A large eddy simulation of an enclosed annular rotor stator cavity is presented. The

geometry is characterized by a large aspect ratio G = (b-a)/h = 18.32 and a small radius

ratio a/b = 0.152, where a and b are the inner and outer radii of the rotating disk and h is

the interdisk spacing. The rotation rate ω under consideration is equivalent to the rotational

Reynolds number Re = ω b^{2 /ν= 9.5x104 , where ν is the kinematic viscosity.}

The main objective of this study is to correctly simulate the rotor stator cavity using a low

order numerical scheme on unstructured grids. The numerical simulations were run on the

software AVBP developed by the Centre Européen de Recherche et de Formation Avancée en Calcul Scientific. The results were compared to the experimental results obtained by Sebastien Poncet of Universit é Aix-Marseille. Two large eddy simulations techniques were used: the Smagorinsky and Wall-adapting local eddy-viscosity models. The simulations were run on three set of grids, each with a different cell resolution-14, 35 and 50- along the thickness of the system. Results from each mesh show a good qualitative agreement of the mean velocity field with Poncet's experimental results. It was found that the Samgorinsky model to be more appropriate for this configuration.

Inviscid vortex models have been used for decades to investigate unsteady aerodynamics. However, real-time use of these models has been hindered by the tradeoff between increasing a model's dynamical capability and reducing its dimensionality. In this work, we present two different solutions to this problem. First, we develop a hybrid model where vortex sheets represent shear layers that separate from the wing and point vortices represent the rolled-up cores of these shear layers and the other coherent vortices in the wake. Instead of rolling up, each vortex sheet feeds its circulation into a point vortex using a circulation transfer procedure we developed. This procedure eliminates the spurious force that manifests when transferring circulation between vortex elements. By tuning the rate at which the vortex sheets are siphoned into the point vortices, we can adjust the balance between the model's dimensionality and dynamical richness. This hybrid model can capture the development and subsequent shedding of the starting vortices in real time, and remain low-dimensional enough to simulate long time horizon events such as periodic bluff-body shedding. Our second solution augments a vortex blob model with surface pressure measurements using the ensemble Kalman filter (EnKF). We adapt our circulation transfer procedure to aggressively aggregate the vortex blobs in order to prevent the dimension of the system from increasing indefinitely. The reduced number of blobs, along with the parallel nature of the EnKF, allow this solution to also run in real time. We find that not only does the data assimilation process compensate for the severe reduction in dimension, it also seems to fill in some missing physics from our inviscid model.

Ischemic stroke is one of the leading causes of death in the US. It is estimated that intracranial stenosis, which refers to the narrowing of the blood vessels within the brain, accounts for 30% - 50% of ischemic strokes. Typical intracranial stenosis includes narrowing of arteries such as the middle cerebral artery (MCA) and the internal carotid artery (ICA). Among alltypes of intracranial stenosis, middle cerebral artery (MCA) stenosis poses a higher risk for ischemic stroke. Collateral circulation occurs when occlusion happens and plays a vital role in sustaining cerebral blood flow during ischemic strokes. However, the factors leading to the formation of collateral circulation remain unclear and are difficult to observe through image data alone. The imaging data provides only a snapshot, thereby limiting our understanding of stenosis development and the formation of collateral circulation. This work developed a computational tool that integrates patient-specific image data with fluid dynamics to investigate the development of MCA stenosis and explore possible causes of collateral circulation between MCA and PComm.

The hemodynamic data of arterial segments is extracted from individual computed tomography angiography (CTA) scan. The extracted network includes parent arterial segment ICA and daughter arterial segments MCA, ACA and PComm. The inlet blood flow profile is calibrated using color-coded duplex ultrasound from young, healthy volunteers, while a three-element Windkessel outlet boundary condition is attached to the end of each daughter arterial segment to simulate artificial wave reflection from the downstream vascular network. Our findings demonstrate that increases in area reduction increase the pressure of the unobstructed pathway and decrease the pressure of the obstructed pathway, which becomes significant after 90% of the MCA area is occluded, aligning with clinical observations. Wave Intensity Analysis (WIA) results reveal that the forward expansion wave (FEW) accelerates the pressure and blood flow from the downstream vascular network, particularly when MCA is highly occluded (90% cross-sectional area is occluded). The pressure/blood flow and WIA results show that the formation of collateral circulation is closely related to the high resistance of the upstream large artery, which aligns with the clinical hypothesis.

Furthermore, this work compares the two types of outlet boundary conditions: the three-element Windkessel 0-D outlet boundary condition and the 1-D binary tree outlet boundary condition. To assess their impact on the pressure and blood flow waveform of the upstream large artery, both types of outlet boundary conditions are connected to the upstream large artery. The findings reveal that the two types of outlet boundary conditions exhibit minimal differences in affecting the pressure/blood flow of upstream large artery. However, it should be noted that the utilization of the 1-D outlet boundary conditions is limited in the presented model due to the long-wave approximation.This work introduces a novel model that includes one parent arterial segment connected to three daughter arterial segments. It is the first time that the collateral circulation between MCA and PComm through the downstream vascular network is investigated. Additionally, the WIA method is employed for the first time to study the relationship between MCA stenosis and the formation of collateral circulation.

In summary, a computational tool is developed to understand the evolution of hemodynamic changes in cerebral blood flow and to reveal pathology of collateral circulation and MCA stenosis.

Lean premixed prevaporized (LPP) gas turbine generators have become popular in energy conversion applications to meet strict emission requirements. Because the combustion process is very lean, combustion instabilities due to acoustic perturbations are more likely to occur than in a less lean fuel combustion process. Current design of damping strategies for mitigating these instabilities is often based on empirical trial and error, which precludes the possibility of determining an optimal configuration. A combustion system whose elements consist of flames, passive dampers, and ducts must be optimized to reduce or completely eliminate combustion instabilities. Hence, a modular simulation tool is developed to examine the interaction of plane acoustic waves with typical combustion system elements. The simulation tool represents these interactions in the form of transfer matrices, which can be modularly arranged for exploring a variety of configurations. In this work, a heuristic gain-delay flame model is represented as a transfer matrix, which can be used to test damping devices. Similarly, transfer matrices representing a Helmholtz resonator and a perforated liner with bias flow are developed, and preliminary results are obtained.

Flapping flight has gained much attention in the past decade driven by the desire to understand capabilities observed in nature and the desire to develop agile small-scale aerial vehicles. Advancing our current understanding of unsteady aerodynamics is an essential component in the development of micro-air vehicles (MAV) intended to utilize flight mechanics akin to insect flight. Thus the efforts undertaken that of bio-mimicry. The complexities of insect wing motion are dissected and simplified to more tractable problems to elucidate the fundamentals of unsteady aerodynamics in biologically inspired kinematics. The MAV's fruition would satisfy long established needs in both the military and civilian sectors. Although recent studies have provided great insight into the lift generating mechanisms of flapping wings the deflection response of such wings remains poorly understood. This dissertation numerically and experimentally investigates the aerodynamic performance of passively and actively deflected wings in hover and rotary kinematics. Flexibility is distilled to discrete lines of flexion which acknowledging major flexion lines in insect wings to be the primary avenue for deformation. Of primary concern is the development of the leading-edge vortex (LEV), a high circulation region of low pressure above the wing to which much of the wing's lift generation is attributed. Two-dimensional simulations of wings with chord-wise flexibility in a freestream reveal a lift generating mechanism unavailable to rigid wings with origins in vortical symmetry breaking. The inclusion of flexibility in translating wings accelerated from rest revealed the formation time of the initial LEV was very weakly dependent on the flexible stiffness of the wing, maintaining a universal time scale of four to five chords of travel before shedding. The frequency of oscillatory shedding of the leading and trailing-edge vortices that develops after the initial vortex shedding was shown to be responsive to flexibility satisfying an inverse proportionality to stiffness. In hover, an effective pitch angle can be defined in a flexible wing that accounts for deflection which shifts results toward trend lines of rigid wings. Three-dimensional simulations examining the effects of two distinct deformation modes undergoing prescribed deformation associated with root and tip deflection demonstrated a greater aerodynamic response to tip deflection in hover. Efficiency gains in flexion wings over rigid wing counterpart were shown to be dependent on Reynolds number with efficiency in both modes increasing with increased Reynolds number. Additionally, while the leading-edge vortex axis proved insensitive to deformation, the shape and orientation of the LEV core is modified. Experiments on three-dimensional dynamically-scaled fruit fly wings with passive deformation operating in the bursting limit Reynolds number regime revealed enhanced leading-edge vortex bursting with tip deflection promoting greater LEV core flow deceleration in stroke. Experimental studies on rotary wings highlights a universal formation time of the leading-edge vortex independent of Reynolds number, acceleration profile and aspect ratio. Efforts to replicate LEV bursting phenomena of higher aspect ratio wings in a unity aspect ratio wing such that LEV growth is no limited by span but by the LEV traversing the chord revealed a flow regime of oscillatory lift generation reminiscent of behavior exhibited in translating wings that also maintains magnitude peak to peak.

Large amplitude flow disturbances and gusts can drastically alter the aerodynamic forces on airfoils or structures. The modeling and control of the aerodynamic response of the flow around the body is complicated by the inherent nonlinearities and high dimensionality of this system, which increase the cost and complexity of the numerical tools that enable this. This work introduces two numerical tools for the efficient modeling of unsteady aerodynamic flows and explores the use of deep reinforcement learning to perform airfoil pitch control during flow disturbances. The first numerical tool we introduce is a grid-based potential flow solver. We focus on the discrete streamfunction and model the flow around bodies and point vortices immersed in a cartesian grid. This tool reformulates the inviscid vortex-in-cell method as a saddle-point problem where constraints such as the no-flow-through and Kutta condition can be added as Lagrange multipliers. The second numerical tool uses a potential flow to model wind tunnel walls and irrotational wind tunnel gust generators in a viscous flow simulation. This technique allows us to still accurately model the flow around the test subject while accounting for the effects of the wind tunnel in a way that doesn't drastically increase the cost of the simulation. Lastly, in our explorative study of reinforcement learning of airfoil pitch control, we try out the training of a control policy in a classical unsteady aerodynamics environment and a viscous, low-Reynolds number flow environment to minimize the lift variations caused by flow disturbances and compare the performance of controllers, or agents, that observe different types of information about the states of the system.

Obstructive sleep apnea (OSA) is a medical condition characterized by repetitive partial or complete occlusion of the airway during sleep. The soft tissues in the airway of OSA patients are prone to collapse under the low pressure loads incurred during breathing. A numerical tool suitable for simulation of air-tissue interactions with patient-specific upper airway geometries has been developed. The tool is designed to utilize non-body-fitted Cartesian grid to avoid grid generation and remeshing processes for the deforming irregular geometries. A sharp-interface embedded boundary method is used resolving the airflow in the airway geometries, while a novel cut-cell non-linear finite element analysis is constructed for structure undergoing large deformation. The separate fluid and structure solvers are strongly coupled with a partitioned iterative algorithm. The interface is represented with the signed distance function in the flow and structural solvers. Also, the explicit interface representation with the union of Lagrangian triangle elements is introduced to manage the data exchange between two solvers. The simulation of deformation of cantilever beam under uniformly distributed load demonstrated the ability of the structural solver. The capability of the developed tool dealing with flow induced deformation problems has been exhibited with several two- and three- dimensional validation cases. To obtain the patient-specific airway and soft tissue geometries, a procedure by which patient-specific airway geometries are segmented and processed from dental cone-beam CT scans into signed distance fields is presented. The results are shown for three-dimensional flow simulation inside two rigid patient-specific airways. The simulation results of fluid--structure interaction of the pressure driven flow surrounded by deformable soft palate and tongue structures are also presented.

Modeling the multiphase mechanics with coupled fluid and elastic material is important for many applications such as blood perfused soft tissues, wicking in porous medium. The liquid-solid interaction results in complicated effect on structure deformation and liquid transportation. The current study aims to develop high visual and physical fidelity simulations of multiphase mechanics, particularly within the context of soft tissue swelling, human injuries, medical treatments, the transport of blood through damaged tissue under bleeding or hemorrhaging conditions and droplet spreading on a fabric. The solid material is considered as a dynamic poro-hyperelastic material with liquid-filled voids. A biphasic formulation---effectively, a generalization of Darcy's law---is utilized, treating the phases as occupying fractions of the same volume. A Stokes-like friction force, a pressure that penalizes deviations from volume fractions summing to unity and the surface tension between multiphase interface, serve as the interaction force between solid and liquid phases. The resulting equations for both phases are discretized with the method of Smoothed Particle Hydrodynamics (SPH). The solver is validated separately on each phase and demonstrates good agreement with exact solutions in test problems. Simulations of oozing, hysteresis, swelling, drying and shrinkage, tissue fracturing and hemorrhage, liquid droplet spreading on a fabric are shown in this work. Besides the physical-based SPH solver, the technique called dynamic mode decomposition (DMD) from data science also applies on the results from SPH solver to extract the system features without any knowledge of governing equations, providing benefits such as data compression and efficient data manipulation, raising the potential of developing data-driven computational solver in the future.