compare the numerical results for different important problems --- such as tsunami hazard mitigation due to presence of coastal vegetation, boulder dislodgement and displacement by long waves, and tsunamis generated by an asteroid impact --- in risk assessment of tsunamis. I employ depth-integrated shallow water equations and Serre-Green-Naghdi equations for solving the problems and compare them to available three-dimensional results obtained by mesh-free smoothed particle hydrodynamics and volume of fluid methods. My results suggest that depth-integrated equations, given the current hardware computational capacities and the large scales of the problems in hand, can produce results as accurate as three-dimensional schemes while being computationally more efficient by at least an order of a magnitude.
  • Thumbnail Image
    Investigation of Momentum and Heat Transfer in Flow Past Suspensions of Non-Spherical Particles
    Cao, Ze (Virginia Tech, 2021-03-11)
    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.
  • Thumbnail Image
    Large Eddy simulation of Trailing Edge Acoustic Emissions of an Airfoil
    Wu, Jinlong; Devenport, William J.; Paterson, Eric G.; Sun, Rui; Xiao, Heng (Virginia Tech, 2015-06)
    The present investigation of trailing edge acoustic emission of an airfoil concerns the effects of the broadband noise generated by the interaction of turbulent boundary layer and airfoil trailing edge, and the tonal noise generated by the vortex shedding of trailing edge bluntness. Large eddy simulation (LES) is performed on an NACA0012 airfoil with blunt trailing edge at a Reynolds number Rec = 400; 000 based on the airfoil chord length for three different configurations with different angles of attack. In order to reproduce and compare with the result from experiment in the literature, numerical tripping is tested and chosen to control the boundary layer development to guarantee a similar boundary layer thickness near the airfoil trailing edge. The near wall region inside the boundary layer is directly resolved by LES simulation with Van Driest damping, in order to obtain the instantaneous data in that region. With these instantaneous data from aerodynamic simulation, the acoustic predication is conducted by the Curle's analogy, which is suitable for stationary surface in free ow. To validate the numerical solutions, both ow simulation and acoustic integration results are compared to experimental data and simulation results available in the literature, and good agreement is achieved. The aerodynamic results show that the similar boundary layer development of experimental result can be reproduced by simulation with a suitable choice of numerical tripping, and the similar instantaneous behavior of ow inside the boundary layer is therefore guaranteed, which is vital for the acoustic prediction. The aeroacoustic results show that the acoustic prediction changes with the lift and drag force provided by the airfoil. Basically speaking, it's a result that the unsteady force around the surface is closely related to the mean force provided by an airfoil, which means that the noise control of a given airfoil is coupled with the optimization of its aerodynamic performance. As for the approximation made in the implemetation of Curle's analogy, it is shown in the aeroacoustic results that the airfoil can be treated as a compact point only if low frequency acoustic emission is of interest, and such kind of approximation can cause obvious problem if very high frequency acoustic emission is concerned.
  • Thumbnail Image
    Machine Learning and Data Fusion of Simulated Remote Sensing Data
    Higgins, Erik Tracy (Virginia Tech, 2023-07-27)
    Modeling and simulation tools are described and implemented in a single workflow to develop a means of simulating a ship wake followed by simulated synthetic aperture radar (SAR) and infra-red (IR) images of these ship wakes. A parametric study across several different ocean environments and simulated remote sensing platforms is conducted to generate a preliminary data set that is used for training and testing neural network--based ship wake detection models. Several different model architectures are trained and tested, which are able to provide a high degree of accuracy in classifying whether input SAR images contain a persistent ship wake. Several data fusion models are explored to understand how fusing data from different SAR bands may improve ship wake detection, with some combinations of neural networks and data fusion models achieving perfect or near-perfect performance. Finally, an outline for a future study into multi-physics data fusion across multiple sensor modalities is created and discussed.