Doctor of Philosophy, The Ohio State University, 2019, Mechanical Engineering

Understanding the complex phenomena occurring inside the catalyst layer of a proton exchange membrane fuel cell (PEMFC) is critical to design of an optimized structure with low platinum loading and high performance. Describing detailed physical and chemical processes in the catalyst layer at the resolution of pore scale, pore-scale simulation is considered as a promising approach for use in understanding the structure-performance relation and subsequent optimization of the catalyst layer. For wide spread use in industry, the computational cost of pore-scale simulation needs to be reduced. To achieve this goal, a multiscale decomposition method that accelerates the convergence of an iteratively-solved variable distribution in porous electrodes is proposed. The multiscale method combines the macroscopic method with pore-scale simulation by decomposing a variable distribution into the macroscopic component and local fluctuations. The decomposition removes the slowly converged, long wavelength components in an iteratively-solved variable distribution, thereby accelerating the convergence. In this research, to reduce the computational cost of multiphase pore-scale simulation, the multiscale method is applied to the electrolyte phase potential and oxygen concentration, both of which converge slowly and limit the overall computational efficiency. The results show that the multiscale method can substantially accelerate the convergence without sacrificing the accuracy. It is also found that the estimation of the effective transport property appearing in the volume-averaged part of the multiscale method influences the convergence rate of the multiscale method. With more accurate estimation of an effective transport property, the multiscale method is shown to work more effectively, especially for a thick porous electrode. Being an important parameter in the application to oxygen concentration, the effective oxygen diffusivity in pores is systematically investigated using po (open full item for complete abstract)

Committee: Seung Hyun Kim Dr. (Advisor); Sandip Mazumder Dr. (Committee Member); Jung Hyun Kim Dr. (Committee Member); Marcello Canova Dr. (Committee Member) Subjects: Mechanical Engineering

Master of Science in Engineering, University of Akron, 2016, Mechanical Engineering

A two-dimensional (2D) and a three-dimensional (3D) lattice Boltzmann (LB) and Cellular Automaton (CA) models were developed to study the buoyancy-induced mechanism of micro-segregation defects leading to formation of freckles in solidification of binary alloys. The channel-like defects that form during solidification have a significant influence on mechanical properties of the cast products. 3D simulations can provide a more profound insight about the mechanism of freckle formation, in comparison with the previous two-dimensional simulations. In the present work, lattice Boltzmann method (LBM) is used to solve for transport phenomena, while CA is employed to capture the solidification interface. Considering the local nature of LB and CA methods, combination of the two methods is a decent choice for large-scale parallelization. The model can be used to simulate dendritic growth, micro-segregation and freckle formation during solidification of metallic alloys in large-scale domains. The computer simulation results indicate that an instability in the fluid density leads to a competition among dendrites that retards growth of some dendrites and accelerates the others. The solute-rich channels which survive during this competition will solidify in the form of freckles.

Committee: Sergio Felicelli (Advisor); Mohsen Eshraghi (Committee Member) Subjects: Materials Science; Mechanical Engineering

Doctor of Philosophy (Ph.D.), University of Dayton, 2015, Electrical Engineering

Boundary extraction for object region segmentation is one of the most challenging tasks in image processing and computer vision areas. The complexity of large variations in the appearance of the object and the background in a typical image causes the performance degradation of existing segmentation algorithms. One of the goals of computer vision studies is to produce algorithms to segment object regions to produce accurate object boundaries that can be utilized in feature extraction and classification. This dissertation research considers the incorporation of prior knowledge of intensity/color of objects of interest within segmentation framework to enhance the performance of object region and boundary extraction of targets in unconstrained environments. The information about intensity/color of object of interest is taken from small patches as seeds that are fed to learn a neural network. The main challenge is accounting for the projection transformation between the limited amount of prior information and the appearance of the real object of interest in the testing data. We address this problem by the use of a Self-organizing Map (SOM) which is an unsupervised learning neural network. The segmentation process is achieved by the construction of a local fitted image level-set cost function, in which, the dynamic variable is a Best Matching Unit (BMU) coming from the SOM map. The proposed method is demonstrated on the PASCAL 2011 challenging dataset, in which, images contain objects with variations of illuminations, shadows, occlusions and clutter. In addition, our method is tested on different types of imagery including thermal, hyperspectral, and medical imagery. Metrics illustrate the effectiveness and accuracy of the proposed algorithm in improving the efficiency of boundary extraction and object region detection. In order to reduce computational time, a lattice Boltzmann Method (LBM) convergence criteria is used along with the proposed self-organized ac (open full item for complete abstract)

Committee: Vijayan Asari (Advisor); Raúl Ordóñez (Committee Member); Eric Balster (Committee Member); Muhammad Usman (Committee Member) Subjects: Computer Engineering; Electrical Engineering

Master of Science, University of Toledo, 2011, Mechanical Engineering

Lattice Boltzmann Method is evolving as a substitute to the prevalent and predominant CFD modeling especially in cases such as multiphase flows, porous media flows and micro flows. This study is aimed at developing simulation model for multiphase flows for practical applications such as cavitation in a journal bearing or lubrication of micro contact. The code is first validated against benchmark single phase flows like Poiseulle flow and flow over a cylinder. In the process, various boundary conditions like velocity, pressure, out-flow, no-slip and periodic boundary conditions are tested. Finally, the Shan-Chen model for multiphase physics, which is based on the interaction force between the fluid particles, is incorporated into the code and is validated.

Committee: Sorin Cioc PhD (Advisor) Subjects: Mechanical Engineering

Doctor of Philosophy, The Ohio State University, 2009, Chemical Engineering

Multiphase flows involving bubbles and droplets are ubiquitous in nature and in many industrial processes. Detailed information of such flows can be acquired from direct numerical simulations that directly resolve the flow on the bubble or droplet scale. In recent years, the lattice Boltzmann method (LBM) has emerged as a novel numerical method for multiphase flow simulation. While having many favorable features such as incorporation of physics on the more fundamental level and efficient algorithm for fast computation, the current multiphase LBM still faces challenges in issues such as numerical instability and narrow parameter window, which severely restrict its application in a broad range of real world engineering problems.This dissertation presents the development of a novel multiphase LBM which significantly expands the application of the method in various flow problems. Specifically, three techniques are developed to achieve enhanced performance in three aspects: First, new interaction potential functions are developed for multi-component LBM model to improve numerical stability at high density ratios between the liquid and gas phase. Second, an adaptive mesh refinement (AMR) scheme is developed to provide sufficient resolution of the gas-liquid interface. Third, the multi-relaxation time (MRT) scheme is incorporated into the interaction potential model to enhance the numerical stability at low viscosities. The above new techniques are presented in detail, and simulations are performed in both 2D and 3D to evaluate their performance. It is demonstrated that the new interaction potential model is able to raise the stable density ratio from below 50 to over 1000. The AMR can provide accurate predictions of the interface, while reduce the computation cost by about 50% in real computations. In addition, with the MRT algorithm the maximum Reynolds number in bubble simulations can be increased from 100 to about 1000. The performance of the newly developed LBM te (open full item for complete abstract)

Committee: Liang-Shih Fan PhD (Advisor); Chiu-Yen Kao PhD (Committee Member); Ly Lee PhD (Committee Member); Kurt Koelling PhD (Committee Member) Subjects: Chemical Engineering

MS, Kent State University, 2009, College of Arts and Sciences / Department of Computer Science

This thesis proposes an inherent parallel scheme for image segmentation of large data sets using the GPU. The method originates from an extended Lattice Boltzmann Model (LBM), and provides a new numerical solution for solving the level set equation. As a local, explicit and parallel scheme, this method lends itself to several favorable features: (1) Very easy to implement with the core program only requiring a few lines of code; (2) Implicit computation of curvatures; (3) Flexible control of generating smooth segmentation results; (4) Strong amenability to parallel computing, especially on the low-cost, powerful graphics hardware (GPU). The parallel computational scheme is also well suited for cluster computing, leading to solution for segmenting very large data sets, which cannot be accommodated by a single machine. While large data sets are typically found in various applications, current level set segmentation algorithms cannot easily operate on such data. This method proposes a new tool adopting distributed computing for the visualization community. Several examples are shown performing segmentation on the GPU and GPU cluster with satisfying results and performance.

Committee: Ye Zhao PhD (Advisor); Paul A Farrell PhD (Committee Member); Arden Ruttan PhD (Committee Member) Subjects: Computer Science

Doctor of Philosophy, Case Western Reserve University, 2006, Mechanical Engineering

A multiphase lattice Boltzmann method (MLBM) based on the HSD model has been adapted for the solution of multiphase fluid dynamics problem. The interactions between particles are expressed through a mean-field approximation and exclusion-volume effect. The behavior of interface is obtained as part of the solution of the lattice Boltzmann equations. No a priori assumptions and artificial treatment are made regarding the shape and dynamic roles of the interface. Interfacial tension dynamics is validated through a series of test running of three-dimensional wave dispersion. The MLBM is also extended to thermal multiphase LBM (TMLBM) which includes the effects of interfacial tension and its dependence on temperature by a hybrid scheme. The key point for this scheme is combining a micro-scale description of the flow with a macroscopic energy transport equation. Applying the TMLBM, a systematic investigation of fluid dynamics in a two-layer immiscible fluid system is undertaken starting with Rayleigh-Benard convection. A parametric study of the effects of thermally induced density change, buoyancy, surface tension variation with temperature on interface dynamics, flow regimes and heat transfer is presented. Further investigation of TMLBM is applied to a two-layer immiscible fluid system with density inversion in which density inverse assumption holds for the lower layer fluid. The evaluation of the effects of density distribution parameter, Rayleigh number, size aspect ratio and Marangoni number on convection flow and heat transfer is presented. Interaction between gravity-induced and vibration-induced thermal convection in a two-layer fluid system has also be studied by TMLBM. The vibrations considered correspond to sinusoidal translations of a rigid cavity at a fixed frequency and is parallel to temperature gradient. The ability of applied vibration to enhance the flow, heat transfer and interface distortion is investigated. Comparisons of two-phase fluid system with si (open full item for complete abstract)

Committee: J. Iwan D. Alexander (Advisor) Subjects: