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

This work is intended towards the development of distributed optimization methods for large-scale networked systems. The advancement in technological fields such as networking, communication and computing has facilitated the development of networks which are massively large-scale in nature. One of the important challenges in these networked systems is the evaluation of the optimal point of operation of the system. The problem is essentially challenging due to the high-dimensionality of the problem, distributed nature of resources, lack of global information and dynamic nature of operation of most of these systems. The inadequacies of the traditional centralized optimization techniques in addressing these issues have prompted the researchers to investigate distributed optimization techniques. This research work focuses on developing techniques to carry out the global optimization in a distributed fashion that explores the fundamental idea of decomposing the overall optimization problem into a number of sub-problems that utilize limited information exchanged over the network. Inspired by price-based mechanisms, the research develops two methods. First, a distributed optimization method consisting of dual decomposition and update of dual variables in the subgradient direction is developed for some different classes of resource allocation problems. Although this method is easy to implement, it has its own drawbacks.To address some of the drawbacks in distributed optimization, in this dissertation, a Newton based distributed interior point optimization method is developed. The proposed approach, which is iterative in nature, focuses on the generation of feasible solutions at each iteration and development of mechanisms that demand lesser communication. The convergence and rate of convergence of both the primal and the dual variables in the system is also analyzed using a benchmark Network Utility Maximization (NUM) problem followed by numerical simulation results. A comp (open full item for complete abstract)

Committee: Manish Kumar Ph.D. (Committee Chair); Sundararaman Anand Ph.D. (Committee Member); Kelly Cohen Ph.D. (Committee Member); David Thompson Ph.D. (Committee Member) Subjects: Mechanics

Master of Science, The Ohio State University, 2024, Computer Science and Engineering

In recent years, there has been significant research and development in Deep Learning (DL) due to its efficiency and extensive applicability across diverse domains, including Computer Vision and Large Language Models. However, the architecture of large Deep Learning models, containing dense layers, makes them compute and memory intensive. Distributed Deep Learning (Distributed DL) is the successful adaption to accelerate and enable training and inference for large-scale DL models, where it also deals with various parallel approaches, inference and training techniques, and communication optimization strategies to enhance performance. In this thesis, we focus on accelerated and memory-efficient techniques to optimize distributed training and inference. It is broadly categorized into three different approaches: 1. Inference for scaled images using quantization, achieving a speedup of 6.5x with integer- only quantization and 1.58x with half-precision, with less than 1% accuracy degradation. 2. MPI4DL: Distributed Deep Learning Parallelism framework encompassing various parallelism techniques with integral components such as Spatial Parallelism, Bidirectional Parallelism, and Hybrid Parallelism 3. Communication optimization by leveraging MCR- DL: A distributed module for DL frameworks with support for mixed-backend communication, dynamic selection of the optimal backend, and communication optimization enhancements such as compression and tensor fusion.

Committee: Prof. Dhabaleswar K. Panda (Advisor); Dr. Aamir Shafi (Committee Member); Prof. Hari Subramoni (Committee Member) Subjects: Computer Science

MS, University of Cincinnati, 2022, Engineering and Applied Science: Environmental Engineering

Managing modern microgrids for the 21st century will require looking beyond the intelligent control of complex microgrids and their coupling with centralized power grids. Incorporating multiple-objective optimization for sustainable decision-making purposes is a step toward providing end-users with reliable, cost-effective heating and power with minimum environmental impacts. A simulation model, based on nonlinear mathematical programming principles, is proposed to optimize a microgrid using economic and environmental impact objectives, using the University of Cincinnati's (UC) combined heat and power (CHP) microgrid as a case study. The economic objective focuses on minimizing operation costs, as subset of variable costs, that include electricity import and natural gas fuel costs to daily operate the microgrid's CHP process. The environmental impact objective concentrates on minimizing greenhouse gas (GHG) emissions (i.e., carbon dioxide, methane, and nitrous oxide) from the supply chain and microgrid system boundary on a cradle-to-grave lifecycle basis using a life cycle assessment (LCA) methodology. Three different analysis applications are investigated to optimize the simulation model of UC's CHP microgrid using the stated single and multiple objectives in MATLAB, harnessing MATLAB's Optimization Toolbox solvers to perform prescriptive analytics. The results of each analysis can assist operation managers with identifying economically or environmentally optimal operating conditions, commodity pricing thresholds and operational trends that can inform the development of optimal operation strategies, as well as Pareto optimal trade-off curves to reduce GHG emissions for the case study process. The fundamental conclusion taken from the multiple-objective optimization analysis is that managers can make sizable reductions (15-30% in the investigated examples) in greenhouse gas emissions while incurring smaller economic penalties (5-15% in the investigated examples) usi (open full item for complete abstract)

Committee: Margaret Kupferle Ph.D. (Committee Member); Stephen Thiel Ph.D. (Committee Member); Patrick Ray Ph.D. (Committee Member); Drew McAvoy Ph.D. (Committee Member) Subjects: Environmental Engineering

Master of Science, The Ohio State University, 2020, Mechanical Engineering

The emissions resulting from fossil fuel consumption across various industries has lead to detrimental effects on the planet's ecosystems. The commercial aircraft industry is attempting to reduce their emissions with the use of more fuel efficient propulsive architectures. The electrification of propulsive systems are considered to be a promising advancement for future aircraft. The architecture under investigation in this study is a turboelectric distributed propulsion system for a regional aircraft that assumes weight, drag, and design improvements that as consistent with a system being deployed in 2035. This system has the ability for hybridization via an onboard energy storage system, comprising of a battery pack, to further reduce the fuel consumption of this advanced architecture. A quasi-static model was developed with the primary objective of predicting the performance of the propulsion system over the course of a mission. Empirical equations, performance maps, and experimental data was used to calibrate the subsystems of this model. In order to expand the fuel reduction potential of the turboshaft engines, a variable speed free power turbine and a variable area nozzle were added to the engines as additional controls to the engine's primary throttle. The optimal control strategy of the turbine's speed and nozzle throat area that minimized the mission fuel consumption was determined and the associated fuel savings have concluded. The results of this study were compared to the same system where the two parameters were held constant at their design values. This analysis was completed for the hybrid and non-hybrid system to determine the optimal control strategy of the engine with and without a secondary source of power and to determine the variation of the control if a secondary source is present. It was concluded that the optimal engine control strategy can obtain a maximum savings of 1.55% over the same system neglecting the two additiona (open full item for complete abstract)

Committee: Meyer Benzakein Prof (Advisor); Marcello Canova Prof (Committee Member) Subjects: Aerospace Engineering; Mechanical Engineering

Doctor of Philosophy, The Ohio State University, 2019, Electrical and Computer Engineering

Large-scale machine learning has shown great promise for solving many practical applications. Such applications require massive training datasets and model parameters, and force practitioners to adopt distributed computing frameworks such as Hadoop and Spark to increase the learning speed. However, the speedup gain is far from ideal due to the latency incurred in waiting for a few slow or faulty processors, called straggler; to complete their tasks. To alleviate this problem, current frameworks such as Hadoop deploy various straggler detection techniques and usually replicate the straggling task on another available node, which creates a large computation overhead. In this dissertation, we focus on a new and more effective technique, called coded computation to deal with stragglers in the distributed computation problems. It creates and exploits coding redundancy in local computation to enable the final output to be recoverable from the results of partially finished workers, and can therefore alleviate the impact of straggling workers. However, we observe that current coded computation techniques are not suitable for large-scale machine learning application. The reason is that the input training data exhibit both extremely large-scale targeting data and a sparse structure. However, the existing coded computation schemes destroy the sparsity and creates large computation redundancy. Thus, while these schemes reduce delays due to the stragglers in the system, they create additional delays because they end up increasing the computational load on each machine. This fact motivates us to focus on designing more efficient coded computation scheme for machine learning applications. We begin by investigating the linear transformation problem. We analyze the minimum computation load (number of redundant computations in each worker) of any coded computation scheme for this problem, and construct a code we name diagonal code; that achieves the above minimum computation (open full item for complete abstract)

Committee: Ness Shroff (Advisor); Atilla Eryilmaz (Committee Member); Abhishek Gupta (Committee Member); Andrea Serrani (Committee Member) Subjects: Computer Science; Electrical Engineering

Doctor of Philosophy, The Ohio State University, 2019, Food, Agricultural and Biological Engineering

Thermal comfort is an important factor in designing high-quality buildings. The well-conditioned environment can keep occupants healthy and productive and ensure workplace safety. The heating, ventilation and air conditioning (HVAC) system plays an important role in providing and maintaining indoor thermal comfort for buildings. The faults in an HVAC system not only waste energy but also cause poor thermal comfort, building-related illnesses, or even safety accidents. This research adopted the model-based method to detect and diagnose the faults in a selected HVAC system. First, a simulation model of the case study building was created and validated based on both energy and thermal performance. Then, by comparing the indoor air temperatures between the simulation model and the real situation, three common types of faults in the HVAC system were detected for summer and winter, including: 1) control fault, 2) facility fault, and 3) design fault. In addition, the simulation fault was identified in the winter time. For each type of faults, the corresponding solutions were proposed, which will help building operators to locate and solve the faults quickly and accurately. As another important factor to designing high-quality buildings, building energy efficiency could reduce building's energy consumption and their environmental footprint. To lower buildings' significant energy consumption and high impacts on environmental sustainability, recent years have witnessed rapidly growing interests in efficient HVAC precooling control and optimization. However, due to the complex analytical modeling of building thermal transfer, rigorous mathematical optimization for HVAC precooling is highly challenging. As a result, progress on HVAC precooling optimization remains limited in the literature. One of the main contributions of this research is to overcome the aforementioned challenge and propose an accurate and tractable HVAC precooling optimization framework. The main results are (open full item for complete abstract)

Committee: Qian Chen (Advisor); Jia Liu (Committee Member); Sandra Metzler (Committee Member); Lingying Zhao (Committee Member) Subjects: Agricultural Engineering; Civil Engineering; Environmental Engineering; Sustainability

Master of Science, The Ohio State University, 2018, Electrical and Computer Engineering

This thesis addresses the coordinated operation of networked microgrids (MGs), distributed energy resources (DERs), and Volt-VAR control devices in the implementation of Volt-VAR optimization (VVO). Although our formulation is focused on implementing conservation voltage reduction (CVR), it can be extended for other VVO objectives e.g. losses minimization, peak demand shaving, or energy consumption reduction. We assume that the distribution network operator (DNO) has to make decisions anticipating the decisions of the MG operators. The hierarchy shown in this problem is related to a Stackelberg game. Hence, we formulate this problem as a bi-level optimization problem where the upper-level problem corresponds to the DNO and the lower-level problems correspond to each MG. The DNO as well as each MG are assumed to be independent entities with their individual objective functions. The objective of the DNO depends on the objective of the VVO objective. In particular, in the case of CVR, the objective is to minimize the load demand and losses of the distribution system. Conversely, we assume that the objective of the MG operators is to minimize the operation costs of dispatching DGs within the MG and buying/selling electricity from/to DNO i.e. an economic dispatch. The integration of DERs at the distribution level changes the response of the grid to a VVO strategy. We consider that DERs are located at the distribution network and within each microgrid. We study the four voltage-power control modes of DERs stated in the IEEE Std. 1547-2018, namely constant power factor, voltage-reactive power, active--reactive power, and constant reactive power. Finally, we validate our formulation in a modified IEEE 33-node test system. The effectiveness of CVR with the different voltage-power control modes of DERs is analyzed. The findings of this work are significant for the implementation of VVO in active distribution systems with networked microgrids.

Committee: Mahesh Illindala Ph.D. (Advisor); Jiankang Wang Ph.D. (Committee Member) Subjects: Electrical Engineering

Doctor of Philosophy, The Ohio State University, 2016, Electrical and Computer Engineering

Large-scale power outages are rare but extreme accidents. They are usually caused by severe weather events and overloading caused cascading failures. Nowadays, with climate change and ever-increasing load demand, power blackouts are happening more frequently. In order to ensure reliable power delivery to customers, resilient distribution systems are envisaged, because of their characteristics of high reliability, power quality, advanced protection, and optimal restoration. During extreme events, they can provide uninterruptible power supply to critical loads, quickly detect and accurately isolate fault areas, and reestablish with an optimal restoration plan. This dissertation first proposes to develop community microgrids within distribution systems by integrating local distributed energy resources (DERs) and neighboring load centers, especially critical loads. Community microgrids can be useful means of providing resilient electricity service by enabling sustainable operations and supporting critical loads in the event of power disruptions. When an extreme event happens, the distribution system can be seamlessly partitioned into several energy-independent community microgrids. Then, the important customers are supplied with uninterrupted power by local DERs. After fault isolation, distribution systems are restored by reconnecting these community microgrids. The DER selection for community microgrids is mainly determined by the levelized cost of energy (LCOE) based quantitative assessment in conjunction with the quality functional deployment (QFD) tool. Subsequently, the capacity planning of dispatchable generation units, like natural gas gensets and battery energy storage system (BESS), is elaborated. The goal of this sizing scheme is to keep adequate reserve margin to ride through unforeseen events, like uncertainties from loads and renewables, loss of generation, etc. This is because when community microgrids work in the islanded mode, the critical loads co (open full item for complete abstract)

Committee: Mahesh Illindala Dr. (Advisor); Jin Wang Dr. (Committee Member); Jiankang Wang Dr. (Committee Member) Subjects: Electrical Engineering

Doctor of Philosophy, The Ohio State University, 2016, Electrical and Computer Engineering

Researchers' prediction about the emergence of very small and very large computing devices is becoming true. Computer users create personal content from their mobile devices and these contents are processed/stored in the remote server. This mobile cloud computing architecture contains millions of smartphone devices as the edge and high-end servers as the cloud, in order to provide data services worldwide. Unlike data services in traditional architectures, data services in the mobile computing architecture is greatly constrained by by energy consumption. Data services running in the cloud consume a large amount of electricity that accounts for 4% of the global energy use. Data processing and transmission in mobiles devices, such as smartphones, quickly drain out their batteries. Therefore, energy is one of the most important criterion in the design of these systems. To address this problem, we need to build an energy modeling and management framework to profile, estimate and manage the energy consumption for data processing in the mobile cloud architecture. We first start with energy profiling of data processing in a single node. The study discovers that there exist possibilities of finding energy-efficient execution plans other than fast plans only. Based on the profile, we propose our online estimation tools for modeling and estimating energy consumption of relational data operations. Further, we provide power performance control for data processing. The control framework provide service level agreement guarantee while reducing the power consumption. The control-theoretic design provide system stability when facing unpredictable workloads. Using the modeling processing, we expand our research to optimize energy-related objectives, such as carbon footprint and cloud expense, in multiple nodes. We carefully study the processing of data in multiple nodes, and find that the processing (i.e., read/write) significantly affects the objectives wh (open full item for complete abstract)

Committee: Xiaorui Wang (Advisor); Fusun Ozguner (Committee Member); Christopher Stewart (Committee Member) Subjects: Computer Engineering; Computer Science

Master of Science in Mechanical Engineering (MSME), Wright State University, 2015, Mechanical Engineering

Surrogate models are commonly used in place of prohibitively expensive computational models to drive iterative procedures necessary for engineering design and analysis such as global optimization. Additionally, surrogate modeling has been applied to reliability based design optimization which constrains designs to those which provide a satisfactory reliability against failure considering system parameter uncertainties. Through surrogate modeling the analysis time is significantly reduced when the total number of evaluated samples upon which the final model is built is less than the number which would have otherwise been required using the expensive model directly with the analysis algorithm. Too few samples will provide an inaccurate approximation while too many will add redundant information to an already sufficiently accurate region. With the prediction error having an impact on the overall uncertainty present in the optimal solution, care must be taken to only evaluate samples which decrease solution uncertainty rather than prediction uncertainty over the entire design domain. This work proposes a numerical approach to the surrogate based optimization and reliability assessment problem using solution confidence as the primary algorithm termination criterion. The surrogate uncertainty information provided is used to construct multiple distributed surrogates which represent individual realizations of a lager surrogate population designated by the initial approximation. When globally optimized upon, these distributed surrogates yield a solution distribution quantifying the confidence one can have in the optimal solution based on current surrogate uncertainty. Furthermore, the solution distribution provides insight for the placement of supplemental sample evaluations when solution confidence is insufficient. Numerical case studies are presented for comparison of the proposed methodology with existing methods for surrogate based optimization, such as expected improvem (open full item for complete abstract)

Committee: Ha-Rok Bae Ph.D. (Advisor); Ahsan Mian Ph.D. (Committee Member); Zifeng Yang Ph.D. (Committee Member) Subjects: Aerospace Engineering; Mechanical Engineering

Doctor of Philosophy, Case Western Reserve University, 2014, Applied Mathematics

The mathematical modeling of brain energy metabolism in the literature has been approached in a spatially lumped framework, where the region of interest is represented in terms of well mixed compartments representing different cell types, extracellular space and capillary blood. These models shed some light on the brain metabolism, but they cannot account for some potentially important factors including, e.g., the locus of the synaptic activity in reference to capillaries, the effect of diffusion, pre- and postsynaptic neurons, and possible variations in mitochondrial density within the cells. In this thesis, we propose a novel multi-domain formalism to assemble a three dimensional distributed model of brain cellular metabolism, which can take into account some of the aforementioned factors. The model is governed by coupled reaction-diffusion equations in different cells and in the extracellular space, and it allows the inclusion of additional details, for example separate mitochondria for each cell type. This formalism allows to track the changes in metabolites and intermediates in mutually interacting domains without the need for detailed geometric modeling of the microscopic tissue structure. Acknowledging the complexity of the multidimensional model and the difficulty of finding suitable values of many parameters which specify it, we propose a way to reduce the complex model to a lower dimensional one, whose parameter values can be compared with literature values. More specifically, we derive a computational model for a brain sample of the size of a Krogh cylinder, with spatial distribution in tissue along the radial component. For this model, the different availability of oxygen and glucose away from the blood vessel could affect the cells' aerobic or anaerobic metabolism and trigger the uptake of lactate, highlighting the important role of diffusion. This spatially distributed model indicates that drawing conclusions about a complex spatially distributed sy (open full item for complete abstract)

Committee: Daniela Calvetti (Advisor); Erkki Somersalo (Advisor); David Gurarie (Committee Member); Joseph LaManna (Committee Member) Subjects: Applied Mathematics; Biology; Mathematics

Master of Science in Electrical Engineering, University of Toledo, 2013, College of Engineering

Processing large graph datasets represents an increasingly important area in computing research and applications. The size of many graph datasets has increased well beyond the processing capacity of a single computing node, thereby necessitating distributed approaches. As these datasets are processed over a distributed system of nodes, this leads to an inter-node communication cost problem (also known as inter-partition communication), negatively affecting the system performance. This research proposes new graph partitioning algorithms to minimize the inter-node communication by achieving a sufficiently balanced partition. Initially, an intuitive graph partitioning algorithm using Random Selection method coupled with Breadth First Search is developed for reducing inter-node communication by achieving a sufficiently balanced partition. Second, another graph partitioning algorithm is developed using Particle Swarm Optimization with Breadth First Search to reduce inter-node communication further. Simulation results demonstrate that the inter-node communication using PSO with BFS gives better results (reduction of approximately 6% to 10% more) compared to the RS method with BFS. However, both the algorithms minimize the inter-node communication efficiently in order to improve the performance of a distributed system.

Committee: Robert Green (Committee Chair); Vijay Devabhaktuni (Committee Co-Chair); William Acosta (Committee Member); Mansoor Alam (Committee Member) Subjects: Computer Engineering; Computer Science

Master of Science, University of Toledo, 2010, Electrical Engineering

Demand-Side Management (DSM) can be defined as the implementation of policies and measures to control, regulate, and reduce energy consumption. This document introduces home energy management through dynamic distributed resource management and optimized operation of household appliances in a DSM based simulation platform. The principal purpose of the simulation platform is to illustrate customer-driven DSM operation, and evaluate an estimate for home electricity consumption while minimizing the customer's cost. A heuristic optimization algorithm i.e. Binary Particle Swarm Optimization (BPSO) is used for the optimization of DSM operation in the platform. The platform also simulates the operation of household appliances as a Hybrid Renewable Energy System (HRES). The resource management technique is implemented using an optimization algorithm, i.e. Particle Swarm Optimization (PSO), which determines the distribution of energy obtained from various sources depending on the load. The validity of the platform is illustrated through an example case study for various household scenarios.

Committee: Dr. Lingfeng Wang PhD (Advisor); Dr. Vijay Devabhaktuni PhD (Advisor); Dr. Gursel Serpen PhD (Committee Member) Subjects: Computer Science; Electrical Engineering; Energy; Technology

Master of Science, The Ohio State University, 2010, Electrical and Computer Engineering

In this research, we investigate designing a smart lighting system. By extending and enhancing the centralized and distributed control algorithms we try to address the lighting control problem and design a robust smart lighting system. The purpose of implementing various control strategies is to come up with a strategy that optimizes the power consumed by the system and its robustness to the problems of cross-illumination, external light disturbances, delays in the communication network, and the network topologies used for communication. The functionalities we try to achieve are uniform lighting, user-defined preference based lighting, maximum lighting mode and energy savings lighting mode by utilizing daylight. We study the performance of each control strategy and present a comparative analysis between the best strategies.

Committee: Kevin Passino PhD (Advisor); Vadim Utkin PhD (Committee Member) Subjects: Electrical Engineering; Energy