Aircraft, turbomachinery, wind turbines, and other systems that generate or rely on aerodynamic forces are designed to operate most efficiently when flows are fully attached. However, especially due to increasing off-design performance requirements, there is significant risk of inefficient operation or failure due to flow separation. This work formulates a procedure for extending the performance envelope of many fluidic systems by delaying flow separation through real time separated flow state estimation and control. The history of active separation control is rich; however the approach presented here is novel in that it employs "real time" dynamical system updates to track nonlinear variations in the flow and provide robustness to flow state conditions. First, the dynamics of the canonical laminar separated flow over a flat plate at Rec=10⁵ are characterized by employing full-field, time-resolved PIV and unsteady surface pressure measurements. Dynamic Mode Decomposition (DMD) is employed on the high dimensional PIV velocity fields to identify the dynamically relevant spatial structure and temporal characteristics of the separated flow. Then, results of various cases of open-loop control using a zero-net mass flux actuator slot located just upstream of separation are presented that show separation reduction occurs for the employed actuation method. Real time estimates of the dynamical characteristics are provided by performing online DMD on measurements from a linear array of unsteady surface pressure transducers. The results show that online DMD of the pressure measurements provides reliable estimates of the modal characteristics of the separated flow subject to forcing. Furthermore, the dynamical estimates are updated at a rate commensurate with the characteristic time scales of the flow. Therefore, as the separated flow reacts to the applied forcing, online DMD applied to the surface pressure measurements provides a time-varying linear estimate of the evolution of the flow. Building upon these results, methods for adaptive control of flow separation based on the model provided by online DMD are formulated and implemented on the separated flow. Feedback control is implemented in which Linear Quadratic Regulator gains are computed recursively as the model provided by online DMD is updated. This physics-motivated, autonomous approach results in more efficient flow reattachment, requiring approximately 30% less actuator effort as compared with the commensurate open loop forcing case. Since this approach relies solely on observations of the separated flow, it is robust to variable flow conditions. Additionally, this approach does not require prior knowledge of the characteristics of the separated flow. / A Dissertation submitted to the Department of Mechanical Engineering in partial fulfillment of the requirements for the degree of Doctor of Philosophy. / Summer Semester 2018. / June 12, 2018. / Adaptive Control, Dynamic Mode Decomposition, Flow Control, Laminar Separation / Includes bibliographical references. / Louis N. Cattafesta, Professor Directing Dissertation; Mark Sussman, University Representative; Kunihiko Taira, Committee Member; Emmanuel Collins, Committee Member; Matthew Moore, Committee Member; Maziar Hemati, Committee Member.

With an increase in the number of space transport applications, the need for larger, more powerful and efficient rocket motors are the need of the day. Thus, optimizing the thrust generated by these rocket motors is of great importance, which can be achieved by bell-shaped nozzles with high area ratios. However, the high area ratios result in significant over-expansion at sea level conditions, that cause flow separation inside the nozzle and generate side-loads. Several novel nozzle designs are being developed to overcome the flow separation phenomenon. Characterizing the thrust output from these nozzles in a laboratory is vital for their successful development and implementation. Conventionally, the thrust from a nozzle is measured using load cells on a thrust stand. A thrust stand can provide limited information on the loads generated by the flow through a nozzle, such as the magnitude and direction of time average loads, but the flow features responsible for generating them remain elusive. In the present study, a novel method to estimate thrust from flow-field data, obtained by Particle Image Velocimetry (PIV) experiments and Pitot pressure survey at the nozzle exit, is proposed. The thrust estimated by this method is then validated with the conventional thrust measurements using load cells. A Mach 4 convergent-divergent nozzle with 12.7 mm throat diameter was tested using compressed air at a range of substantially over-expanded operating conditions with Nozzle pressure ratio (NPR) of 4, 5, 6, 7 and 7.5, at two temperature ratios (TR =1.2 and TR =1.5). The flow field at the nozzle exit was surveyed at these conditions using a Pitot tube mounted on a 2-D traverse system and the stereo PIV technique. Using the data obtained from both the flow surveys in the rocket thrust equation, the values of thrust are estimated. The thrust estimated from the flow field data showed identical levels at both temperature ratios. This suggested that temperature ratio has a negligible impact on the thrust measured at respective NPR. Using a load cell, the thrust produced by the nozzle was measured for each NPR at isothermal condition (TR=1). A comparison between the thrust obtained by the two methods verified that PIV and pressure surveys could be used to determine the time-averaged thrust of a nozzle to within 6% of the load cell readings. This experimental study provides a reliable alternative method for rocket nozzle thrust measurements / A Thesis submitted to the Department of Mechanical Engineering in partial fulfillment of the requirements for the degree of Master of Science. / Spring Semester 2018. / February 2, 2018. / Flow-filed diagnostics, load cell measurements, Over-expanded nozzles, Particle Image Velocimetry (PIV), Rocket Nozzle, Thrust Measurements / Includes bibliographical references. / Rajan Kumar, Professor Directing Thesis; William S. Oates, Committee Member; Neda Yaghoobian, Committee Member.

An experimental study is conducted on the interaction of a swept-shock wave with a turbulent boundary layer. The shock wave is generated by a sharp un-swept fin in a Mach 2 flow, where the strength of the interaction is varied from weak to moderate by changing the fin angle of attack from 10° to 15°, which corresponds to a normal Mach number of 1.3 and 1.4, respectively. Surface oil-flow visualization is used to study the mean characteristics of the interaction where surface features such as the upstream influence and separation line are identified. By taking advantage of the quasi-conical symmetry of the flowfield, two- and three-component velocity field measurements are acquired in the conical reference frame for the interaction of moderate interaction strength (Mn ~ 1.4) at two locations from the fin apex. Flowfield features such as the λ-shock structure, slip line, and the separation bubble with the reverse flow are clearly visible in the in-plane velocity fields. These results are also examined in the spherical coordinate frame, and good agreement in the spatial location of the critical features is found, providing direct quantitative, experimental evidence of quasi-conical symmetry of this flowfield above the surface. An examination of the velocity field downstream of the rear-foot of the λ-shock shows a region - a 'streamtube,' bounded on one side by the slip line emanating from the triple point - where the flow accelerates to transonic and supersonic speeds. This flow eventually turns towards and impinges upon the flat plate, a phenomenon referred to as an 'impinging jet' in literature and is believed to be the principal cause of the high mean, and unsteady pressures, very high heating and skin friction coefficients near impingement. The out-of-plane velocity fields unveiled the presence of a significant radially outward velocity component distinctly showing the presence of an 'open' separation bubble with a flattened conical vortex, a typical characteristic of a 3-D SBLI flowfield. The interaction dynamics are explored through unsteady surface pressure measurements at strategic locations. The highest unsteadiness is observed near the intermittent separation and underneath the open separation bubble. Further insights into the interaction dynamics is sought by examining the contributions to unsteady pressures in three spectral regimes - low-frequency (Stδ < 0.01), mid-frequency (0.01 < Stδ < 0.2) and high-frequency (Stδ > 0.2) to separate the contributions of each band to the total pressure fluctuations. Mid-frequency fluctuations dominate the current 3-D interaction flowfield, in contrast to 2-D SBLI where low-frequency disturbances are shown to be prominent. The spectral behavior shows no discrete peaks. However, relatively high coherence is observed between the intermittent separation region and underneath the separation bubble at Stδ ~ 0.013. It is plausible that the separation and rear shock are undergoing a low-frequency correlated motion, but the energy in this low-frequency periodic motion is found to be much lower than the mid-frequency unsteadiness that dominates this flowfield. Finally, the interaction is visualized using high-frame-rate conical shadowgraphy (24000 fps), where a shock detection scheme is utilized to identify the primary shock features from the instantaneous conical shadowgraphy images. The PDFs of their mean-subtracted spatial locations revealed that the separation shock undergoes the highest range of motion compared to other two shock features. It is inferred that the smaller extent of the λ-shock is more probable, which is further confirmed by the conditional sampling of the separation and rear shock slope based on the upstream or downstream movement of the separation shock. The response of this interaction to unsteady REM perturbations is also studied using three-component velocity and z-vorticity fields. Various REM configurations are tested (A1, A2, and A123) and it is distinctly seen that the A123 actuation most affected the flowfield. From the out-of-plane velocity fields, a significant impact of the REM induced CVPs on the flowfield inside the separation bubble is observed, whereas the inviscid region appears to remain unaltered. A general trend of increase in near-wall vorticity when compared to the baseline case, upstream of the intermittent separation is seen from the z-vorticity fields. Finally, the path of these induced CVPs in this highly 3-D flowfield became evident when these vorticity fields are visualized in conjunction with the surface flow maps from our earlier work. / A Dissertation submitted to the Department of Mechanical Engineering in partial fulfillment of the requirements for the degree of Doctor of Philosophy. / Fall Semester 2018. / November 2, 2018. / 3-D SBLI, SBLI, Separated Flows, Sharp fin-generated Interactions, Shock Boundary Layer Interactions, Shock Wave Boundary Layer Interactions / Includes bibliographical references. / Farrukh S. Alvi, Professor Directing Dissertation; Okenwa Okoli, University Representative; Rajan Kumar, Committee Member; Emmanuel G. Collins, Committee Member.

Control of aerodynamic noise generated by high-performance jet engines continues to remain a serious problem for the aviation community. Intense low frequency noise produced by large-scale coherent structures is known to dominate acoustic radiation in the aft angles. A tremendous amount of research effort has been dedicated towards the investigation of many passive and active flow control strategies to attenuate jet noise, while keeping performance penalties to a minimum. Unsteady excitation, an active control technique, seeks to modify acoustic sources in the jet by leveraging the naturally-occurring flow instabilities in the shear layer. While excitation at a lower range of frequencies that scale with the dynamics of large-scale structures, has been attempted by a number of studies, effects at higher excitation frequencies remain severely unexplored. One of the major limitations stems from the lack of appropriate flow control devices that have sufficient dynamic response and/or control authority to be useful in turbulent flows, especially at higher speeds. To this end, the current study seeks to fulfill two main objectives. First, the design and characterization of two high-frequency fluidic actuators ($25$ and $60$ kHz) are undertaken, where the target frequencies are guided by the dynamics of high-speed free jets. Second, the influence of high-frequency forcing on the aeroacoustics of high-speed jets is explored in some detail by implementing the nominally 25 kHz actuator on a Mach 0.9 ($Re_D = 5\times10^5$) free jet flow field. Subsequently, these findings are directly compared to the results of steady microjet injection experiments performed in the same rig and to prior jet noise control studies, where available. Finally, limited acoustic measurements were also performed by implementing the nominally 25 kHz actuators on jets at higher Mach numbers, including shock containing jets, and elevated temperatures. Using lumped element modeling as an initial guide, the current work expands on the previous development of low-frequency (2-8 kHz) Resonance Enhanced Micro-actuators (REM) to design actuators that are capable of producing high amplitude pulses at much higher frequencies. Extensive benchtop characterization, using acoustic measurements as well as optical diagnostics using a high resolution micro-schlieren setup, is employed to characterize the flow properties and dynamic response of these actuators. The actuators produced high-amplitude output a range of frequencies, $20.3-27.8$ kHz and $54.8-78.2$ kHz, respectively. In addition to providing information on the actuator flow physics and performances at various operating conditions, the benchtop study serves to develop relatively easy-to-integrate, high-frequency actuators for active control of high-speed jets for noise reduction. Following actuator characterization studies, the nominally 25 kHz ($St_{DF} \approx 2.2$) actuators are implemented on a Mach 0.9 free jet flow field. Eight actuators are azimuthally distributed at the nozzle exit to excite the initial shear layer at frequencies that are approximately an order of magnitude higher compared to the \textit{jet preferred frequency}, $St_P \approx 0.2-0.3$. The influence of control on the mean and turbulent characteristics of the jet, especially the developing shear layer, is examined in great detail using planar and stereoscopic Particle Image Velocimetry (PIV). Examination of cross-stream velocity profiles revealed that actuation leads to strong, spatially coherent streamwise vortex pairs which in turn significantly modify the mean flow field, resulting in a prominently undulated shear layer. These vortices grow as they convect downstream, enhancing local entrainment and significantly thickening the initial shear layer. Azimuthal inhomogeneity introduced in the jet shear layer is also evident in the simultaneous redistribution and reduction of peak turbulent fluctuations in the cross-plane near the nozzle exit. Further downstream, control results in a global suppression of turbulence intensities for all axial locations, also evidenced by a longer potential core and overall reduced jet spreading. The resulting impact on the noise signature is estimated via far-field acoustic measurements. Noise reduction was observed at low to moderate frequencies for all observation angles. Direct comparison of these results with that of steady microjet injection revealed some notable differences in the initial development of streamwise vorticity and the redistribution of peak turbulence in the azimuthal direction. However, despite significant differences in the near nozzle aerodynamics, the downstream evolution of the jet appeared to approach near similar conditions with both high-frequency and steady microjet injection. Moreover, the impact on far-field noise was also comparable between the two injection methods as well as with others reported in the literature. Finally, for jets at higher Mach numbers and elevated temperatures, the effect of control was observed to vary with jet conditions. While the impact of the two control mechanisms were fairly comparable on non-shock containing jets, high-frequency forcing was observed to produce significantly larger reductions in screech and broadband shock-associated noise (BBSN) at select under-expanded jet conditions. The observed variations in control effects at different jet conditions call for further investigation. / A Dissertation submitted to the Department of Mechanical Engineering in partial fulfillment of the requirements for the degree of Doctor of Philosophy. / Fall Semester 2017. / September 19, 2017. / Active Flow Control, Actuators, High-frequency excitation, High-speed Jets, Jet Noise Control, Particle Image Velocimetry / Includes bibliographical references. / Farrukh Alvi, Professor Directing Dissertation; M. Yousu Hussaini, University Representative; Rajan Kumar, Committee Member; Jonathan Clark, Committee Member; Jonas P. R. Gustavsson, Committee Member.

<p> Within the field of Guidance, Navigation, and Control, the navigation process refers to accurately determining one's position in space. The quest of accurate navigation has shaped human history. Early navigation techniques involved dead reckoning with infrequent measurement updates from line-of-sights to stars or landmarks on the shore. The first practical inertial navigation system (INS) attributed to the German V-2 missile in 1942. In the early 1960's the Kalman filter was developed to aid in the merging of mathematical models and measurement updating. Throughout the space age and continuing into today's remote systems the hardware has made vast improvements; however, the navigation filtering theory has remained mostly stagnant with the multiplicative Extended Kalman Filter (MEKF) still being the workhorse of most modern INS applications. </p><p> In most INS applications, the state vector usually consists of the attitude, position, velocity, and Inertial Measurement Unit (IMU) calibration parameters such as biases and scale factors. Because position-type measurements are usually only given, such as pseudoranges to GPS satellites, the observability of the attitude and gyroscope calibration parameters is weak. Since the early days of employing the MEKF for INS applications, and even modern-day applications, the state errors are defined as a simple algebraic difference between the truth and the estimate. </p><p> It has been argued that a new state-error definition is required in which state-error quantities are defined using elements expressed in a common frame. This provides a realistic framework to describe the actual errors. In previous work, the errors were put into a common frame using the estimated attitude error, which led to the ``geometric EKF'' (GEKF). The GEKF provides extra transport terms, due to error-attitude coupling with the states, in the filter. This previous work focused strictly on attitude estimation, which incorporated only ``body-frame" errors. In this work the GEKF is extended to the INS formulation. Here, errors are considered for both the body frame and the filter's navigation frame. </p><p> In this work, it is shown how these body-frame and navigation-frame errors are related through a similarity transformation. The body-frame error, and the navigation-frame error through this similarity transformation, are first examined in a Planar Inertial Navigation (PIN) problem. For the PIN problem, the MEKF and GEKF algorithms are derived using the same kinematic and measurement models. These algorithms are then studied for a single simulation test case; these results were then verified via Monte Carlo analysis. For this example, it was shown that the body-frame errors had a faster convergence for the GEKF; however, the navigation-frame states showed slower convergence. It is argued here that the direct comparison of these results is inconclusive since both filters employ different error definitions; therefore, the errors being examined utilize a different error metric. It can be stated that the errors realized by the GEKF are more representative of the real errors experienced by the system. </p><p> The body-frame and navigation-frame errors are also used to derive the GEKF for three navigation filters. Specifically, this work examines the absolute Earth-Centered-Earth-Fixed (ECEF) navigation, relative ECEF navigation, and North-East-Down (NED) navigation filters. The counterpart MEKF filters are also derived in this work to illustrate the differences seen in the state matrices due to the additional coupling terms. These filters are also studied via simulation studies. However, now the body-framed vectors do not show faster convergence for the GEKF. This is because the measurement update for this specific example is unaffected by the new error definition. The measurements are not affected by the new error definition because these filters only utilize pseudo-GPS position measurements, and it is shown that the position error still utilizes the old error definition due to its kinematic relation to the velocity error. </p><p> Finally, this work conducts a linearize analysis of a simplified INS where the GEKF in the NED frame is considered. It is shown via a stationary analysis that the fundamental frequencies of the GEKF are the same as those of the MEKF. This is because during the stationary analysis all of the additional coupling terms seen in the state error matrix are neglected due to assumptions made about the vehicle's motion.</p><p>

<p> This thesis studies the feasibility of using neural networks to ''learn" the vortex panel method. This study is motivated by the desire for the rapid and accurate prediction of fluid flows during the preliminary design of engineering systems, where traditional computational fluid dynamics (CFD) are too computationally costly. The results show that a two-layer neural network can estimate the pressure coefficient and elements in the vortex-panel influence-coefficient matrix. However, when the neural-network-predicted influence-coefficient matrix is used to estimate the pressure coefficients, the results are in poor agreement with the baseline prediction, although general trends are captured. </p><p>

<p> This research presents new methods to apply safety standards to Detect and Avoid (DAA) functions for Unmanned Aircraft Systems (UAS), using maneuvering target tracking and encounter models. </p><p> Previous DAA research methods focused on predefined, linear encounter generation. The new estimation and prediction methods in this research are based on the target tracking of maneuvering intruders using Multiple Model Adaptive Estimation and a realistic random encounter generation based on an established encounter model. </p><p> When tracking maneuvering intruders there is limited knowledge of changes in intruder behavior beyond the current measurement. The standard Kalman filter (KF) with a single motion model is limited in performance for such problems due to ineffective responses as the target maneuvers. In these cases, state estimation can be improved using MMAE. It is assumed that the current active dynamic model is one of a discrete set of models, each of which is the basis for a separate filter. These filters run in parallel to estimate the states of targets with changing dynamics. </p><p> In practical applications of multiple model systems, one of the most popular algorithms for the MMAE is the Interacting Multiple Model (IMM) estimator. In the IMM, the regime switching is modeled by a finite state homogeneous Markov Chain. This is represented by a transition probability matrix characterizing the mode transitions. A Markov Chain is a stochastic model describing a sequence of possible events in which the probability of each event depends only on the previous event. </p><p> This research uses the hazard states estimates (which are derived from DAA standards) to analyze the IMM performance, and then presents a new method to predict the hazard states. To reduce the prediction error, this new method accounts for maneuvering intruders. The new prediction method uses the prediction phase in the IMM algorithm to predict the future intruder aircraft states based on the current and past sensor measurements. </p><p> The estimation and prediction methods described in this thesis can help ensure safe encounters between UAS and manned aircraft in the National Airspace System through improvement of the trajectory estimation used to inform the DAA sensor certification process.</p><p>

<p> The variable pitch quadrotor is not a new concept but has been largely ignored in small unmanned aircraft, unlike the fixed pitch quadcopter which is controlled only by changing the RPM of the motors and only has about 30 minutes of total flight time. The variable pitch quadrotor can be controlled either by the change of the motor RPM or rotor blade pitch angle or by the combination of both. This gives the variable pitch quadrotor potential advantages in payload, maneuverability and long endurance flight. This research is focused on the design methodology for a variable pitch quadrotor using a single motor with potential applications for a long endurance flight. This variable pitch quadcopter uses a single power plant to power all four rotors through a power transmission system. All four rotors have the same rpm but vary the blade pitch angle to control its attitude in the air. A proof of concept variable pitch quadcopter is developed for testing the drivetrain mechanism on the vehicle and evaluating performance of the vehicle through numbers of testing. </p><p>

<p> In today&rsquo;s space industry, satellite formation flying has become a cost-efficient alternative solution for science, on-orbit repair and military time-critical missions. While in orbit, the satellites are exposed to the space environment and unpredictable spacecraft on-board disturbances that negatively affect the attitude control system&rsquo;s ability to reduce relative position and velocity error. Satellites utilizing a PID or adaptive controller are typically tune to reduce the error induced by space environment disturbances. However, in the case of an unforeseen spacecraft disturbance, such as a fault in an IMU, the PID based attitude control system effectiveness will deteriorate and will not be able to reduce the error to an acceptable magnitude. </p><p> In order to address the shortcomings a PID-Extended Memory RNN (EMRNN) adaptive controller is proposed. A PID-EMRNN with a short memory of multiple time steps is capable of producing a control input that improves the translational position and velocity error transient response compared to a PID. The results demonstrate the PID-EMRNN controller ability to generate a faster settling and rise time for control signal curves. The PID-EMRNN also produced similar results for an altitude range of 400 km to 1000 km and inclination range of 40 to 65 degrees angles of inclination. The proposed PID-EMRNN adaptive controller has demonstrated the capability of yielding a faster position error and control signal transient response in satellite formation flying scenario. </p><p>

<p> Carbon-Fiber Reinforced Polymers (CFRPs) provides superior mechanical properties and low weight, enabling their extensive use in the aerospace industry. Susceptibility to internal damage due to out-of-plane loads and poor electrical properties are some of their major challenges that require to be addressed in order to increase the utilization of composites in further aerospace structures. Lightning strikes can lead to catastrophic damage, inflicting high repair and certification costs. Lightning Strike Protection (LSP) solutions such as integration of metallic meshes or foils into the composite structures, even though effective, impose extra costs and hinders the aircraft performance due to the increased weight of the aircraft. </p><p> This research aims at the development of a different LSP solution, by enhancing the electrical conductivity of composite, while maintaining a sufficient degree of mechanical properties. The use of non-woven conductive interlayers was proposed for manufacturing of conductive composites. Highly-conductive, low-aerial-weight carbon veil was utilized to manufacture prepreg-based CF/Epoxy laminates, which are generally toughened, in order to improve their conductivity using vacuum bag only (VBO) and heat-pressing techniques. Further, a bi-functional interlayer of graphene coated Polyamide (PA) was developed using interfacial trapping method. This conductive thermoplastic interlayer was then utilized for manufacturing Benzoxazine (BZ) infused carbon fabric laminate with Vacuum-assisted resin transfer molding (VARTM) method, which acted as a conductive toughener and improves the Inter-laminar Fracture Toughness (ILFT) as well as to increase the electrical conductivity. </p><p> The effects of the incorporation of non-woven interlayers on the electrical conductivity, thermal behavior of composites, and mechanical properties such as shear strength, compressive strength, and the ILFT (Mode-I and Mode-II) were investigated in this study. In both types of composites, an increase in electrical properties, as well as mechanical properties, were observed. The only exception was in the Mode-I ILFT of the CF/Epoxy prepregs, which decreased with the increase of the areal weight of the interleaved carbon veils. The mechanical properties increased in the range of 9%&ndash;138% with the only decrement observed in Mode-I ILFT of CF/Epoxy with carbon veils of 25%. The volume resistivity of the CF/Epoxy samples decreased significantly by approximately 50% due to the incorporation of the conductive interlayer. This added feature was used to develop a structural health monitoring (SHM) procedure. The conductive composite showed an increased sensitivity in detecting the pre-identified damage location in the composites.</p><p>