Mechanical Engineering Theses and Dissertations
Permanent URI for this collectionhttp://hdl.handle.net/1903/2795
Browse
15 results
Search Results
Item Denoising the Design Space: Diffusion Models for Accelerated Airfoil Shape Optimization(2024) Diniz, Cashen; Fuge, Mark D; Mechanical Engineering; Digital Repository at the University of Maryland; University of Maryland (College Park, Md.)Generative models offer the possibility to accelerate and potentially substitute parts of the often expensive traditional design optimization process. We present Aero-DDM, a novel application of a latent denoising diffusion model (DDM) capable of generating airfoil geometries conditioned on flow parameters and an area constraint. Additionally, we create a novel, diverse dataset of optimized airfoil designs that better reflects a realistic design space than has been done in previous work. Aero-DDM is applied to this dataset, and key metrics are assessed both statistically and with an open-source computational fluid dynamics (CFD) solver to determine the performance of the generated designs. We compare our approach to an optimal transport GAN, and demonstrate that our model can generate designs with superior performance statistically, in aerodynamic benchmarks, and in warm-start scenarios. We also extend our diffusion model approach, and demonstrate that the number of steps required for inference can be reduced by as much as ~86%, compared to an optimized version of the baseline inference process, without meaningful degradation in design quality, simply by using the initial design to start the denoising process.Item DISSECTION AND MODELING OF AEDC WIND TUNNEL 9 CONTROL LAW AND FACILITY DURING BLOW PHASE(2023) Gigioli, Samuel George; Gupta, Ashwani K; Mechanical Engineering; Digital Repository at the University of Maryland; University of Maryland (College Park, Md.)This work presents the progress towards a mathematical modeling of the Arnold Engineering Development Complex (AEDC) Wind Tunnel 9 control law during the blow phase of a given tunnel run, composing of electrical analog physics, ideal gas control volume physics, incompressible fluid mechanics, and force balance kinematics. This work is unique to Tunnel 9 and unique in respect to other works, as no other existing models of the current control law exist. The primary goal of this work is to provide enhanced support to the Tunnel 9 engineers with the ability to model different run conditions. Key facility measurements can be estimated, aiding in the determination if proposed non-standard run conditions will meet or maintain the facility capabilities, and if the facility can be operated under safe operating limits. The secondary goal of this model is to progress toward a digitally controlled valve system to replace the current analog system. Such will help provide advantages in the facility (1) performance, (2) health monitoring, (3) maintainability, and (4) sustainment.Item Ordering Non-Linear Subspaces for Airfoil Design and Optimization via Rotation Augmented Gradients(2023) Van Slooten, Alec; Fuge, Mark; Mechanical Engineering; Digital Repository at the University of Maryland; University of Maryland (College Park, Md.)Airfoil optimization is critical to the design of turbine blades and aerial vehicle wings, among other aerodynamic applications. This design process is often constrained by the computational time required to perform CFD simulations on different design options, or the availability of adjoint solvers. A common method to mitigate some of this computational expense in nongradient optimization is to perform dimensionality reduction on the data and optimize the design within this smaller subspace. Although learning these low-dimensional airfoil manifolds often facilitates aerodynamic optimization, these subspaces are often still computationally expensive to explore. Moreover, the complex data organization of many current nonlinear models make it difficult to reduce dimensionality without model retraining. Inducing orderings of latent components restructures the data, reduces dimensionality reduction information loss, and shows promise in providing near-optimal representations in various dimensions while only requiring the model to be trained once. Exploring the response of airfoil manifolds to data and model selection and inducing latent component orderings have potential to expedite airfoil design and optimization processes. This thesis first investigates airfoil manifolds by testing the performance of linear and nonlinear dimensionality reduction models, examining if optimized geometries occupy lower dimensional manifolds than non-optimized geometries, and by testing if the learned representation can be improved by using target optimization conditions as data set features. We find that autoencoders, although often suffering from stability issues, have increased performance over linear methods such as PCA in low dimensional representations of airfoil databases. We also find that the use of optimized geometry and the addition of performance parameters have little effect on the intrinsic dimensionality of the data. This thesis then explores a recently proposed approach for inducing latent space orderings called Rotation Augmented Gradient (RAG) [1]. We extend their algorithm to nonlinear models to evaluate its efficacy at creating easily-navigable latent spaces with reduced training, increased stability, and improved design space preconditioning. Our extension of the RAG algorithm to nonlinear models has potential to expedite dimensional analyses in cases with near-zero gradients and long training times by eliminating the need to retrain the model for different dimensional subspacesItem DESIGN AND EXPERIMENTAL CHARACTERIZATION OF METAL ADDITIVE MANUFACTURED HEAT EXCHANGERS FOR AEROSPACE APPLICATION(2020) Battaglia, Fabio; Ohadi, Michael; Mechanical Engineering; Digital Repository at the University of Maryland; University of Maryland (College Park, Md.)High temperature heat exchangers are key to the success of emerging high-temperature, high-efficiency solutions in energy conversion, power generation and waste heat recovery applications. When applied to the aerospace applications, the main objective is to develop heat exchangers that can realize significant performance improvement in terms of gravimetric heat exchange density (kW/kg). In the present study, two air-to-air crossflow heat exchangers were designed, built and tested to determine their potential for high performance, pre-cooling heat exchanger for aircraft applications. A novel design based on manifold-microchannel technology was chosen as it provided localized and optimum distribution of the flow among the heat transfer surface micro channels, offering superior heat transfer performance and low pressure drops, when compared to conventional, state of the art heat exchangers for the chosen application. However, fabrication of the manifold microchannel design for high temperature with super alloys as the heat exchanger material presents serious manufacturing challenges fabrication techniques. To overcome this limit, direct metal laser sintering (DMLS) additive manufacturing technique was selected for the fabrication of the Ni-based superalloy manifold-microchannel heat exchangers in the present study. Extensive work was performed to characterize the printing capability of different metal 3D-printers in terms of printing orientation, printing accuracy and structure density. Based on the knowledge acquired, two units were printed, with overall size of 4”x4”x4” and 4.5”x4”x3.5” and fin thickness of 0.220 mm and 0.170 mm, respectively. The printed units were the largest additively printed, superalloy-based manifold-microchannel heat exchangers found in the literature. The experimental characterization was carried at high temperature (600°C) and the model prediction of the performance was updated to characterize the behavior of the heat exchangers in this operational conditions. Based on the experimental results, a gravimetric heat duty of 9.4 kW/kg for an effectiveness (ε) of 78% was achieved, which corresponds to an improvement of more than 50% compared to the conventional designs. The characterization of the performance at high temperature was then completed by analyzing the thermo-mechanical stress generated by the simultaneous presence of temperature gradient and pressures. The current study is the first to characterize the behavior of manifold-microchannel heat exchanger under high temperature in terms of performance prediction and thermo-mechanical analysis.Item PHYSICS OF LAMINAR PREMIXED CH4 − O2 FLAMES AT CRYOGENIC CONDITIONS - A COMPUTATIONAL STUDY(2019) Gopal, Abishek; Larsson, Johan; Mechanical Engineering; Digital Repository at the University of Maryland; University of Maryland (College Park, Md.)With increased commercial spaceflight activity, methane has found adoption in the next generation of liquid rocket engines (LREs). In a liquid rocket engine with cryogenic propellants, such as methane and oxygen, the propellants are stored in their tanks at low temperatures. As they are injected into the combustion chamber at high pressures, the fluid is close to its thermodynamic critical point where there are drastic changes in fluid properties like density, heat capacity, surface tension, and solubility. The ideal gas law is inapplicable at such extreme conditions, and real gas thermodynamic and transport properties are required to accurately model the combustion physics at supercritical conditions. Much of the previous work applying real gas models in computational simulations of reacting flows have focused on non-premixed flames or cold-flow mixing configurations. In this study, we investigate the effects of real gas property estimation on planar, unstretched, laminar premixed methane-oxygen flames at transcritical conditions. The computational framework used in this study integrates real gas property estimation into the steady-state, freely-propagating flame solver available in the Cantera combustion suite. The Peng-Robinson equation of state provides thermodynamic property closure. High-pressure transport properties are modeled by the Chung and Takahashi correlations, respectively. The effects on laminar flame structure are presented. We find that enhanced real gas reactant densities have a significant impact on flame propagation, lowering flame speeds by a factor of ∼ 5 near the critical region. Real gas caloric properties lower mass burning rates by 10%. The consequence of using low-pressure transport properties with the Peng-Robinson EOS at variable Lewis numbers is discussed.Item EFFECTS OF GLASS/EPOXY INTERPHASES ON ELECTRO-CHEMICAL FAILURES IN PRINTED CIRCUIT BOARDS(2018) Sood, Bhanu Pratap; Pecht, Michael G; Mechanical Engineering; Digital Repository at the University of Maryland; University of Maryland (College Park, Md.)Reduction in printed circuit board line spacing and via diameters and the increased density of vias with higher aspect ratios (ratio between the thickness of the board and the size of the drilled hole before plating) are making electronic products increasingly more susceptible to material and manufacturing defects. One failure mechanism of particular concern is conductive anodic filament formation, which typically occurs in two steps: degradation of the resin/glass fiber bond followed by an electrochemical reaction. The glass-resin bond degradation provides a path along which electrodeposition occurs due to electrochemical reactions. Once a path is formed, an aqueous layer, which enables the electrochemical reactions to take place, can develop through the adsorption, absorption, and capillary action of moisture at the resin/fiber interphase. This study describes the experimental and analytical work undertaken to understand the glass-resin delamination and the methods used for analyzing this critical interphase. This study shows that a smaller conductor spacing in reduces the time to failure due to conductive anodic filament formation and that the plated-through-hole to plated-through-hole conductor geometry is more susceptible to conductive anodic filament-induced failures than plated through hole to plane geometries. The results also show that laminates with similar materials and geometries with a 45-degree angle of weave demonstrate a higher resistance to conductive anodic filament formation compared with a 90-degree angle of weave. The study is the first of its kind conducted on FR-4 printed circuit board materials where the pathway formation due to breakage of the organosilane bonds at the glass/resin interphase was evaluated. Using techniques such as force spectroscopy, micro-Fourier transform infrared spectroscopy, scanning quantum interface device microscopy and focused ion beam, evidence of bond breakage and a pathway formation was revealed, poor glass treatment, hydrolysis of the silane glass finish (adsorption of water at the glass fiber/epoxy resin interphase) or repeated thermal cycling contribute to the bond breakage. The technique of applying in-situ resistance measurements during cross-sectioning analysis of printed circuit boards suspected of conductive anodic filament is the first time this method is described in the open literature. This solution addresses the potential problem in destructive physical analysis of grinding away the evidence of the CAF filament and ultimately loosing evidence at the failure site. By applying a subset of the evaluation criteria described in this research, an upfront evaluation of printed circuit board materials can be performed for susceptibility to electro-chemical migration and other failure causes in PCBs that are attributable to the glass/resin interfacial adhesion. Manufacturers can identify board suppliers based on answers to and validation of a series of questions. These questions focus on the necessary requirements of reliable board material manufacturing and are independent of the specifications of the product.Item Strategies For Enhancing Performance of Flapping Wing Aerial Vehicles Using Multifunctional Structures and Mixed Flight Modes(2018) Holness, Alex; Bruck, Hugh A; Mechanical Engineering; Digital Repository at the University of Maryland; University of Maryland (College Park, Md.)Biological flapping wing flight offers a variety of advantages over conventional fixed wing aircraft and rotor craft. For example, flapping propulsion can offer the speed of fixed wing aircraft at similar scales while providing the maneuverability of rotor craft. Avian species easily display feats of perching, payload carrying, endurance flying, and transition behavior. In light of these characteristics, emulating and recreating flapping flight in biomimetic or bioinspired work is important in the development of next generation aerial systems. Unfortunately, recreating flapping wing flight is not easily achieved despite numerous efforts to do so. This is in large part due to technological deficiencies. With emerging technologies, it has been possible to begin to unravel the intricacies of flapping flight. Despite technological advancements, offsetting weight with mechanical systems robust enough to provide power and torque while sustaining loading remains difficult. As a result platforms either have simple flapping kinematics with fair payload or have more complex kinematics with limited excess power which in turn limits payload. The former limits capabilities to mirror biological performance characteristics and the latter limits the energy available to power flight which ultimately negatively impacts mission capabilities. Many flapping wing systems are subpar to traditional flying vehicles. Flapping systems can become more competitive in achieving various mission types with increased system performance. In particular, if endurance is coupled with desirable features such as those displayed in nature, i.e., avian perching, they may become superior assets. In this work, four strategies for increasing performance were pursued as follows: (1) increases to maneuverability and payload via a mixed mode approach of flapping wing used in conjunction with propellers, (2) rapid deceleration and variation of flight envelope via inertial control using the battery, (3) increased endurance via integrated energy storage in the wings, and (4) providing endurance to the point of complete energy autonomy using a design framework considering flapping wings with integrated high efficiency solar cells.Item IMPROVED PREDICTION OF FLAPPING WING AERIAL VEHICLE PERFORMANCE THROUGH COMPONENT INTERACTION MODELING(2018) Gerdes, John William; Gupta, Satyandra K; Mechanical Engineering; Digital Repository at the University of Maryland; University of Maryland (College Park, Md.)Flapping wing aerial vehicles offer the promise of versatile performance, however prediction of flapping wing aerial vehicle performance is a challenging task because of complex interconnectedness in vehicle functionality. To address this challenge, performance is estimated by using component-level modeling as a foundation. Experimental characterization of the drive motors, battery, and wings is performed to identify important functional characteristics and enable selection of appropriate modeling techniques. Component-level models are then generated that capture the performance of each vehicle component. Validation of each component-level model shows where errors are eliminated by capturing important dynamic functionality. System-level modeling is then performed by creating linkages between component-level models that have already been individually validated through experimental testing, leading to real-world functional constraints that are realized and correctly modeled at the system level. The result of this methodology is a system-level performance prediction that offers the ability to explore the effects of changing vehicle components as well as changing functional properties, while maintaining computational tractability. Simulated results are compared to experimental flight test data collected with an instrumented flapping wing aerial vehicle, and are shown to offer good accuracy in estimation of system-level performance properties.Item Development and Application of Mach 10 PIV in a Large Scale Wind Tunnel(2018) Brooks, Jonathan; Gupta, Ashwani K; Marineau, Eric C; Mechanical Engineering; Digital Repository at the University of Maryland; University of Maryland (College Park, Md.)This dissertation presents the development of particle image velocimetry (PIV) for use in a large-scale hypersonic wind tunnel to measure the turbulent boundary layer (TBL) and shock turbulent boundary layer interaction (STBLI) on a large hollow cylinder flare (HCF) test article. The main feature of this application of PIV is the novel local injector which injects seeding particles into the high-speed section of the flow. Development work began sub-scale in a Mach 3 wind tunnel where the seeding particle response was characterized and the local injectors were demonstrated. Once the measurement technique was refined, it was scaled up to hypersonic flow. The particle response was characterized through PIV measurements of Mach 3 TBLs under low Reynolds number conditions, $ Re_\tau=200{-}1,000 $. Effects of Reynolds number, particle response and boundary layer thickness were evaluated separately from facility specific experimental apparatus or methods. Prior to the current study, no detailed experimental study characterizing the effect of Stokes number on attenuating wall normal fluctuating velocities has been performed. Also, particle lag and spatial resolution are shown to act as low pass filters on the fluctuating velocity power spectral densities which limit the measurable energy content. High-speed local seeding particle injection has been demonstrated successfully for the first time. Prior to these measurements, PIV applications have employed global seeding or local seeding in the subsonic portion of the nozzle. The high-speed local seeding injectors accelerate the particle aerosol through a converging/diverging supersonic nozzle which exits tangentially to the wall. Two methods are used to measure the particle concentration which shows good agreement to the CFD particle tracking codes used to design the injector nozzle profiles. Based on the particle concentration distribution in the boundary layer a new phenomenon of particle biasing has been identified and characterized. PIV measurements of a Mach 10 TBL and STBLI have been performed on a large (2.4-m long, 0.23-m dia.) HCF at a freestream unit Reynolds number of 16 million per meter. These are the highest Mach number PIV measurements reported in the literature. Particles are locally injected from the leading edge of the test article and turbulent mixing dispersed the particles for a relatively uniform high concentration of particles at the measurement section 1.83-m downstream of the leading edge. The van Driest transformed mean velocity in the TBL agrees well with incompressible zero pressure gradient log law theory. Morkovin-scaled streamwise velocity fluctuations agree well with the literature for the majority of the boundary layer.Item DESIGN AND TESTING OF A HIGH-POWER PNEUMATIC ANKLE EXOSKELETON(2018) Geating, Josh; Wereley, Norman; Mechanical Engineering; Digital Repository at the University of Maryland; University of Maryland (College Park, Md.)A pneumatic ankle exoskeleton is designed, constructed, and tested to study the use of pneumatic artificial muscles (PAMs) and force sensitive resistors (FSRs) for increased running and jumping performance. PAMs are selected for their low weight, high power density, and natural compliance, all advantageous for exoskeleton applications. FSRs are selected as they can be packaged between the sole of the wearer’s foot and a ground plate, enabling toe push-off force to be measured. Toe force measurement is an uncommon parameter to be measured on most lower extremity exoskeletons. A closed loop force controller is implemented using the FSRs as an input force and a strain sensor embedded in the PAM as the feedback sensor. This architecture is shown to achieve a high bandwidth, capable of following trajectories similar to that of a running gait or jumping force profile. The FSRs are shown to exhibit low hysteresis and high dynamic response, and moderate linearity compared to traditional strain gauge based transducers (SGBTs). Special attention is given to keep the structure realistically lightweight, as well as select and characterize sensors that could be realistically packaged in an end-use application.