Doctor of Philosophy, The Ohio State University, 2021, Physics

In this dissertation, we develop models for biological processes at several spatial and temporal scales. Codon optimization is a procedure in which genetic sequences are altered (without affecting protein identity) in an attempt/effort to increase protein expression. Our goals are to identify if a given single input sequence (for example, of a pathogenic protein of interest) has been codon optimized, and if so, to identify the target organism. In Chapter 2, we present multiple metrics that we have devised to identify codon optimization. Using information from publicly available databases, we define methods both on the scale of an entire sequence/genome and on the scale of individual codon differences between two matched sequences; these methods are shown to perform with high levels of success (>85%) on optimization routines centered around codon usage as well as maximization of the codon adaptive index. It is known experimentally that information about different external stimuli to cells are transmitted to the interior through the temporal patterns of transcription factors (TFs). In Chapter 3, we address the question of how genes can decode information contained in different aspects of the temporal patterns of single transcription factors and initiate downstream responses with specificity. We focus on amplitude and duration variation of the TF signals and construct a two-gene module that produces protein distribution that have minimal overlap for different input signals; it can distinguish between four types of signals reliably (>90% success) in the presence of intrinsic stochastic fluctuations inherent in biochemical reactions and extrinsic temporal fluctuations. We provide information-theoretic measures of the performance including capacity obtaining values consistent with experimental measurements on yeast. In Chapter 4, we define a model which explores an interesting observation: replication of influenza A virus in infected epithelial cells on a cell plate pro (open full item for complete abstract)

Committee: Ciriyam Jayaprakash (Advisor); Mohit Randeria (Committee Member); Ratnasingham Sooryakumar (Committee Member); Stuart Raby (Committee Member) Subjects: Bioinformatics; Biophysics; Physics

Master of Science, The Ohio State University, 2021, Public Health

The second largest Ebola Virus Disease outbreak in history was declared on August 1, 2018, by the Ministry of Health of The Democratic Republic of Congo. This epidemic affected the eastern most part of DRC, spanning the provinces of North Kivu, South Kivu, and Ituri. Lasting over 15 months, the outbreak resulted in 3470 cases (probable and confirmed) and 2287 deaths (CDC 2019). In collaboration with the University of Kinshasa, we obtained individual-level data spanning almost the entirety of the epidemic, presenting us with the unique opportunity of analyzing long-term Ebola epidemic dynamics and the effect of public health intervention. Exploratory analysis uncovered that this epidemic comprised many smaller, more isolated outbreaks, with pronounced spatial-temporal patterns. To reflect this, the data was split into three temporal segments. A statistical model for the data analysis was based on the new methodology known as Dynamic Survival Analysis (DSA), derived from the general stochastic model of spread of infection across a network of interconnected individuals (nodes). A key feature of the statistical model is that, unlike general stochastic network models, it does not require knowledge of the susceptible population size, the disease prevalence in the community, or the epidemic curve shape. The DSA-based model was applied to all three segments of the full epidemic, attempting to combine information across the three distinct waves of infections. The fitting of the model was based on individual data of infection and recovery times in each wave, and the estimated parameter values suggested that the epidemic was brought to an end because of increased effort in Ebola cases identification and prompt isolation. According to our findings, the time from infection onset to hospitalization was significantly decreased over the three waves, helping to contain the spread of disease.

Committee: Grzegorz Rempala (Advisor); Eben Kenah (Committee Member) Subjects: Biostatistics; Epidemiology

Doctor of Philosophy, The Ohio State University, 2014, Physics

In this dissertation we develop a stochastic multi-scale model of the early (the first 11 hours post infection) innate immune response of dendrite cells to viral infection. Using this model we identify a mechanism by which cellular noise can be transiently enhanced via extracellular communication. This full model of the early innate immune response of dendritic cells to a viral challenge is both complex (containing thousands of parameters) and computationally expensive, which makes sensitivity analysis intractable. In order to overcome these barriers we developed a coarse-grained model of the early innate immune response of dendritic cells to viral infection that both executes five to ten times faster than the full model (depending on the simulation time point of interest) and allows us to identify key model components (e.g. the distribution of times cells activate a gene). The coarse-grained model reproduces the mean, standard deviation, distributions, and correlations between the key observables in the full model. The utility of our approach in coarse-graining computationally expensive models in immunology stems from the tractability of more intelligent and comprehensive sensitivity analysis that may provide immunologists with insights that can facilitate the development of immunotherapies. The dendritic cells whose early response to a viral challenge we model, eventually have to communicate information about the virus to T-cells, which are part of the adaptive immune system. This motivated us to sharpen our understanding of the effect of noise on the transmission of information in extracellular signals to proteins characteristic of cellular response. We quantify information transmission in cellular systems using a well established quantity from information theory, mutual information, and establish its connection to error rates in cellular response. We demonstrate that (1) small changes in mutual information can lead to potentially important changes in cellular (open full item for complete abstract)

Committee: Ciriyam Jayaprakash (Advisor) Subjects: Biophysics; Cellular Biology; Immunology; Information Science; Molecular Biology; Physics

MS, University of Cincinnati, 2001, Engineering : Environmental Engineering

A framework is developed to quantify the susceptibility of drinking water distribution systems to intrusion events. The framework integrates infrastructure information, hydraulic modeling, and demographic data. These elements are managed within a geographic information system (GIS). Using criteria that reflect system pressure, hydraulic intrusion pathways, and contaminant sources, the framework identifies locations within the distribution system susceptible to intrusion events. Locations found to be susceptible to intrusions are prioritized for attention based on proximity to sensitive populations, such as young children and the elderly. The proposed method is demonstrated with a case study based on a real distribution system. The study area encompasses approximately 38 square miles, includes three service areas, contains over 280 miles of water main serving 18,900 connections with a total average demand of five to six million gallons per day. Susceptibility conditions exist at some locations throughout the system; however, only rarely do all three conditions coincide. Hence very few locations were deemed susceptible to intrusion events. The framework may support capital improvement programs, operational decisions, and distribution system sampling designs. Methods such as this have been suggested as part of a larger distribution system management approach to improve water quality and at the same time reduce regulatory sampling requirements.

Committee: Steven Buchberger (Advisor) Subjects:

Master of Science (MS), Wright State University, 2021, Computer Science

A great percentage of documents in scientific and engineering disciplines include mathematical formulas and/or algorithms. Exploring the mathematical formulas in the technical documents, we focused on the mathematical operations associations, their syntactical correctness, and the association of these components into attributed graphs and Stochastic Petri Nets (SPN). We also introduce a formal language to generate mathematical formulas and evaluate their syntactical correctness. The main contribution of this work focuses on the automatic segmentation of mathematical documents for the parsing and analysis of detected algorithmic components. To achieve this, we present a synergy of methods, such as string parsing according to mathematical rules, Formal Language Modeling, optical analysis of technical documents in forms of images, structural analysis of text in images, and graph and Stochastic Petri Net mapping. Finally, for the recognition of the algorithms, we enriched our rule based model with machine learning techniques to acquire better results.

Committee: Nikolaos G. Bourbakis Ph.D. (Advisor); Euripides G.M. Petrakis Ph.D. (Committee Member); Soon M. Chung Ph.D. (Committee Member) Subjects: Computer Science

Doctor of Philosophy, University of Toledo, 2021, Engineering

The importance of modeling the spreading processes through a population has led to the development of several mathematical models. A number of empirical studies have collected and analyzed data on contacts between individuals which also shows patterns of contacts in an ever-evolving network. Contagious processes on networks, such as the spread of disease through physical proximity or information diffusion over social media, are continuous-time processes that depend upon the pattern of interactions between the individuals in the network. Continuous-time stochastic epidemic models are a natural fit for modeling the dynamics of such processes. However, prior works on such continuous-time models do not consider the dynamics of the underlying interaction network which involves the addition and removal of edges over time. In this work, firstly, we investigate the effects of different contact network models with varying levels of complexity on the outcomes of simulated epidemics. Secondly, we incorporate continuous-time network dynamics (addition and removal of edges) into continuous-time epidemic simulations and propose two rejection-sampling based approaches coupled with the well-known Gillespie algorithm and Thinning algorithm that enables exact simulation of the continuous-time epidemic process. Thirdly, we propose a continuous-time contact network model which takes into account the duration of contacts for inference procedure.

Committee: Kevin Xu (Committee Chair); Devinder Kaur (Committee Member); Rong Liu (Committee Member); Ahmad Javaid (Committee Member); Defne Apul (Committee Member); Ezzatollah Salari (Committee Member) Subjects: Computer Science; Epidemiology; Mathematics; Sociology; Statistics

Doctor of Philosophy, Case Western Reserve University, 2019, Physics

In the last 20 years, there have been huge advances in the field of single-molecule force spectroscopy experimental techniques. With such advances, there is a plethora of raw single-molecule data. In those data, one of the most intriguing observations is the existence of biphasic bond lifetime behavior (catch bonds) in the protein-ligand interaction system under the application of the force. The first part of the thesis focuses on a theoretical way of understanding the origin of catch bonds in the cadherin-catenin-actin (CCA) and L-selectin-ligand systems. In both cases, we show that only a mode with two degrees of freedom is sufficient to reproduce catch-bond behavior. We also fit our model to the experimental bond lifetime data and learn that the value of the free parameters extracted from fitting corroborates with observed values from the protein structure. Furthermore, we also explore the non-Markovian behavior observed in the L-selectin-ligand system under the application of force at different ramping rates. In the case of varying the ramping rate, we learn that ramping behavior induces changes to the protein-ligand interface analogous to introducing mutations in the lectin domain of the L-selectin protein. In the second part of the thesis, we present the mapping between machine learning dynamics and non-equilibrium statistical mechanics. We focus on stochastic gradient descent (SGD) learning dynamics and map SGD to the Fokker-Planck dynamics. Using Fokker-Planck dynamics, we characterize the steady state probability distribution of the weights (the parameters of the machine learning algorithm). We learn that steady state probability distribution is non-Boltzmannian, which means that the SGD dynamics leads to a non-equilibrium steady state. By forcing thermalization, we also get a notion of temperature for the ML system and a weight update rule similar to natural gradient descent.

Committee: Michael Hinczewski (Committee Chair); Philip Taylor (Committee Member); Lydia Kisley (Committee Member); Alkan Kabakcioglu (Committee Member); Steven Izen (Committee Member) Subjects: Biophysics; Physics

Doctor of Philosophy, The Ohio State University, 2018, Physics

Respiratory epithelial cells are an important, initial target of the human influenza A virus during infection. An important question arises as to what factors determine whether the innate immune response of these cells is able to contain the infection. This is determined by the complex interplay of viral replication (where the virus hijacks the host cell machinery to replicate itself and ultimately infect other cells), the immune response (which detects, contains and eliminates the virus both through intracellular responses to limit viral replication and through intercellular communication by diffusing cytokines to trigger an antiviral response in uninfected cells) and viral antagonism (where the virus has evolved to counteract the host immune response). All these processes occur on different timescales from minutes to hours and on different length scales from subcellular to across a large population of cells. How this complex spatio-temporal dynamics determines the outcome of the competition between viral replication and immune response on the scale of days and at the level of tissues remains an open problem. Recent advances in biological techniques have allowed experimentalists to measure the ex vivo response of epithelial cells in detail both at the level of individual cells and at the level of a layer of epithelial cells. This thesis describes a successful attempt at developing a spatially explicit, stochastic model that incorporates the molecules and reactions known to play a key role in the contest between viral antagonism and immune response, retaining those details required to understand measured quantities. The model is validated using experiment data. Our simulation results for our spatial, stochastic agent-based model successfully reproduces measures of infection at the population level including the observed saturation in the number of virions as a function of time and size of viral inoculum, as well as the rapid rise in the percent of infected cells (open full item for complete abstract)

Committee: Ciriyam Jayaprakash (Advisor); Ralf Bundschuh (Committee Member); Comert Kural (Committee Member); Mohit Randeria (Committee Member) Subjects: Biophysics; Physics

Doctor of Philosophy, The Ohio State University, 2017, Electrical and Computer Engineering

This dissertation delivers new theoretical and computational frameworks for systematically modeling the aggregate dynamics of large-scale cyber-physical systems. Particularly focused on the hierarchical demand response management system in smart grid, we develop both control-oriented and optimization-oriented aggregate models for coordinating a large population of responsive loads, including both thermostatically controlled loads (TCLs) and deferrable loads. For control-oriented modeling, we develop a unified stochastic hybrid system (SHS) framework to derive the partial differential equations (PDE) that characterize the dynamical evolution of the load distribution. A deterministic hybrid system is proposed for modeling general individual responsive load. An SHS is proposed for modeling the population dynamics after accounting for different uncertainties.Existing literature usually derives the PDE based on the physical principles and specifies the associated boundary conditions heuristically. Our method is based on the adjoint relation between the differential operator associated with the PDE and the extended generator of the SHS process. In particular, it enables us to determine the PDE boundary conditions directly from the boundary condition satisfied by the SHS generator. The obtained PDE model systematically generalizes many existing aggregate models. It is fundamentally important for designing various aggregate control strategies. The optimization-oriented modeling is to characterize the constraint sets satisfied by the aggregate load power, also known as the aggregate flexibility. We show that the individual power flexibility can be modeled by a polytope and the aggregate flexibility is the Minkowski sum of the individual flexibility polytopes. Exact Minkowski sum of these polytopes is computationally prohibitive. Therefore, we develop optimization-based algorithms to approximate the aggregate flexibility. For TCLs, we propose to approximate indivi (open full item for complete abstract)

Committee: Wei Zhang (Advisor); Kevin Passino (Committee Chair); Abhishek Gupta (Committee Member) Subjects: Electrical Engineering

Doctor of Philosophy (PhD), Wright State University, 2016, Engineering PhD

The efficacy of nurses is impacted by their availability to their patients and the occurrence of both beneficial and detrimental interruptions. Using system engineering tools, this work addresses open challenges in (i) methods for effective matching of nurse availability to non-stationary stochastic demand, (ii) differentiation of beneficial and detrimental interruptions, and (iii) modeling of nurses' work with interruptions to provide an objective method of testing interruption interventions. First, we propose both qualitative and quantitative approaches to evaluate and then model the impact of resource scheduling on patient wait time in a Level I trauma center for a highly specialized nurse, the advanced practice provider (APP). Our findings revealed mismatches during evenings and weekends, which prompted the trauma manager to implement a schedule similar to one proposed by our model. This schedule reduced the patent wait time by over 73% at the cost of a 10.5% increase in APP hours. Applying a simulation-optimization approach, we obtained near-optimal schedules that reduced the wait time to over 78% with no increase in APP hours. Second, we proposed a novel patient-centered framework for classifying observed interruptions as detrimental or beneficial. We utilize a mixed-method approach that involved analysis of data collected via direct observation, surveys, and analysis of retrospective data for hands-free devices. With comfort and time as performance measures, we show that beneficial interruptions include those returning the nurse's focus to the patient, and detrimental interruptions those breaking the delivery of steady treatment or attention to the patient. Finally, using this differentiation, we provide a model of nurse's workflow with interruptions that captures the underlying stochastic, non-stationary nature of interruptions and their onset through actual observation of trauma center nurses. This model provides a deeper understanding of how inter (open full item for complete abstract)

Committee: Parikh Prakik Ph.D. (Advisor); Frank Ciarallo Ph.D. (Committee Member); Jennie Gallimore Ph.D. (Committee Member); Nan Kong Ph.D. (Committee Member); Mary McCarthy M.D. (Committee Member) Subjects: Engineering; Health Care; Industrial Engineering; Systems Science

Doctor of Philosophy (PhD), Ohio University, 2016, Mechanical and Systems Engineering (Engineering and Technology)

Global supply chain decisions, such as facility location, manufacturing system design, resource allocation, and distribution center location are long-term strategic decisions in nature and involve many uncertainties. Traditionally, a hierarchical approach is used design supply chain networks and manufacturing systems. First, the location of the facilities are determined, and then the manufacturing systems are designed at the selected locations. In this dissertation, a multi-stage supply chain network model is developed where locations of the plants and inner manufacturing system design are determined simultaneously for labor-intensive manufacturing companies. This dissertation aims to develop a decision making framework to integrate manufacturing systems and supply chain network design decisions considering optimal operator assignment and layered cellular manufacturing in mind. The industry studied is fashion jewelry manufacturing where labor cost is one of the major cost factors. Hence, optimizing the number of workers required for each operation, cell, and plant is critical for the cost efficiency of the entire supply chain. The optimal number of operators are determined for each manufacturing process, and then the optimal cell sizes are found for each manpower level using a heuristic procedure. The optimal number of manufacturing cells required to cover the uncertain demand is determined with mathematical modeling, and the designed layered cellular manufacturing systems for manufacturing stages are evaluated using Arena simulation models. The results of these models and methods are used as inputs while finding the optimal locations of the plants and allocating the optimal number of cells, workers, and machines for each selected plant. Different supply chain design alternatives considering various factors such as the shortest lead times, minimum capacity allocations, and multiple shifts are also studied.

Committee: Gursel A. Suer Ph.D. (Advisor) Subjects: Industrial Engineering; Operations Research

PhD, University of Cincinnati, 0, Engineering and Applied Science: Environmental Engineering

In essence, modern drinking water distribution systems (DWDSs) exist to continuously satisfy the demand of their users while complying with water quality regulations. It stands to reason that the tasks of quantifying, estimating, and forecasting water consumption are critical components of resource management, planning and operation in the urban water industry. Yet, due to the complex stochastic nature of water demands, such important tasks are typically performed in an oversimplified deterministic manner which at best produces conservative results. Of critical inter- est, therefore, is the adoption of quantitative methods and technologies for accurately estimating and forecasting water consumption. The concomitant benefits may include the reduction of energy costs, residence times, pressure, and leakage in the DWDS through the optimal operation of pumps, reservoirs, and supply. Computational models of DWDSs have widely been developed by water utilities and researchers and applied mainly in design and offline analyses. It is clear that the industry and research com- munity recognize the usefulness of hydraulic models as tools to analyze the complex interaction among the generally massive number of system's components. Significant efforts are sometimes devoted to the refinement of the models to ensure their parameters reflect reality as close as possi- ble. Curiously, however, the parameters that most greatly influence the model's behavior, i.e., the water demands, are normally overlooked. It is not uncommon to assume a single arbitrary daily pattern for the totality of the nodes in a network model. This research considers that a more valid approach is to combine a reliable hydraulic model of a DWDS with realistic stochastic models of water use developed from fine-resolution consumption data. The intent is to abandon the time-pattern paradigm and take benefit of the opportunity to ac- cess large volumes of automatic meter reading (AMR) data at the level of the (open full item for complete abstract)

Committee: James Uber Ph.D. (Committee Chair); Robert Janke M.S. (Committee Member); David Kelton Ph.D. (Committee Member); Dominic Boccelli Ph.D. (Committee Member); Steven Buchberger Ph.D. (Committee Member) Subjects: Environmental Engineering

PhD, University of Cincinnati, 2007, Engineering : Electrical Engineering

Biological pathways provide a comprehensive view of a biological phenomenon, in the form of a network of inter-related reactions or processes. Modeling the biochemical reactions helps in studying and analyzing a biological pathway. This is done through parameter extraction and development of mathematical models of the biological systems. The importance of such modeling lies in the ability to easily perform mathematical mutations and optimizations to achieve a specific result, which can then be duplicated in the laboratory. The ability to control the outputs of biological reactions increases the possibilities for new applications, such as developing crops resistant to infection and bio-engineering drugs for diseases like Hepatitis and HIV-AIDS. Studying random mutations through practical experimentation is time consuming and expensive. Mathematical modeling definitely provides an affordable and convenient virtual experimental platform. However current methods are limited, as they produce results that may be difficult to be reproduced by biologists. The typical results do not address the practical constraints and feasibilities of the proposed mathematical mutation. Hence, there is a definite need for efficient algorithms and software, which not only help study the effect of mutations in a mathematical setting, but also provide practical methods to control biological pathways in a laboratory setting. In this dissertation, we develop an algorithm named Box which addresses this issue. The Box algorithm encompasses all the steps needed, from modeling a pathway to producing the biological controls needed to achieve desired mutations. The Box algorithm can be explained in terms of six logical steps: bio-modeling development language, bio-control database integration, sensitivity analysis, bio-rules formation, output optimization and comparison. The first step, the bio-modeling development language BMDL, is a new type of representation for a biological model. It is a weighte (open full item for complete abstract)

Committee: Dr. Carla Purdy (Advisor) Subjects:

PhD, University of Cincinnati, 2006, Engineering : Environmental Engineering

A new computer model, ADRNET, is developed to predict the spatial and temporal distribution of disinfectant in a pipe network, considering stochastic water demands and unsteady mass dispersion. An Eulerian-Lagrangian scheme is combined with a numerical Green's Function technique to solve the advection-dispersion-reaction equation efficiently in network conditions. In a comparison with the industry standard advection-reaction water quality model (EPANET), ADRNET exhibits better agreement with field observations at locations where laminar flow is prevalent. Implementation of the ADRNET model is preceded by three ancillary studies. The first study investigates the effect of temporal averaging on stochastic pipe flows to identify the appropriate time scales for water quality modeling of distribution networks. For this purpose, a non-homogeneous Poison Rectangular Pulse (PRP) process is utilized to simulate high resolution residential water demands in a distribution network. Two water demand models are successfully established to demonstrate variability and frequency of regimes for PRP flows as function of time scale. The results show that the variance of time-averaged PRP random flows is inversely proportional to the time scale; the frequency of flow regimes depends on both the time scale and the mean of the random flows. The second study investigates the conditions under which mass dispersion is important in pipe networks through comparison of numerical simulations with and without dispersive transport. The results show that mass dispersion is always important in laminar flow zones, and the importance of dispersion increases with increasing pipe diameter but decreases with increasing of reaction rate coefficient. Finally, the effect of temporal scale on unsteady dispersion is studied through both theoretical analyses with periodic binary flow pulses and numerical simulation with PRP random laminar flows. For small diameter tubes, unsteady dispersion decreases with incr (open full item for complete abstract)

Committee: Dr. Steven Buchberger (Advisor) Subjects:

Master of Science, The Ohio State University, 2011, Electrical and Computer Engineering

In the last couple of decades, various industries have either reduced or eliminated the power of the Government, creating more competition within the industry. In such a deregulated economy, load forecasting plays a very vital role in the planning and operation of electric utilities. Load forecasting finds in use in the sales, marketing, planning and manufacturing divisions of every industry. Literature review indicates the need to consider several factors such as time of a day, weather data and possible customer classes for effective one-step ahead and day ahead load forecasting on a feeder. This work focuses on analyzing the effects of temperature and humidity on different customer classes namely: Residential, Commercial and Industrial loads. Having analyzed their effects, a nominal load sequence from the weather sensitive load sequence was constructed. One-step-ahead and day ahead load forecasting of these sequences were then carried out using stochastic modeling.

Committee: Ali Keyhani PhD (Advisor); Mahesh Illindala PhD (Committee Member) Subjects: Electrical Engineering

Master of Science, The Ohio State University, 2011, Industrial and Systems Engineering

The paradigm of cloud computing has started a new era of service computing. While there are many research efforts on developing enabling technologies for cloud computing, few focuses on how to strategically set price and capacity and what key components are leading to success in this emerging market. In this thesis, we present quantitative modeling and optimization approaches for assisting such decisions in cloud computing services. We first show that learning curve models help in understanding the potential market of cloud services and explain quantitatively why cloud computing is most attractive to small and medium businesses. We then present Single Instance model to depict a particular type of cloud networks and aid in resource provisioning for the cloud service providers. We further present Multiple Instance model to depict any generic cloud network. We map the resource provisioning problem to Kelly's Loss Network and propose Genetic Algorithm to solve it. The approach provides the cloud service provider a quantitative framework to obtain management solutions and to learn and react to the critical parameters in the operation management process by gaining useful business insights.

Committee: Dr. Cathy Xia (Advisor); Dr. Theodore Allen (Committee Member) Subjects: Operations Research

Doctor of Philosophy, The Ohio State University, 2004, Industrial and Systems Engineering

Before production, experience and deterministic simulations are used to design the dies for the chosen part. Experience is mainly used for process design and input parameter selection. During production few parts are manually inspected to see if they are within specifications and have no defects. Control charts are also used by operators to make process changes based on experience if parts are out of tolerance. Conventionally, deterministic finite element methods (FEM) are used for die design in production systems and experience is mainly used for process design. However in reality, process conditions vary with time and hence with the same nominal process settings it is likely that the final forged part geometry varies with time. The simulations will give us solutions which may not match with the actual part geometries. In order to solve this problem, an approach to integrating numerical and response surface models for robust process design is described in this research. The integrated model is a virtual production model (VPM) which integrates numerical modeling techniques with response surface methodology (RSM). The VPM is driven by a combination of FEM, RSM and stochastic simulations. The production system is modeled as a series of processes with input parameters having distributions and output attributes which vary with time. The tasks to develop the model include deterministic FEM simulations, development of response surfaces of attributes and stochastic simulations. This model will be used to design the existing production process by designing the best input parameter settings. Its main aim is reducing process variability, reducing defect rates and improving process capability by robust process design. In this dissertation two approaches for integrating numerical models and response surface models have been described. The first approach integrates well established empirical relations, Taguchi methods of experimental design and FEM to arrive at robust roll pass (open full item for complete abstract)

Committee: Rajiv Shivpuri (Advisor) Subjects:

Doctor of Philosophy, The Ohio State University, 2002, Industrial and Systems Engineering

This dissertation contains the first proof of convergence of a genetic algorithm in the context of stochastic optimization. The class of stochastic optimization problems includes formulations in which the objective is an expected value, which can be evaluated using Monte Carlo methods. Growing computer power combined with methods presented here and elsewhere makes feasible the solution of many stochastic optimization problems with applications ranging from process design to facility location. The dissertation also describes the proposed stochastic optimization method that combines a sequential ranking and selection procedure with an elitist genetic algorithm. A batching procedure is included to assure that batch means of solutions achieve approximate normality. The proposed method is proven under the normality assumption to converge in the long run to identify and maintain solutions with objective values within an acceptable difference, D, from the global optimal solution with probability greater than an acceptable probability, P*. Computational results illustrate that the proposed algorithm achieves promising performance compared with alternatives for a variety of problems with minimal changes. The first application is on the stochastic optimization for “robust” engineering process design decisions making. By robust we mean designs that maximize the expected utility taking into account variation of “noise factors”. A methodology for robust process design is presented based on direct minimization of the expected loss in some cases using the proposed optimization heuristics. The proposed methods are compared with alternatives including methods based on Taguchi's signal-to-noise ratios. Several formulations of the loss are explored. The method is illustrated through its application to the design of robotic gas metal arc-welding parameter settings. The second application is a simulation optimization method applied to decision making about where to locate facilities and (open full item for complete abstract)

Committee: Theodore Allen (Advisor) Subjects:

Master of Science (MS), Ohio University, 2004, Chemical Engineering (Engineering)

The two-dimensional (2-D) stochastic model, which describes the balance of two processes: corrosion (leading to metal loss) and precipitation (leading to metal protection), is able to predict localized corrosion, which is the most serious type of corrosion attack found in practice. The model uses uniform corrosion rate and surface-scaling tendency predicted by a 1-D mechanistic corrosion model as the inputs and can predict the possibility of localized corrosion as a function of primitive parameters such as temperature, pH, partial pressure of CO2, velocity, etc. The maximum penetration rate as well as uniform corrosion rate can be predicted and used to describe the severity of the localized attack.

Committee: Srdjan Nesic (Advisor) Subjects:

Doctor of Philosophy, Case Western Reserve University, 1990, Civil Engineering

Fatigue of metals has been recognized as an important cause of failure of engineering structures. The experiments show that the fatigue life of real mechanical components is characteristically random. The random nature of the fatigue process is most obvious if a structure is subjected to time-varying random loading. This work develops a stochastic phenomenological model for crack growth which incorporates the effects of material inhomogeneity and random loading as well as including deterministic models which try to explain experimentally observed behavior, thus removing a majority of the shortcomings in existing stochastic models.

Committee: Fred Moses (Advisor) Subjects: Engineering, Civil