Doctor of Philosophy with a Major in Aerospace Engineering

Series

Doctor of Philosophy with a Major in Aerospace Engineering

Permanent Link

https://hdl.handle.net/1853/72952

Series Type

Degree Series

Associated Organization(s)

Organizational Unit

Daniel Guggenheim School of Aerospace Engineering

Full item page

Publication Search Results

Now showing 1 - 10 of 642

Optimization-based methods for deterministic and stochastic control: Algorithmic development, analysis and applications on mechanical systems & fields

(Georgia Institute of Technology, 2019-12-17) Boutselis, Georgios

Developing efficient control algorithms for practical scenarios remains a key challenge for the scientific community. Towards this goal, optimal control theory has been widely employed over the past decades, with applications both in simulated and real environments. Unfortunately, standard model-based approaches become highly ineffective when modeling accuracy degrades. This may stem from erroneous estimates of physical parameters (e.g., friction coefficients, moments of inertia), or dynamics components which are inherently hard to model. System uncertainty should therefore be properly handled within control methodologies for both theoretical and practical purposes. Of equal importance are state and control constraints, which must be effectively handled for safety critical systems. To proceed, the majority of works in controls and reinforcement learning literature deals with systems lying in finite-dimensional Euclidean spaces. For many interesting applications in aerospace engineering, robotics and physics, however, we must often consider dynamics with more challenging configuration spaces. These include systems evolving on differentiable manifolds, as well as systems described by stochastic partial differential equations. Some problem examples of the former case are spacecraft attitude control, modeling of elastic beams and control of quantum spin systems. Regarding the latter, we have control of thermal/fluid flows, chemical reactors and advanced batteries. This work attempts to address the challenges mentioned above. We will develop numerical optimal control methods that explicitly incorporate modeling uncertainty, as well as deterministic and probabilistic constraints into prediction and decision making. Our iterative schemes provide scalability by relying on dynamic programming principles as well as sampling-based techniques. Depending upon different problem setups, we will handle uncertainty by employing suitable concepts from machine learning and uncertainty quantification theory. Moreover, we will show that well-known numerical control methods can be extended for mechanical systems evolving on manifolds, and dynamics described by stochastic partial differential equations. Our algorithmic derivations utilize key concepts from optimal control and optimization theory, and in some cases, theoretical results will be provided on the convergence properties of the proposed methods. The effectiveness and applicability of our approach are highlighted by substantial numerical results on simulated test cases.
An architecture model of the U.S. air transportation network

(Georgia Institute of Technology, 2019-11-26) Song, Kisun

For almost a century, the U.S. Air Transportation Network (ATN) has continuously and successfully adapted to its changing environment as if it were a living organism. Today, the complexity of the network encompasses various exogenous as well as endogenous factors: fuel price, socioeconomic and political climates, atmospheric conditions, varying interests of stakeholders, and growing dependence on technology, to name a few. Its sophisticated interactions among diverse factors affecting the ATN have captivated many network researchers. Some researchers have attempted to retrieve an order out of seemingly chaotic constructions, while others have analyzed historical variations in its properties to understand the ATN’s behavioral mechanisms. However, its mathematical representation led by the known components and rules is yet to be developed. Thus, this thesis develops an architecture model of the ATN that mathematically represents the components and rules with realism. In the model, the network evolves in a virtual environment comprising three time-variant components – demand, airport, and aircraft technology – built upon extensive realistic datasets. Then the network is constructed by the active agents – airlines – performing multi-tiered network evolutionary processes and evolves into a strong hub-and-spoke (H&S) structure network that mimics the function of its reference: real-world ATN. The validated model provides various opportunities to conduct extensive analyses and studies on the past, current, and future of the ATN. Finally, a case study has been performed: forecasting the future ATN disruption caused by the technological revolution of civil supersonic transports. It provided an opportunity to experience the exploratory and interpretative capability of the architecture model, which shed light on performing future researches with better realism.
Ignition, topology, and growth of turbulent premixed flames in supersonic flows

(Georgia Institute of Technology, 2019-11-12) Ochs, Bradley Alan

Supersonic combustion ramjets (scramjets) are currently the most efficient combustor technology for air breathing hypersonic flight, however, lack of fundamental understanding and numerous engineering challenges hinder regular deployment of these devices. This work addresses scramjet-relevant knowledge gaps in supersonic turbulent premixed combustion, including laser ignition, numerical modeling, and flame-compressibility interaction. One of the main contributions of this work is introduction of a new turbulent premixed flame arrangement where flame-compressibility interaction can be systematically explored: flame kernels in an expanding flow field. The scramjet flow path is replaced by a simplified channel geometry with a well characterized mean flow acceleration that mimics flow field expansion typically imposed on scramjet combustors to avoid thermal choking. Spherically expanding flames are created via laser ignition and subsequent flame growth and morphology are investigated using combined physical and numerical experiments. Pressure-density misalignment due to flame-compressibility interaction produces vorticity at the flame surface through baroclinic torque, i.e. flame-compressibility interaction acts like a turbulence source. The flame ultimately evolves into a reacting vortex ring that increases the flame speed and enhances reactant consumption. To explore the relative importance of turbulence and compressibility on flame dynamics, the Mach number (M=1.5,1.75,2), equivalence ratio (φ= 1.0,0.9,0.8,0.7), and root-mean-squared turbulent velocity (u'=3.98,4.14,4.45 m/s) are varied systematically. This work also introduces flame kernels in an expanding flow field as a canonical numerical validation test case for flame-compressibility interaction. Inaccuracies in simulation results are easily identified due to high flow velocity and simplicity of the problem. The numerical setup and models are scrutinized to minimize errors. Using the appropriately verified numerical models, simulation results show very reasonable agreement with experimental data. Validated simulations are instrumental in enhancing understanding of the underlying physics of supersonic flame kernels. Laser ignition studies in supersonic flows have historically focused on ignition of non-premixed fuels within cavity flame holders. This work introduces a far simpler and more tractable problem: laser ignition of a fully premixed supersonic gas. Ignition experiments with a range of laser settings are performed to determine supersonic breakdown and ignition probabilities, length of time the ignition event influences flame growth, and Mach number influence on the ignition process. The ignition event has a long-lasting effect on kernel growth, but the influence can be minimized by properly selecting the laser energy. Mach number has a minimal impact on the ignition process, but does affect the initial kernel shape due to flow field variations with Mach number. Kernel growth matches low speed studies closely at early times, but deviates at later times due to vortex ring topology. It is not obvious how the turbulent flame speed will scale for flows with mean compressibility. Therefore, the combined physical and numerical experiments are leveraged to explore this question. The vortex ring causes significant errors in the line of sight-measured burned volume, hence correction factors to convert from line of sight to volumetric measurements are presented. Conditions for displacement and consumption speed equivalency are shown to depend heavily on the particular diagnostic used; which progress variable isocontour is measured and where it is measured within the flame brush must be considered carefully during interpretation of experimental data. Scaling with the RMS turbulent velocity cannot collapse these flame speed data, i.e. previously established flame speed scalings are inappropriate for flames interacting with compressibility. Drawing motivation from vortex ring literature, a new flame speed scaling based on the ring propagation velocity is proposed. The proposed scaling collapses the data and produces a nearly linear scaling regime, which suggests turbulence plays a secondary role to the hydrodynamic instability created by flame-compressibility interaction. In summary, flame kernels are a new and effective canonical configuration for exploring flame-compressibility interactions in supersonic flows.
Using sample-based continuation techniques to efficiently compute subspace reachable sets and Pareto surfaces

(Georgia Institute of Technology, 2019-11-11) Brew, Julian

For a given continuous-time dynamical system with control input constraints and prescribed state boundary conditions, one can compute the reachable set at a specified time horizon. Forward reachable sets contain all states that can be reached using a feasible control policy at the specified time horizon. Alternatively, backwards reachable sets contain all initial states that can reach the prescribed state boundary condition using a feasible control policy at the specified time horizon. The computation of reachable sets has been applied to many problems such as vehicle collision avoidance, operational safety planning, system capability demonstration, and even economic modeling and weather forecasting. However, computing reachable volumes for general nonlinear systems is very difficult to do both accurately and efficiently. The first contribution of this thesis investigates computational techniques for alleviating the curse of dimensionality by computing reachable sets on subspaces of the full state dimension and computing point solutions for the reachable set boundary. To compute these point solutions, optimal control problems are reduced to initial value problems using continuation methods and then solved. The sample-based continuation techniques are computationally efficient in that they are easily parallelizable. However, the distribution of samples on the reachable set boundary is not directly controlled. The second contribution presents necessary conditions for distributed computation convergence, as well as necessary conditions for curvature- or uniform coverage-based sampling methods. Solutions to multi-objective optimization problems are generally defined using a set of feasible solutions such that for any one objective to improve it is necessary for other objectives to degrade. This suggests there is a connection between the two fields with the potential of cross-fertilization of computational techniques and theory. The third contribution explores analytical connections between reachability theory and multi-objective optimization with investigation into properties, constraints, and special cases.
A reduced-order modeling methodology for the multidisciplinary design and analysis of boundary layer ingestion configurations

(Georgia Institute of Technology, 2019-11-08) Bozeman, Michael Dwain

In response to the increasingly stringent requirements for subsonic transport aircraft, NASA has established aggressive goals for the noise, emissions, and fuel burn of the next generations of aircraft. This has led to the investigation of a variety of unconventional configurations and new technologies. Boundary Layer Ingestion (BLI) propulsion has been identified as a promising technology to reduce fuel burn. Preliminary studies show that BLI propulsion can offer 3-12% reduction in fuel burn, depending on the configuration. Traditionally, the design and analysis of the airframe and propulsion system has been performed in a decoupled manner. For BLI configurations, the propulsion system is tightly integrated into the airframe resulting in strong interactions between the airframe aerodynamics and propulsion system performance. As a result, the design and analysis of BLI configurations requires coupled multidisciplinary analysis (MDA) consisting of an aerodynamic analysis in an iterative loop with a propulsion system analysis. This is a very expensive analysis considering the requirement for high-fidelity models. Additionally, the design of highly-coupled configurations cannot rely solely on intuition to make design decisions. Advanced methods including Multidisciplinary Analysis and Optimization (MDAO) and design space exploration are needed to allow for the design decisions to be made based directly on a system-level objective (e.g., fuel burn) and to allow for design studies to provide insight into the multidisciplinary trades associated with BLI configurations. However, MDAO and design space exploration using coupled, high-fidelity analysis models are not practical. In this work, reduced-order modeling (ROM) is proposed as a potential solution to reduce the computational cost associated with the coupled MDA of BLI configurations and to enable these advanced design methods. An interpolation-based POD ROM is developed based on the CFD analysis to allow for predictions of the aerodynamics over a range of propulsor operating conditions for a simplified tail-cone thruster (TCT) configuration. The resulting ROM is then coupled to a propulsion model to perform ROM-based, coupled MDA. Finally, the ROM-based, coupled MDA approach is employed for coupled MDAO to assess the performance benefit offered relative to equivalent CFD- and adjoint-based approaches. The results show that the ROM-based, coupled MDA approach offers an improvement in performance relative to the current state of the art. Relative to the equivalent CFD-based approach, the ROM-based, coupled MDA method demonstrated significant computational savings for even a single optimization. However, the ROM-based approach requires multiple optimizations to offer a computational benefit over the adjoint-based approach. This result highlights the benefit of the proposed approach for optimization studies and design space exploration.
Optimal covariance steering: Theory and its application to autonomous driving

(Georgia Institute of Technology, 2019-11-08) Okamoto, Kazuhide

Optimal control under uncertainty has been one of the central research topics in the control community for decades. While a number of theories have been developed to control a single state from an initial state to a target state, in some situations, it is preferable to simultaneously compute control commands for multiple states that start from an initial distribution and converge to a target distribution. This dissertation aims to develop a stochastic optimal control theory that, in addition to the mean, explicitly steers the state covariance. Specifically, we focus on the control of linear time-varying (LTV) systems with additive Gaussian noise. The task is to steer a Gaussian-distributed initial system state distribution to a target Gaussian distribution, while minimizing a state and control expectation-dependent quadratic cost under probabilistic state constraints. Notice that, in such systems, the system state keeps being Gaussian distributed. Because Gaussian distributions can be fully described by the first two moments, the proposed optimal covariance steering (OCS) theory allows us to control the whole distribution of the state and quantify the effect of uncertainty without conducting Monte-Carlo simulations. We propose to use a control policy that is an affine function of filtered disturbances, which utilizes the results of convex optimization theory and efficiently finds the solution. After the OCS theory for LTV systems is introduced, we extend the theory to vehicle path planning problems. While several path planning algorithms have been proposed, many of them have dealt with deterministic dynamics or stochastic dynamics with open-loop un- certainty, i.e., the uncertainty of the system state is not controlled and, typically, increases with time due to exogenous disturbances, which may lead to the design of potentially conservative nominal paths. A typical approach to deal with disturbances is to use a lower-level local feedback controller after the nominal path is computed. This unidirectional dependence of the feedback controller on the path planner makes the nominal path unnecessarily conservative. The path-planning approach we develop based on the OCS theory computes the nominal path based on the closed-loop evolution of the system uncertainty by simultaneously optimizing the feedforward and feedback control commands. We validate the performance using numerical simulations with single and multiple vehicle path planning problems. Furthermore, we introduce an optimal covariance steering controller for linear systems with input hard constraints. As many real-world systems have input constraints (e.g., air- craft and spacecraft have minimum/maximum thrust), this problem formulation will allow us to deal with realistic scenarios. In order to incorporate input hard constraints in the OCS theory framework, we use element-wise saturation functions and limit the effect of disturbance to the control commands. We prove that this problem formulation leads to a convex programming problem and demonstrate the effectiveness using simple numerical examples. Finally, we develop the OCS-based stochastic model predictive control (CS-SMPC) theory for stochastic linear time-invariant (LTI) systems with additive Gaussian noise subject to state and control constraints. In addition to the conventional terminal cost and terminal mean constraints, we introduce terminal covariance constraints in the stochastic model predictive control theory. The OCS theory efficiently computes the control commands that satisfy the terminal covariance constraints. The key benefit of the CS-SMPC algorithm is its ability to ensure stability and recursive feasibility of the controlled system. In addition, thanks to the efficient OCS theory, the proposed CS-SMPC theory is computationally less demanding than previous SMPC approaches. In order to verify the effectiveness, the CS-SMPC approach is also applied to the problem of self-driving vehicle control under uncertainty.
Experimental investigation of nitrogen oxide production in premixed reacting jets in a vitiated crossflow

(Georgia Institute of Technology, 2019-09-03) Sirignano, Matthew Davis

The presented work describes the experimental investigation of nitrogen oxide (NOx) emissions from reacting jets in a vitiated crossflow (RJICF). It is motivated by interest in axial staging of combustion as an approach to reduce undesirable NOx emissions from gas turbine combustors operating at high flame temperatures (>1900K). In lean-premixed combustion, NOx levels are exponential functions of temperature and linear functions of residence time. Consequently, NOx production rates are high at such temperatures, and conventional combustor architectures are unable to simultaneously deliver low NOx and part-load operability. A RJICF is a natural means of implementing axial staging. Therefore, a fuller understanding of the governing processes and parameters regarding pollutant formation within this complex flow field is critical to the next generation of gas turbine technology advancement. It is clear that RJICF NOx production is a highly coupled process. A key challenge was decoupling the interdependent jet parameters in order to observe fundamental NOx production sensitivities. Data is presented for premixed jets injected into a vitiated crossflow of lean combustion products. The jets varied in: fuel selection (methane or ethane or a combination), equivalence ratio (0.8≤ϕjet≤9.0), momentum flux ratio (2≤J≤40), and exit geometry (pipe or nozzle). The crossflow temperatures ranged from 1350K – 1810K, and the reacting jets induced a bulk averaged temperature rise on the flow (ΔT) ranging from 75K – 350K. In addition, several data series were replicated with varied ethane/methane ratios at constant ϕjet to influence flame lifting independent of other parameters. Similarly, the jet exit geometry was varied to influence shear layer vortex growth rates. Overall, these data indicate that NOx emissions are largely determined by ΔT. However, significant variation was observed at constant ΔT levels. The data is consistent with the idea that this variation is controlled by the stoichiometry at which combustion actually occurs, referred to as ϕFlame. ϕFlame is influenced by ϕjet and pre-flame mixing of the jet and crossflow that, in turn, is a function of flame lift-off distance (LO), nozzle geometry, and crossflow temperature. The data highlights the importance of flame lifting as well as the potential importance of post-flame mixing effects. Both are complex problems and are not directly addressed in this work. Further work in these areas would significantly deepen understanding of the relevant phenomena in RJICF NOx production.
Investigation of high-pressure methane and syngas autoignition delay times

(Georgia Institute of Technology, 2019-08-28) Karimi, Miad

This thesis reports methane (CH4) and a syngas mixture (H2/CO=95:5) autoignition delay measurements relevant to operating conditions of supercritical carbon dioxide (sCO2) power cycle (100 to 300 bar) combustors. To acquire data at these conditions as part of this thesis, a new high-pressure shock tube is designed, fabricated and commissioned. The experiments are conducted for diluted carbon dioxide environments at 100 and 200 bar and at temperatures within the range of approximately 1100–1400 K. To investigate the chemical effect of CO2 at supercritical conditions, experiments are conducted at similar pressures and temperatures by substituting CO2 with an inert bath gas, Ar (argon). Obtaining ignition delay times in Ar bath gas allows to systematically study the chemical effect of CO2 on ignition chemistry. Methane ignition delay times are compared to several chemical kinetic models, such as Aramco 2.0, FFCM-1, HP-Mech, USC Mech II and GRI 3.0. For the conditions of this study, predictions of the Aramco 2.0 kinetic model show the overall best agreement with experimental measurements. Following the experimental data, brute-force sensitivity analyses and reaction pathway flux analyses are utilized to gain insight into details of the ignition chemistry of the fuels (CH4 and H2/CO=95:5). These analyses indicate that methyl (CH3) recombination to form ethane (C2H6) and oxidation of CH3 to form methoxide (CH3O) are the most important reactions controlling the ignition behavior of methane at temperatures greater than approximately 1250 K. However, at temperatures below approximately 1250 K, an additional reaction pathway for methyl radicals is found through CH3+O2+M=CH3O2+M, which leads to formation of methyldioxidanyl (CH3O2). This reaction pathway plays a distinct role in dictating the ignition trends at lower temperature conditions. Replacing CO2 with argon as the bath gas reveals that CO2 does not have major effects on ignition chemistry of CH4. A similar approach is taken to obtain experimental data at 100 bar and 200 bar for a syngas fuel mixture of 95% H2 (hydrogen) and 5% CO (carbon monoxide) in CO2 and Ar bath gasses. Aramco 2.0 kinetic model, FFCM-1 kinetic model, HP-Mech and USC Mech II show good agreement with the measured ignition delay times. Detailed sensitivity analyses of these kinetic models highlight the importance of the third-body reaction between hydrogen atoms (H) and oxygen molecules (O2) through H+O2+M=HO2+M to form hydroperoxyl (HO2). In both cases, irrespective of the diluents, this reaction is the most influential reaction to hinder ignition. Ignition delay times obtained from both mixtures not only show a similar trend, but also the same magnitude when compared to the CO2 mixture. While this observation may suggest that CO2 has no chemical effect on ignition chemistry, it is found to play a counterbalancing role on syngas ignition at the elevated pressures and temperatures of this study. CO2 increases the OH (hydroxyl) radical production by colliding with hydrogen peroxide (H2O2) through H2O2+M=OH+OH+M. However, it reduces OH production through HO2+H=OH+OH due to a lower amount of H radical production compared to the Ar mixture. Therefore, these two effects cancel out the change of OH productions, and CO2 does not change the ignition delay time of the syngas mixture considered in this study upon comparison with the mixture with Ar bath gas.
An architecture-based growth approach for industrial gas turbine product development

(Georgia Institute of Technology, 2019-07-30) Fu, Haoyun

In this research, an architecture-based methodology has been developed to understand and interpret the ascending performance trajectory of industrial gas turbines from a growth perspective. Historical data depicting the product evolvement are examined to reveal trends and features that can be tied to the published design philosophy and practices in this industry. Quantifiable growth metrics are introduced and deployed in an established framework that offers a scientific product development environment to emulate the prevalent product development practices. Furthermore, the capability established by this methodology is expected to support performance prediction and planning for future gas turbine products.
Investigating physics of nanosecond-pulsed argon plasma discharges for a VLF plasma antenna

(Georgia Institute of Technology, 2019-07-30) Liu, Connie Y.

Very low frequency (VLF) waves (3-30 kHz) are useful in communication and navigation, can deeply penetrate the ground and ocean surface, and help with satellite protection by removing energetic charged particles in the Van Allen radiation belts that damage satellite electronics. However, current VLF antenna arrays take 1000s of acres because they are efficiency-limited – the signal propagates down the antenna and reflects faster than the signal period, which interferes with and cancels the outgoing signal. A top-hat loaded antenna is a solution that radiates more efficiently but is constrained to a small bandwidth. Replacing the metal conductor in a conventional antenna with a series of individually-controlled plasma cells in a plasma antenna could overcome both efficiency and bandwidth limitations. Modulating the plasma conductivity in each segment would turn a portion of the antenna on or off and suppress reflected waves in the time-domain by removing the necessary electrically conducting pathway. The two main research goals were to further understand the physics of pulsed plasmas by investigating ionization and recombination of pulsed plasmas on the nanosecond timescale and how operating conditions affect the time-resolved conductivity of a pulsed plasma. A single plasma cell was investigated by generating nanosecond-pulsed, argon plasma at various pulse frequencies, widths, and pressures. Argon emission lines were analyzed with an ICCD-spectrometer assembly gating at 4 ns, and relative intensities of strong argon neutral and ion lines were used in line-ratio calculations. These experimentally-determined ratios were compared to theoretical ratios generated from PrismSPECT, a collisional-radiative spectral analysis software, to obtain time-resolved electron temperature (~ 1 eV), electron density (1014 − 1015cm−3), and plasma frequency (~200 GHz). Those results were used to discover trends and extract sets of plasma parameters for the rapid ionization and recombination needed for a successful VLF plasma antenna design. Further investigations into the physics of nanosecond-pulsed plasmas could include analysis of wavelength transitions and processes as well as the effects of electrode geometry on plasma properties. Additional future work needed for a VLF plasma antenna demonstration would entail developing the signal propagation technology needed for transmission through the plasma antenna cell.