Organizational Unit:
Daniel Guggenheim School of Aerospace Engineering

Permanent Link
https://hdl.handle.net/1853/70742
Research Organization Registry ID
Description
Previous Names
Parent Organization
Parent Organization
Organizational Unit
Includes Organization(s)
ArchiveSpace Name Record
https://finding-aids.library.gatech.edu/agents/corporate_entities/106

Filters

992
992
992
992
Reset filters

Settings
End date: 2021 × search.filters.applied.f.isSeriesOfPublicationTitle: Doctor of Philosophy with a Major in Aerospace Engineering ×

Publication Search Results

Now showing 1 - 10 of 992
Thumbnail Image
Item

DEEP-LEARNING-ENHANCED MULTIPHYSICS FLOW COMPUTATIONS FOR PROPULSION APPLICATIONS

2021-12-17 , Milan, Petro Junior

Numerical simulation is a critical part of research into and development of engineering systems. Engineers often use simulation to explore design settings both analytically and numerically before prototypes are built and tested. Even with the most advanced high performance computing facility, however, high-fidelity numerical simulations are extremely costly in time and resources. For example, a survey of the design parameter space for a single-element injector for a propulsion application (such as the RD-170 rocket engine) using the large eddy simulation technique may require several tens of millions of CPU-hours on a major computer cluster. This is because the flowfields can only be fully characterized by resolving a multitude of strongly coupled fluid dynamic, thermodynamic, transport, multiphase, and combustion processes. The cost is further increased by grid resolution requirements and by the effects of turbulence and high-pressure phenomena, which require treatment of real-fluid physics at supercritical conditions. If such models are used for statistical analysis or design optimization, the total computation time and resource requirements may render the work unfeasible. Recent developments in deep learning techniques offer the possibility of significant advances in dealing with these challenges and significant shortening of the time-to-solution. The general scope of this thesis research is to set the foundations for new paradigms in modeling, simulation, and design by applying deep learning techniques to recent developments in computational science. More specifically, the research aims at developing an integrated suite of data-driven surrogate modeling approaches and software for large-scale simulation problems. The techniques to be put into practice include: (1) deep neural networks for function approximation and solver acceleration, (2) deep autoencoders for nonlinear dimensionality reduction, and (3) spatiotemporal emulators based on multi-level neural networks for simulator approximation and rapid exploration of design spaces. A hierarchy of benchmark cases has been studied to generate databases to enable and support the development and verification of the proposed approaches. Emphasis is placed on canonical examples, as well as on engineering problems for aerospace and automotive applications, including supercritical turbulent flows in a rocket-engine swirl injector, and multiphase cavitating flows in a diesel engine injector.

View
Thumbnail Image
Item

MODEL-BASED APPROACH TO THE UTILIZATION OF HETEROGENEOUS NON-OVERLAPPING DATA IN THE OPTIMIZATION OF COMPLEX AIRPORT SYSTEMS

2021-12-15 , Nikoue, Harold

Simulation and optimization have been widely used in air transportation, particularly when it comes to determining how flight operations might evolve. However, with regards to passengers and the services provided to them, this is not the case in large part because the data required for such analysis is harder to collect, requiring the timely use of surveys and significant human labor. The ubiquity of always--connected smart devices and the rise of inexpensive smart devices has made it possible to continuously collect passenger information for passenger-centric solutions such as the automatic mitigation of passenger traffic. Using these devices, it is possible to capture dwell times, transit times, and delays directly from the customers. The data; however, is often sparse and heterogeneous, both spatially and temporally. For instance, the observations come at different times and have different levels of accuracy depending on the location, making it challenging to develop a precise network model of airport operations. The objective of this research is to provide online methods to sequentially correct the estimates of the dynamics of a system of queues despite noisy, quickly changing, and incomplete information. First, a sequential change point detection scheme based on a generalized likelihood ratio test is developed to detect a change in the dynamics of a single queue by using a combination of waiting times, time spent in queue, and queue-length measurements. A trade-off is made between the accuracy of the tests, the speed of the tests, the costs of the tests, and the value of utilizing the observations jointly or separately. The contribution is a robust detection methodology that quickly detects a change in queue dynamics from correlated measurements. In the second part of the work, a model-based estimation tool is developed to update the service rate distribution for a single queue from sparse and noisy airport operations data. Model Reference Adaptive Sampling is used in-the-loop to update a generalized gamma distribution for the service rates within a simulation of the queue at an airport’s immigration center. The contribution is a model predictive tool to optimize the service rates based on waiting time information. The two frameworks allow for the analysis of heterogeneous passenger data sources to enable the tactical mitigation of airport passenger traffic delays.

View
Thumbnail Image
Item

Control and game-theoretic methods for secure cyber-physical-human systems

2021-12-13 , Kanellopoulos, Aris

This work focuses on systems comprising tightly interconnected physical and digital components. Those, aptly named, cyber-physical systems will be the core of the Fourth Industrial Revolution. Thus, cyber-physical systems will be called upon to interact with humans, either in a cooperative fashion, or as adversaries to malicious human agents that will seek to corrupt their operation. In this work, we will present methods that enable an autonomous system to operate safely among human agents and to gain an advantage in cyber-physical security scenarios by employing tools from control, game and learning theories. Our work revolves around three main axes: unpredictability-based defense, operation among agents with bounded rationality and verification of safety properties for autonomous systems. In taking advantage of the complex nature of cyber-physical systems, our unpredictability-based defense work will focus both on attacks on actuating and sensing components, which will be addressed via a novel switching-based Moving Target Defense framework, and on Denial-of-Service attacks on the underlying network via a zero-sum game exploiting redundant communication channels. Subsequently, we will take a more abstract view of complex system security by exploring the principles of bounded rationality. We will show how attackers of bounded rationality can coordinate in inducing erroneous decisions to a system while they remain stealthy. Methods of cognitive hierarchy will be employed for decision prediction, while closed form solutions of the optimization problem and the conditions of convergence to the Nash equilibrium will be investigated. The principles of bounded rationality will be brought to control systems via the use of policy iteration algorithms, enabling data-driven attack prediction in a more realistic fashion than what can be offered by game equilibrium solutions. The issue of intelligence in security scenarios will be further considered via concepts of learning manipulation through a proposed framework where bounded rationality is understood as a hierarchy in learning, rather than optimizing, capability. This viewpoint will allow us to propose methods of exploiting the learning process of an imperfect opponent in order to affect their cognitive state via the use of tools from optimal control theory. Finally, in the context of safety, we will explore verification and compositionality properties of linear systems that are designed to be added to a cascade network of similar systems. To obfuscate the need for knowledge of the system's dynamics, we will state decentralized conditions that guarantee a specific dissipativity properties for the system, which are shown to be solved by reinforcement learning techniques. Subsequently, we will propose a framework that employs a hierarchical solution of temporal logic specifications and reinforcement learning problems for optimal tracking.

View
Thumbnail Image
Item

A Framework for Selecting Multi-Attribute Optimal Renewable Energy Driven Desalination Architectures

2021-12-08 , Brooks, Joshua Daniel

The World is drifting into a near-future of unprecedented water resource management challenges. Growing water demand is projected to be met by limited, unpredictable, and in many locations shifting freshwater resources. Near-future populations are projected to face widespread water stress, most immediately and severely encountered in the form of hydrological drought. Desalination systems offer resilience in the form of additional water supplies which are insensitive to drought. However, desalination systems are currently limited by their costs, water inefficiency, greenhouse gas (GHG) emissions, energy requirements, and quality and environmental impacts, and are thus not used on the wider scale necessary to appropriately mitigate the risk of projected water stress. This work aimed to help overcome desalination’s core barriers to adoption by introducing an original framework for the quantitative performance-based selection of multi-attribute optimal desalination architectures. This framework enables an expansive desalination architecture design space exploration across both desalting and energy subsystems. Desalination architectures were valuated by mapping their barriers to adoption to their quantifiable performance attributes: cost, GHG emissions, and freshwater recovery. A superstructure flowsheet model was constructed to include reverse osmosis, multi-stage flash, multi-effect distillation, and thermal vapor compression desalting technologies. This model was situated inside of an optimization routine and used to both explore an unprecedented desalting design space and to identify designs which often outperformed those identified in similar efforts. An individual desalination architecture alternative in this work is defined as any desalting subsystem alternative connected to an optimal energy subsystem. An energy system model was therefore constructed to include photovoltaic arrays, wind energy converters, concentrated solar power plants, battery energy storage, and a connection to a conventional electrical grid and steam generator. Incorporating renewable energy sources (RES) enabled the identification of energy subsystems which lowered cost, GHG emissions, and water consumption compared to traditional grid and dedicated steam generation systems. High speed metamodels were successfully used to represent the full energy system model in order to make desalination architecture evaluation and optimization exercises computationally tenable. The full desalination architecture evaluation environment, consisting of the integrated desalting and energy subsystem models, was situated within an optimization routine. Cost-driven optimization exercises consistently identified RES-driven desalination alternatives which outperformed conventional alternatives identified in similar efforts. In addition, multiple cases were demonstrated wherein the simultaneous consideration of both energy and desalting subsystem performance in desalination architecture optimization exercises identified alternatives which were uncompetitive using the traditional selection approach. This thesis effort provides decision makers with a quantitative performance-based, tailorable framework for rapidly exploring the desalination architecture design space and selecting multi-attribute optimal systems regarding their unique preferences and system requirements. The constructed framework is flexible enough to accommodate different optimization and decision making techniques, and approaches are discussed for incorporating additional technologies into the desalting and energy subsystem modeling environments. This quantitative architecture selection framework, specifically its capability in allowing novel architectural and conceptual trades, is the core outcome of this work.

View
Thumbnail Image
Item

A Micromechanically-Informed Model of Thermal Spallation with Application to Propulsive Landing

2021-12-15 , Hart, Kenneth Arthur

During the propulsive landing of spacecraft, the retrorocket exhaust plume introduces the landing site surface to significant pressure and heating. Landing site materials include concrete on Earth and bedrock on other bodies, two highly brittle materials. During a landing event, defects and voids in the material grow due to thermal expansion and coalesce, causing the surface to disaggregate or spall. After a spall is freed from the surface, the material beneath it is exposed to the pressure and heat load until it spalls, continuing the cycle until engine shutdown. Spalls and debris entrained in the exhaust plume risk damaging the lander or nearby assets- a risk that increases for larger engines. The purpose of this work is to develop a micromechanically-informed model of thermal spallation to improve understanding of this process, in the context of propulsive landing. A preliminary simulation of landing site spallation, utilizing an empirical thermal spallation model, indicates that spallation may occur for human-scale Mars landers. This model, however, was developed for drilling through granite, which has a fundamentally different microstructure compared to typical landing sites, necessitating a more general approach. To that end, highly-detailed simulations of thermomechanical loading, applied to representative microstructures, inform a functional relationship between applied heat flux and spallation rate. These representative microstructures can be generated using an algorithm that has been validated for a wide variety of materials, including basalt from Gusev Crater, Mars.

View
Thumbnail Image
Item

Development of a Sonic Sensor for Aircraft Applications

2021-12-14 , Carroll, Jonathan D.

The field of aeroacoustics has been an area of constant research over the past six decades. Acoustic waves have some special characteristics that allow for heating, cooling, and even active flow control over airfoil shapes using synthetic jets and other methods. They can also be used to measure properties of the flow over an aircraft, including the free-stream pressure ratio, density ratio, and total temperature. The current measurement techniques to obtain these parameters applied to aircraft require a specific probe. It is desired to apply knowledge of acoustics to develop an aircraft sensor that can measure multiple flow properties with minimal impact to the flow field. Adding a sensor that can read total temperature, static temperature, airspeed, and angle of attack will have the added benefit of reducing the number of sensors sticking into the flow and may result in a reduction in failure mode analysis due to the minimization of the number of sensors on the aircraft. This work explores the applicability of sonic anemometry to aircraft for high subsonic and sonic speeds. A computational simulation is developed as a validation of the concept and low speed experiments are shown to validate the theory. This effort identifies the underlying issues associated with applying sonic anemometry to high-speed flows and provides methods to overcome them. This work investigates the use of phased array technology to increase the accuracy and applicability at the higher speeds and smaller footprints (lighter and fewer systems). Phased arrays use the constructive and destructive interference to boost and direct the desired signal, in this case, acoustic waves. These acoustic waves have been shown to provide haptic feedback and levitate small particles utilizing a relatively inexpensive ultrasonic phased array system. It is shown that the ultrasonic phased array overcomes the hydrodynamic noise to produce a strong signal for use in the calculation of the flow parameters up to the maximum speed tested. It is also shown that the signal is strong enough to produce consistent time delay estimations, via cross-correlation, with a 0.05 second sample time to integrate into modern air data systems.

View
Thumbnail Image
Item

RETROSPECTIVE AND EXPLORATORY ANALYSES FOR ENHANCING THE SAFETY OF ROTORCRAFT OPERATIONS

2021-12-13 , Chin, Hsiang-Jui

From recent safety reports, the accident rates associated with helicopter operations have reached a plateau and even have an increasing trend. More attention needs to be directed to this domain, and it was suggested to expand the use of flight data recorders on board for monitoring the operation. With the expected growth of flight data records in the coming years, it is essential to conduct analyses and provide the findings to the operator for risk mitigation. In this thesis, a retrospective analysis is proposed to detect potential anomalies in the fight data for rotorcraft operations. In the study, an algorithm is developed to detect the phases of flight for segmenting the flights into homogeneous entities. The anomaly detection is then performed on the flight segments within the same flight phases, and it is implemented through a sequential approach. Aside from the retrospective analysis, the exploratory analysis aims to efficiently find the safety envelope and predict the recovery actions for a hazardous event. To facilitate the exploration of the corresponding operational space, we provide a framework consisting of surrogate modeling and the design of experiments for tackling the tasks. In the study, the autorotation, a maneuver used to land the vehicle under power loss, is treated as a used case to test and validate the proposed framework.

View
Thumbnail Image
Item

PERFORMANCE ADVANTAGES AND RESONANCE ANALYSIS OF A VARIABLE SPEED ROTOR USING GEOMETRICALLY EXACT BEAM FORMULATIONS

2021-12-15 , Chandrasekaran, Ruthvik

The efficiency and operating envelope of a rotorcraft is constrained by the speed of the rotor. Most of the helicopters operate at a constant rotor speed. Varying the speed of the rotor based on the operating condition could significantly improve the rotor performance. In this study, a hingeless rotor model with elastic blades is built in Dymore to study various aspects of Variable Speed Rotor (VSR) technology. The rotor blades are modeled as one-dimensional beams using state of the art beam theory known as geometrically exact beam theory. An unsteady aerodynamics model with dynamic stall and finite-state dynamic inflow is used to obtain the aerodynamic loads acting on the rotor. A wind tunnel trim procedure is adopted to trim the rotor for a given thrust, roll and pitch moment. An auto-pilot controller is used to trim the rotor during time marching based on the wind tunnel trim values. The rotor model and trim procedure is validated using results from literature. The power savings that can be achieved at various advance ratios by varying the speed of the rotor is evaluated. However, varying the rotor speed leads to vibration issues as the rotor passes through the resonance regions. In this region, the rotor blade's natural frequency coincides with the multiple of rotor's operating frequency. This leads to an increase in vibratory loads. All the resonance points are identified from the fan plot of the rotor blade. It is observed that the lead-lag moment at the blade root increases significantly compared to the nominal value during lag resonance. It is also observed that the flap and torsional moments increase during lag resonance. Transition dynamics of the rotor blade for different operating conditions were analyzed. Load reduction studies during resonance were carried out by changing the transition times and blade properties. The longer the rotor took to traverse a resonance region, greater were the resonance loads. Increasing the structural damping was a very effective way of mitigating resonance loads. An active system called as the Anti-Resonance System (ARS) was conceptualized and modeled in Dymore. The ARS system was able to effectively remove the resonance loads.

View
Thumbnail Image
Item

The Improvement of Multi-Satellite Orbit Determination Through the Incorporation of Intersatellite Ranging Observations

2021-12-14 , Davis, Byron Taylor

For many satellite remote sensing and communications missions, particularly those involving a formation or constellation of satellites, having precise knowledge of the satellites’ positions in both an absolute and relative sense is essential. However, the capabilities of Global Navigation Satellite Systems (GNSS)-based precise orbit determination (POD) alone may not be enough to fulfill the mission’s requirements. This thesis examines potential gains to POD when additional Intersatellite Range (ISR) observations (range magnitude only, not range direction or rate) are combined with standard GNSS observables. These ISR observations can be obtained from simple radio frequency (RF) or optical sensors. The methodology behind the combination approach is described and illustrated through a series of simulated case studies involving multiple satellites in low Earth orbit (LEO) using realistic hardware-derived (where possible) measurement noise. The results demonstrate that substantial improvements (factor of two or better) in the POD of the constellation satellites can be obtained with even intermittent ranging measurements, and with only millimeter-level ranging precision. This improved positioning capability enables new mission concepts for small-satellite constellations and formations, and makes these multi-satellite systems resilient to disruptions in GNSS signal availability. This GNSS-denial could be due to a variety of factors, such as intermittent or total hardware failure, power-related duty cycling, or ground-based jamming. Results show that under appropriate phasing of periodic GNSS-denial, combined with the new information from the ISR observations, POD levels approaching the non-GNSS-denied case can be achieved. For the cases of region-specific or single-satellite total GNSS-denial, constellations with ISR capability can be designed to completely compensate for the loss of GNSS observations and perform at levels better than with GNSS alone. Furthermore, the GNSS-denied case has an extended application for providing ISR-only POD for constellations around planetary bodies through the inversion of the invariant non-spherical gravity fields. Case studies are presented using high resolution invariant Earth and lunar gravity fields. In these example cases, ISR-only POD is demonstrated at the sub-meter level with the same millimeter precision of ISR. This research provides opportunities for new mission concepts that require precise positioning, improvements to mission operations, and enables new paradigms for orbit determination without access to GNSS.

View
Thumbnail Image
Item

A DATA-DRIVEN METHODOLOGY TO ANALYZE AIR TRAFFIC MANAGEMENT SYSTEM OPERATIONS WITHIN THE TERMINAL AIRSPACE

2021-12-10 , Corrado, Samantha Jane

Air Traffic Management (ATM) systems are the systems responsible for managing the operations of all aircraft within an airspace. In the past two decades, global modernization efforts have been underway to increase ATM system capacity and efficiency, while maintaining safety. Gaining a comprehensive understanding of both flight-level and airspace-level operations enables ATM system operators, planners, and decision-makers to make better-informed and more robust decisions related to the implementation of future operational concepts. The increased availability of operational data, including widely-accessible ADS-B trajectory data, and advances in modern machine learning techniques provide the basis for offline data-driven methods to be applied to analyze ATM system operations. Further, analysis of ATM system operations of arriving aircraft within the terminal airspace has the highest potential to impact safety, capacity, and efficiency levels due to the highest rate of accidents and incidents occurring during the arrival flight phases. Therefore, motivating this research is the question of how offline data-driven methods may be applied to ADS-B trajectory data to analyze ATM system operations at both the flight and airspace levels for arriving aircraft within the terminal airspace to extract novel insights relevant to ATM system operators, planners, and decision-makers. An offline data-driven methodology to analyze ATM system operations is proposed involving the following three steps: (i) Air Traffic Flow Identification, (ii) Anomaly Detection, and (iii) Airspace-Level Analysis. The proposed methodology is implemented considering ADS-B trajectory data that was extracted, cleaned, processed, and augmented for aircraft arriving at San Francisco International Airport (KSFO) during the full year of 2019 as well as the corresponding extracted and processed ASOS weather data. The Air Traffic Flow Identification step contributes a method to more reliably identify air traffic flows for arriving aircraft trajectories through a novel implementation of the HDBSCAN clustering algorithm with a weighted Euclidean distance function. The Anomaly Detection step contributes the novel distinction between spatial and energy anomalies in ADS-B trajectory data and provides key insights into the relationship between the two types of anomalies. Spatial anomalies are detected leveraging the aforementioned air traffic flow identification method, whereas energy anomalies are detected leveraging the DBSCAN clustering algorithm. Finally, the Airspace-Level Analysis step contributes a novel method to identify operational patterns and characterize operational states of aircraft arriving within the terminal airspace during specified time intervals leveraging the UMAP dimensionality reduction technique and DBSCAN clustering algorithm. Additionally, the ability to predict, in advance, a time interval’s operational pattern using metrics derived from the ASOS weather data as input and training a gradient-boosted decision tree (XGBoost) algorithm is provided.

View