Browsing by Subject "CFD"

Now showing 1 - 13 of 13

Advanced simulation and modeling of turbulent sprays
(2014-03) Liu, Wanjiao
Sprays have wide applications in agriculture, pharmaceutical synthesis, engines, ink jet printing and so on. The successful spray applications and the control of spray param- eters require a thorough understanding towards the physical mechanisms. Numerical tools have been developed in the past few years for simulating the multiphase turbu- lent flows like sprays. Several researchers have successfully carried out direct numerical simulations (DNS) to investigate the primary breakup in such flows. DNS is accurate but requires extensive computational resources. In comparison, large eddy simulation (LES) is more practical, resolving only the large-scale flow structures and modeling the small-scale effects. The major difficulty with LES of multiphase turbulent flows is the need to model the interfacial subgrid-scale terms. Subgrid surface tension force, for ex- ample, plays an important role in the small droplet formation process. Subgrid surface tension force is, however, a highly non-linear term and can be difficult to model. In this research, we propose a new approach that combines the filtered density function (FDF) approach with the large eddy simulation. The major advantage of FDF is that the non-linear surface tension force appears in a closed form and thus needs no sub- grid modeling. The FDF transport equation is solved conveniently via a Lagrangian Monte-Carlo method. The Lagrangian approach is attractive in that it facilitates the transport of the liquid-gas interface without the diffusive or dispersive errors found in the Eulerian approaches. The surface tension source term in the momentum equation is closed using a Lagrangian volume of fluid (LVOF) approach. We utilize concepts from the smoothed particle hydrodynamics (SPH) in the LVOF approach to obtain the surface tension source term based on the Lagrangian particles. Several modifications have been made towards the original SPH formulation such that it is more suitable for the large-scale, turbulent multiphase flow simulations. Multiple particles are seeded in each Eulerian cell to achieve higher statistical accuracy, while the original SPH seeds one particle in each cell. What's more, a weighted SPH formula for the color function is adopted and is shown to be capable of handling variable particle number density. Performance assessment is via the rotation of Zalesak's disk and an oscillating elliptical droplet. Results show that the modified approach is able to handle the variable particle number density case appropriately. The simulations of multiphase turbulent flows are then performed with the proposed FDF-VOF methodology. At the same time, results from the simulations are compared with the DNS approach for validation and com- parison. Results show that the FDF-LES based approach can be a promising method, in that it models the flow with lower computational cost than DNS, yet maintaining accuracy in a model-free manor.
Boiling of Dilute Emulsions.
(2010-06) Roesle, Matthew Lind
Although boiling in pure liquids has been studied thoroughly, boiling in other circumstances is less well understood. One area that has received little attention is boiling of dilute emulsions in which the dispersed component has a lower boiling point than the continuous component. These mixtures exhibit several surprising behaviors that were unknown until the 1970's. Generally, boiling of the dispersed component enhances heat transfer over a wide range of surface temperatures without transition to film boiling, but a high degree of superheat is required to initiate boiling. In single-phase convection the dispersed component has little effect on heat transfer. These behaviors appear to occur in part because few droplets in the emulsion contact nucleation sites on the heated surface. No detailed and physically consistent model of boiling in dilute emulsions exists at present. The unusual behavior of boiling dilute emulsions makes them potentially useful for high heat flux cooling of electronics. High-power electronic devices must be maintained at temperatures below ~85 °C to operate reliably, even while generating heat fluxes of 100 W/cm2 or more. Current research, generally focusing on single phase convection or flow boiling in small diameter channels, has not yet identified an adequate solution. An emulsion of refrigerant in water would be well-suited to this application. The emulsion retains the high specific heat and thermal conductivity of water, while boiling of the refrigerant enhances the heat transfer coefficient at temperatures below the saturation temperature of water. To better understand boiling dilute emulsions and expand the experimental database, an experimental study of boiling heat transfer from a horizontal heated wire, including visual observations, is performed. Emulsions of pentane in water and FC-72 in water are studied. These emulsions have properties suitable for practical use in high heat flux cooling applications, unlike most emulsions that have previously been studied. The range of the experimental study is extended to include enhanced boiling of the continuous component, which has not previously been observed, in addition to boiling of the dispersed component. In both regimes the heat transfer coefficient is enhanced compared to that of water. Visual observations reveal the presence of large attached bubbles on the heated wire, the formation of which coincides with the inception of boiling in the heat transfer data. At very low dispersed component fractions and low temperatures, boiling of individual dispersed droplets is not observed. The large attached bubbles represent a new boiling mode that has not been reported in previous studies and is, under some circumstances, the dominant mode of boiling heat transfer. A model of boiling dilute emulsions is developed based upon the Euler-Euler model of multiphase flows. The general balance equations as developed by Drew and Passman are applied to the present situation, thus providing a rigorous and physically consistent framework. The model contains three phases that represent the continuous component, liquid droplets of the dispersed component, and bubbles that result from boiling of individual droplets. Mass, momentum, and energy transfer between the phases are modeled based upon the behavior of and interaction between individual elements of the dispersed phases. One-dimensional simulations of a single boiling droplet in superheated liquid are also performed, and the results are used to develop the closure equations of the larger model. Droplet boiling is assumed to occur when a droplet contacts a heated surface or a vapor bubble. Collisions between droplets and bubbles and chain-boiling of closely-spaced droplets are considered. The model is limited to the dispersed component boiling regime, and thus it does not account for phase change of the continuous component. The model also does not include the large attached bubbles revealed in the visualization experiments. However, simulations of boiling match several trends observed in the experimental data. The model thus provides a physically consistent and partially validated platform for future analytical and numerical work.
CFD Simulations and Thermal Design for Application to Compressed Air Energy Storage
(2015-06) Zhang, Chao
The present computational research focuses on fluid flow analysis and heat transfer enhancement in support of the design of a hydraulic Compressed Air Energy Storage (CAES) system. A CAES system compresses air to high pressure during high power generation periods, stores the compressed air, and expands it to generate power during high power demand periods. The main benefit of using CAES is that it overcomes the mismatch between power generation and power demand. An innovative liquid piston method is used in the present research, where liquid (water) is pumped into the lower section of a compression chamber, and the gas (air) is compressed by the rising liquid-gas interface. Important to the efficient operation of CAES is reducing the temperature rise during compression. The work input process for compressing the air requires two main steps. In the first step, compression work compresses the air to a high pressure. The air temperature rises during compression, and this leads to a second step, where the compressed air cools. In order to maintain the work potential, the pressure of the compressed air is maintained during cooling, volume decreases and cooling work is done. As a result, higher temperature rise during compression requires greater amount of total work input for a given amount of air and a given pressure ratio. For similar reasons, it is also desirable to reduce the temperature drop during expansion of compressed air. The use of a liquid piston offers opportunities to insert heat exchanger matrices into the compression chamber to improve heat transfer. Computational Fluid Dynamics (CFD) and design analyses on the heat exchangers are done in the present research. Two main types of heat exchangers, a commercially available, open-cell metal foam and an in-house designed interrupted plate matrix, are investigated, mainly through computational methods, and with experimental validations. CFD modeling of the liquid piston chamber inserted with exchanger matrices requires closure models that characterize the heat transfer and flow resistance characteristics of the heat exchanger elements. For the metal foam matrix, characterization is done by measuring the flow pressure drop and comparing existing heat transfer models to a liquid piston experiment. For the interrupted plate matrix, large numbers of CFD simulations on the unit cells of the exchanger and experiments are applied for developing correlations for these terms. Based on the unit cell simulations, models for three-dimensionally anisotropic heat transfer behavior of porous media have been developed. Using these models, 3-D global-scale CFD simulations of the liquid piston chamber have been done. As the liquid chamber represents an application of two-phase flow through porous media, the simulation combines a VOF (Volume of Fluid) method and a two-energy-equation modeling method for porous media. The choice of the interrupted plate matrix offers flexibility to vary the shape (e.g. plate height, thickness and separation distance) based on an optimum design to further improve CAES efficiency. Design sensitivity analyses typically require large numbers of computational runs and would demand extraordinary computational resources if combined with 2-D or 3-D CFD simulations. Developed in the present research is a simplified, one-dimensional (1-D) code that is much less computationally expensive, and preserves the main physics of the two-phase flow in porous media. The 1-D code is used for the design analysis of the heat exchanger shape distribution along the axial direction of the chamber. CFD Simulations have been done to also compare a no-insert chamber to chambers with exchanger inserts, for both compression and expansion processes. The expansion process allows the expanding compressed gas to push the liquid out of the chamber to generate power. It is shown that in both the compression and expansion processes, using a heat exchanger matrix in the liquid piston chamber can significantly reduce the rise or drop in gas temperature, thus reducing the losses. Using the CFD modeling tools, a design exploration is also done to investigate the effect of changing the profile of the chamber's cross sectional radius along the axial direction to design a gourd-like shaped chamber to agitate flow and enhance heat transfer.
Computational Study of Shock/Plume Interactions Between Multiple Jets in Supersonic Crossflow
(2016-08) Tylczak, Erik
The interaction of multiple jets in supersonic crossflow is simulated using hybrid Reynolds- Averaged Navier Stokes and Large Eddy Simulation turbulence models. The blockage of a jet generates a curved bow shock, and in multi-jet flows, each shock impinges on the other fuel plumes. The curved nature of each shock generates vorticity directly, and the impingement of each shock on the vortical structures within the adjacent fuel plumes strengthens vortical structures already present. These stirring motions are the major driver of fuel-air mixing, and so mixing enhancement is predicted to occur in multi-port configurations. The primary geometry considered is that of the combustion duct at the Calspan- University of Buffalo Research Center 48” Large Energy National Shock (LENS) tunnel. This geometry was developed to be representative of the geometry and flow physics of the Flight 2 test vehicle of the Hypersonic International Flight Research Experimenta- tion Program (HiFIRE-2). This geometry takes the form of a symmetric pair of external compression ramps that feed an isolator of approximately 4” × 1” cross-section. Nine interdigitated flush-wall injectors, four on one wall and five on the other, inject hydrogen at an angle of 30 degrees to the freestream. Two freestream flow conditions are consid- ered: approximately Mach 7.2 at a static temperature of 214K and a density of 0.039 kg/m3 for the five-injector case, and approximately Mach 8.9 at a static temperature of 167K and density of 0.014 kg/m3 for the nine-injector case. Validation computations are performed on a single-port experiment with an imposed shock wave. Unsteady calculations are performed on five-port and nine-port configura- tions, and the five-port configuration is compared to calculations performed with only a single active port on the same geometry. Analysis of statistical data demonstrates enhanced mixing in the multi-port configurations in regions where shock impingement occurs.
Conjugate Heat Transfer Simulations for Hypersonic Vehicles
(2020-08) Reinert, John
The accurate prediction of thermal responses is important for optimizing the design and operability for hypersonic flight vehicles. In order to efficiently simulate this process, a loosely coupled conjugate heat transfer solver was developed. Conjugate heat transfer simulations involve fluid and solid solvers. The fluid solver computes the flow field over the vehicle, and the solid solver calculates the transient heat conduction into the vehicle body. The two solvers are ``loosely'' coupled because both solvers exchange information at the surface of the vehicle, but operate on different time scales. The present work details the derivation of the conjugate heat transfer solver. The simulations were performed with US3D, an implicit finite volume unstructured compressible flow solver, with a newly developed implicit finite element transient heat conduction solver. The finite element solver is verified by comparing with analytical solutions for a bar, cylinder, and sphere. Validation cases for two geometries are shown: a fin-cone and HIFiRE-1. Both cases were shown to match well with the experimental data and flight test data. Additionally, the finite element method is compared to a finite volume method for solving the transient heat conduction equation. The comparison showed the benefits of the finite element method, such as refined temperature distribution and improved grid independence. Finally, the boundary layer transition (BoLT) vehicle is simulated for a segment of the trajectory. Results show the heating of the leading edge through time and the three-dimensional heating of the vehicle. At a specific time in the trajectory, the boundary layer and flow field are investigated. A comparative study is performed for the variable wall temperature and isothermal wall flow fields. The variable wall temperature was found to affect the wall heat flux and flow field structures. These results highlight the importance of performing conjugate heat transfer simulations when comparing to flight tests and experimental data.
Design optimization of a diffuser augmented, dual-rotor hydrokinetic turbine utilizing 2D actuator disk theory
(2024-12) Walz, Andrew
An estimated 58,400 ± 109 TWh/year of untapped energy exists in rivers across the globe.Installing renewable technology at this capacity would mean significant reductions in greenhouse gas emissions and a slowdown of global warming. Hydrokinetic turbines have gained attraction in recent years to fulfill this need as they offer competitive power production with minimized environmental impact when compared to conventional fully dammed hydropower. Even more recently, researchers and manufactures have investigated how diffuser-augmentation of a hydrokinetic turbine can positively increase the power coefficient of these devices. Several design optimizations of these diffusers have been completed and some generalizations of their suggested shape have been made. Often neglected; however, are the adverse outcomes of diffuser augmentation, such as increased structural loading, increased manufacturing costs, and a reduction in compactness for array installation. The present study has developed a robust methodology for quantifying and implementing these adverse effects into the design optimization process. A total of 40, two-dimensional CFD numerical simulations are conducted, testing a wide range of diffuser shapes from flange style to airfoil style. Additionally, the actuator disk method is utilized across all design cycles to replicate the presence of a turbine rotor, creating a more realistic flow field; a step often neglected in existing two-dimensional design optimizations. The output parameter of maximizing interest is efficiency; a metric calculated by how effectively a diffuser shape can accelerate fluid through at the rotor plane. The output parameters of minimizing interest are diffuser pressure, diffuser material volume, and wake deficit; all metrics significantly influencing manufacturing costs and deployment constraints. A maximum of 29.4% and near minimum 24.9% efficiency values are recorded by the sharp flange and airfoil diffusers, respectively. However, where the flange diffuser excelled in maximizing efficiency, it under-performed in reducing the adverse effects. The contrary can be stated for the airfoil diffuser. An ‘optimal’ design is identified that shares characteristics to both diffuser shapes. Discussion includes insight on why this optimal design (DC 16) is most effective at this and how a future economic analysis could reveal a new optimal design dependent on the turbine deployer’s needs and manufacturing costs.
DSMC simulations of near-continuum hypersonic flows: code acceleration techniques and comparisons with state-of-the-art CFD solutions
(2021-10) Bhide, Paritosh
The modeling and simulation of hypersonic flow is challenging due to the fact that molecular time-scales, such as for vibrational energy relaxation and chemical reactions, become comparable to the characteristic flow time of the bulk gas moving past the vehicle. This results in gas flow under strong nonequilibrium conditions. In some cases, such as high-altitude flight, wake flow, flow over sharp leading edges, or flow regions involving sharp gradients, even the translational and rotational modes of the gas may be in nonequilibrium. Continuum Computational Fluid Dynamics (CFD) methods, solving the Navier-Stokes equations, have been extended to model chemical and vibrational nonequilibrium. However, such extensions are often not rigorous and continuum closure models are known to become inaccurate under translational and rotational nonequilibrium. Instead, the Direct Simulation Monte Carlo (DSMC) particle method can be used to simulate the Boltzmann equation, which is accurate for conditions ranging from fully continuum, to near-continuum, to free-molecular flow. The problem is that for the simulation of near-continuum flows, where only isolated regions exhibit strong nonequilibrium effects, DSMC becomes very expensive to simulate the entire flow domain. Although CFD solutions are much more efficient, they may be inaccurate. As a result, significant research has investigated the consistency between DSMC and CFD solutions and the development of hybrid CFD-DSMC methods to solve such near-continuum flows. This thesis addresses several key challenges associated with near-continuum hypersonic flow modeling, in the context of enabling accurate and efficient hybrid CFD-DSMC platforms. Specifically, this thesis investigates the phenomena of slip flow near sharp leading edges by comparing CFD solutions with and without slip-models against DSMC simulations that naturally predict velocity-slip and temperature-jump at the surface. While much CFD research and almost all previous hybrid CFD-DSMC research has employed no-slip conditions for CFD, it is concluded that the addition of slip-models leads to significantly improved consistency between CFD and DSMC, possibly enabling hybrid CFD-DSMC interfaces to extend to the wall in future hybrid simulations. This thesis also investigates the consistency of CFD and DSMC solutions for hypersonic flow involving the interaction between an attached boundary layer and a control surface, which induces a separated flow region. Although clear differences in boundary layer structure and separation size are identified (where discrepancy increases with increasing altitude), the predictions by CFD and DSMC for surface properties are remarkably similar, even up to flow conditions often considered to be rarefied. These results provide confidence that robust information transfer between CFD and DSMC regions of a hybrid solver can be obtained and provides a practical understanding of relevant flow conditions that motivate hybrid methodologies. Finally, to address the computational cost inherent in DSMC simulations of near-continuum flow regions, this thesis investigates the accuracy and efficiency of subcell acceleration techniques for the DSMC method. Techniques from existing literature are studied, implemented, and applied to near-continuum flows involving chemical reactions; challenging conditions not tested in previous research. Similar to previous research, it is determined that the level of subcell resolution dictates the accuracy of the solution. However, missing from previous research is the fact that the local number of simulated particles is equally important for solution accuracy. Based on a range of DSMC simulations, “best-practices” recommendations are made for the subcell method. The results of this thesis indicate that a 4x reduction in cell resolution (in each co-ordinate direction) and a 4x reduction in the overall number of simulated particles is able to achieve the same accuracy as baseline (fully-resolved) DSMC simulations. The contributions made by this thesis should enable the development of more efficient and robust hybrid CFD-DSMC simulation tools.
Effects of Manifold Deformation and Permeability on the Performance of the Flexible and Rigid Porous Stratification Manifolds for Solar Storage Tanks
(2016-12) Wang, Shuping
Promoting and maintaining a high degree of thermal stratification in solar storage tanks has well documented benefits for increasing the solar energy gain from solar heating systems [1–3]. This dissertation investigated the flexible fabric porous manifold and rigid porous-tube manifold proposed and tested in prior studies [4–10] for stratification enhancement. A mathematical model for the flexible fabric manifold is developed that accounts the interaction between the surrounding fluid in the tank and the flexible manifold wall. The relationship between the tube deformation and the pressure difference across the wall is described by the semi-empirical “tube law”. This manifold model provides a physical understanding of the working mechanism of the flexible fabric manifold and disapproves the widely held hypothesis that the deformation of the manifold is beneficial. Modeling results indicate that the dimensionless permeability can be optimized for improved performance. Following above findings, the effect of dimensionless permeability on the performance of the rigid porous-tube manifold are investigated in conditions representative of residential solar hot water systems. 2-D CFD simulation of the charging process reveals that optimizing the selection of the dimensionless permeability can improve the effectiveness of the manifold by eliminating suction (during intermediate charging) and releasing the fluid in the upper portion of the tank (during top charging). Furthermore, I show that the 1-D manifold model provides adequate prediction of the radial velocity distribution on the manifold wall compared with the results from the 2-D simulation. Therefore, the 1-D model can be used as an efficient design tool. With simulation results from the 1-D model over a wide range of the Richardson number and the dimensionless permeability, a design guideline for selection of the dimensionless permeability according to the Richardson number and the charging mode are developed. A prototype of the rigid porous-tube manifold is constructed and tested in comparison to conventional inlets. Temperature distribution in the tank is measured by thermocouple trees and the velocity field near the manifold is measured by a PIV system. For intermediate charging, the manifold achieves 0.5 dimensionless exergy efficiency after 90 minutes of charging, while the exergy efficiency for inlet diffuser and inlet pipe are 0.15 and 0, respectively. For top charging, the performance of the manifold is comparable to the best performing inlet diffuser. The measured velocity field is consistent to the model predication, indicating a 90% reduction of the suction rate during intermediate charging compared with conventional inlet pipe, and showing that the fluid is released approximately in the upper 25% of the tank during top charging.
Investigation of particle effects on a hypersonic Mars entry
(2023-06) Kroells, Michael
Large dust storms periodically form in the Martian atmosphere and pose a threat to future NASA missions. This threat arises from the current lack of understanding of how a Martian dust storm will affect the Thermal Protection System (TPS) of a planetary entry vehicle. While these storms occur infrequently, the long travel-times associated with Martian missions make avoiding these events nearly impossible and therefore the impact of a dust storm on the TPS must be estimated conservatively. However, excessive TPS margins increase the overall entry mass and diminish the allowable mass allocation for mission payload. The overall goal of this thesis is to identify, investigate, and improve underlying computational modeling assumptions relevant to dusty hypersonic Martian entries in order to reduce the conservative margins associated with TPS sizing. The first portion of this thesis covers a recently developed generalized drag coefficient for spherical particles, relevant for Martian dust particles interacting with a hypersonic flow. The proposed model incorporates simple physics-based scaling laws and is valid for a large range of Mach and Knudsen numbers. Additionally, the model retains an explicit dependence on gas type, which is useful for understanding the effect of the Martian atmosphere on particle drag. Next, several studies are performed that utilize Lagrangian particle-tracking to characterize atmospheric particles impacting the surface of high-speed flight vehicles. Two of the studies investigate the effect of Mars entry missions flying through a severe dust storm. The first involves determining the sensitivity of particle-induced surface erosion to underlying particle modeling for the Mars 2020 mission and the second targets an inflatable design that could potentially be used to support human missions to Mars. An additional study is performed in order to understand the impact characteristics of stratospheric particles in Earth's atmosphere on a representative hypersonic flight vehicle. Lastly, a comparison of several numerical strategies for colliding hard-sphere particles is performed. While particle-particle collisions are not likely to play an important role for atmospheric particles because of their low concentrations, particle collisions can play a more important role in characterizing ground experiments that typically have higher particle mass loadings. Specifically, a collision procedure based on the direct-simulation Monte Carlo (DSMC) method is compared to event-driven and time-driven methods for two numerical setups, where the DSMC-inspired collision method is found to be preferable to the other approaches considered because of its improved accuracy and efficiency.
A method for the development of an effective flow diverting device for the treatment of cerebral aneurysms
(2013-09) Chow, Ricky
Intracranial aneurysms are malformations that occur in the complex network of blood vessels supplying oxygen and nutrients to the brain. Weakening of the blood vessel wall leads to a bulge that ruptures in more than 30 000 Americans every year. Prognosis is very poor. Patients often die or suffer a greatly reduced quality of life. Two predominant methods for treating aneurysms are (1) surgical clipping, where part of the skull is temporarily removed and a metallic clip is placed to circumvent the aneurysm neck, and (2) coiling, where metallic coils are snaked through the blood vessels and packed into the aneurysm.For large aneurysms or those with poorly defined necks, a new class of medical device has recently emerged as a more effective treatment than coiling. A flow diverter is placed inside the parent vessel, spanning the aneurysm neck. The diverter's braided structure keeps most of the blood from entering the aneurysm. The risk of rupture is eliminated when stagnant pools of blood thrombose inside the aneurysm, cutting the aneurysm off from the rest of the circulatory system. However, complications related to the presence of flow diverters are observed clinically. Aneurysms with incomplete clot formation after placement of the flow diverter are still at risk of rupture. The high metallic content of the device presents a risk of in-stent thrombosis and require a lifetime of anti-coagulants for its management. Subarachnoid hemorrhage after placement of the flow diverter is observed, but the underlying mechanism is not well understood. Therefore, a greater understanding of the fluid mechanics underlying flow diversion is needed to facilitate the design of the next generation of flow diverters. Research was pursued in three parallel synergistic paths. (1) Benchtop experiments using a technique called particle imaging velocimetry (PIV) were used to characterize the flow diversion accorded by the Pipeline Embolization Device (PED, designed by Covidien) in a variety of geometries. (2) The computational fluid dynamic (CFD) simulation methods were verified with PIV results, and then applied towards a wider range of vessel geometries to predict how the PED will perform at various locations of the human neurovasculature. (3) Animal studies were pursued to develop surgical techniques for device evaluation in the future. The implementation of PIV was found to be a labor and computationally intensive process. Previous researchers who have used PIV to experimentally investigate the flow diverting effect of the device occasionally interrogated the fluid domain at several planes, but typically only at the center plane bisecting the aneurysm. This limited information was found to be insufficient for verification of CFD simulations or to calculate bulk properties such as flow rate of fluid entering the aneurysm. Evaluation of intraaneurysmal flow was also found to be problematic after placement of the flow diverter. The significantly reduced flow highlighted the difference in densities between the seeded reflective particles and the flowing fluid. Particles also accumulated on the glass model wall in regions of low flow. These complications introduced challenges to the PIV measurement technique.Detailed sets of PIV results were collected in three flow domains by interrogating the flow at parallel planes 400 microns apart. The flow rates of fluid entering the aneurysm before (QUT) and after (QT) placement of the flow diverters were calculated. CFD simulations were conducted with the openings, or pores, of the PED modeled as an array of diamond shaped pores connecting the aneurysm to the parent artery. Since the deployed shape of the PED was variable and depended on the deployment technique, simulations with different diamond pore dimensions were conducted. The QT and QUT values predicted from CFD were in reasonable agreement with the PIV results.CFD simulations were then conducted in an array of idealized blood vessel geometries that typified a portion of the vessel curvatures found in the human neurovasculature. It was discovered that the performance of the PED varied depending on the curvature of the parent vessel, the location of the aneurysm along the curve, and the geometry of the aneurysm neck. The claim of "85% reduction in circulation" made by Covidien (who designed the PED) is a somewhat ambiguous statement. An ~85% reduction in vorticity was observed on the center planes of the aneurysms evaluated in this research effort, but the reduction in flow rate entering the aneurysm was on average only around 65%, and dipped as low as 50% in the most tortuous bends. However, the shapes of the deployed PED, the vessel geometries, and inlet conditions examined in this thesis may have been different than those used to substantiate Covidien's marketing claim. The term "circulation" was also not defined in Covidien's literature. Further research is needed to identify the source of this discrepancy.The present research also provides insight into the fluid mechanics of blood entering aneurysms created in a rabbit model. Residence time was defined as the volume of the aneurysm divided by the flow rate of blood entering the aneurysm. Blood velocities acquired using an ultrasound probe were used as input to CFD simulations. The varying volumes of the aneurysms and the varying angles of the aneurysms relative to their parent arteries led to residence times that varied from rabbit to rabbit. Knowledge of the initial flow conditions is important for an apples to apples comparison of new flow diverter designs. More animal studies combined with clinical data of the PED are needed to determine the minimum threshold in flow reduction, the minimum residence time, or some other metric that will predict healing of the aneurysm. In summary, a comprehensive platform of evaluation techniques was developed and implemented for use in optimizing the design of the next generation of flow diverters. The reduction in flow entering the aneurysm after placement of the PED was found to be less than the claimed "reduction in circulation" and presents an opportunity for a flow diverter that restricts flow more severely. Moving from a metallic braid to a polymeric stent graft platform would allow for easier manipulation of flow diversion characteristics while taking into account other design requirements such as device stiffness, force required to advance it through the catheter, radiopacity, thrombogenicity, stent migration, and others. A better understanding of the underlying mechanism by which flow diverters heal aneurysms will lead to wider adoption and on-label use (officially approved by the European Commission and the Food and Drug Administration) of this class of device as a first-line treatment for all aneurysms.
Numerical and Theoretical Studies of Air Entrainment and Bubble Acoustics under Breaking Waves
(2021-12) Gao, Qiang
Bubbles and breaking waves play a critical role in many physical processes. However, bubble formation mechanism, trajectories, and their acoustic signatures are still poorly understood due to the complex process of breaking waves. To study the bubble transport dynamics and their formation mechanism, a technique for Lagrangian tracking of bubbles and detecting their time-evolution behaviors is developed. Five possible behaviors are considered: formation, extinction, continuity, binary fragmentation, and binary coalescence. The technique is based on establishing a network of mappings between bubbles identified at adjacent time instants. The accuracies for continuity, binary fragmentation, and binary coalescence are estimated to be 99.5%, 90%, and 95%, respectively. The algorithm is proved to be accurate and robust by extensive validations using the breaking wave cases. Bubble entrainment mechanism and bubble trajectory are investigated. Air filaments and cavities in plunging breaking waves, generically cylinders, produce bubbles through an interface instability. A generalized dispersion relation is obtained that spans the Rayleigh–Taylor and Plateau–Rayleigh instabilities as cylinder radius varies. The analysis provides insight into the role of surface tension in the formation of bubbles from filaments and cavities. Small filaments break up into bubbles through a Plateau–Rayleigh instability driven through the action of surface tension. Large air cavities produce bubbles through a Rayleigh–Taylor instability driven by gravity and moderated by surface tension, which has a stabilizing effect. Surface tension, interface curvature, and gravity are all important for cases between these two extremes. Bubble trajectories and their interaction with breaking wave flow fields are also studied here. A simulation framework for bubbly flow and the sound radiated by breaking waves is presented. It consists of a two-phase flow solver, an algorithm to track bubbles and bubble creation rates, and a module to compute the sound generated by newly-formed bubbles. The sounds from breaking, third-order Stokes waves of 0.25m wavelength and two slopes are calculated. The results show encouraging agreement with existing laboratory observations and identify the importance of air cylinder breakup in bubble creation.
Numerical Modeling And Optimization Of Thermofluid Systems: Heat Pumps, Turbocompressors, Porous Media
(2020-03) Goldenberg, Vlad
In this dissertation, three types of thermofluid systems: an air Brayton cycle heat pump, a centrifugal compressor stage, and a porous media heat pipe, are investigated. In each of the investigations, numerical modeling is used as the basis that underpins the analyses. Furthermore, the goal of each investigation is to develop a framework for the design and optimization of practical engineering systems. The parameterization of each system is explored and defined. A thermodynamic model of a recuperated air cycle heat pump is developed and used to parametrically study the effects of component performance, operating environment, and design parameters. A numerical optimization is conducted to maximize the heating COP of the air cycle heat pump while maintaining robust performance across a wide operating envelope. Comparison is made to a conventional vapor compression cycle heat pump. It is found that a judicious choice of pressure ratio and maximization of component performance enables a recuperated air cycle heat pump to be comparable in COP to a vapor cycle heat pump for high temperature ratio duty. The recommended pressure ratio is determined to be 1.4. Such a heat pump requires high performance compressor, expander, and heat exchangers. A novel method of the flow path synthesis of a centrifugal compressor stage is revealed. A preliminary design procedure that enables fast and efficient candidate designs is reported. Computational fluid dynamics in conjunction with optimization algorithms, surrogate modeling, and machine learning is used to analyze the fundamental fluid mechanics and to automatically optimize the designs. A single stage performance improvement of over 4% points of isentropic efficiency gain is demonstrated using numerical methods. The microstructure of a flat porous media heat pipe consisting of layers of wire mesh is characterized using numerical techniques. The analysis encompasses the characterizations of the flow-induced pressure drop and interfacial heat transfer for liquid and vapor water phases in a 16-gauge and 200-gauge wire mesh porous domain.
Wind-driven external aerodynamics around buildings and buoyancy-driven fluid motion and heat transfer in internal flow passages
(2017-08) Bettenhausen, Daniel
Two areas of focus are considered with respect to the resilience of buildings to withstand environmental forces and to the enhancement of building energy efficiency. First, wind-driven pressure and velocity fields surrounding a model building are calculated by means of computational fluid dynamics (CFD) in two- and three- dimensions using the Shear Stress Transport (SST) turbulence model. The sensitivity of pressure coefficient distributions over each building surface to the placement of the solution domain and to the applied boundary conditions is determined. Pressure coefficient magnitudes were found to be particularly sensitive to the distance separating the upstream boundary of the solution domain, where the atmospheric boundary layer velocity profile is specified, from the location of the building. The magnitude of pressure coefficients at each building surface tended to decrease with increasing upstream distance of the applied velocity profile. At the building roof, the three-dimensional representation of the building resolved off-roof-centerline periodic transient pressure coefficient variations whereas the two-dimensional model predicted steady-state pressure coefficients over the entire rooftop. Building energy efficiency was studied via CFD to determine buoyancy-based heat transfer and velocity distributions in an asymmetrically heated vertical-wall channel representing a double-walled building with and without obstructions placed within the vertical channel. Specifically, catwalks consisting of an array of rectangular slats were deployed at two floor levels of a three-story building. The numerical model employed is validated by comparison with experimental data from a literature source. Channel Nusselt numbers and Reynolds numbers were found to increase with the openness of the catwalk elements to flow passage and also with the channel Rayleigh number. A porous-medium model employing the Darcy-Forchheimer equation is considered as a means to represent the pressure drop of each catwalk grating. This approach yielded results of only moderate accuracy for the velocity field because the porous-medium model does not faithfully reproduce turbulence. On the other hand, an approximation-free model was successfully employed to yield highly accurate fluid flow and heat transfer results for the channel flow.

University Digital Conservancy

Browse by Subject

Browsing by Subject "CFD"