Browsing by Subject "Computational Fluid Dynamics"

Now showing 1 - 12 of 12
  • Thumbnail Image
    Item
    Air-Carbon Ablation for Hypersonic Flow Environments
    (2022-05) Prata, Krishna Sandeep
    Recent molecular beam experiments of high-velocity O, N, and O2 impacting carbon material at high temperatures produced detailed surface chemistry data relevant for carbon ablation processes. New data on O and N reactions with carbon has been published using a continuous molecular beam with lower velocity (2000 m/s) and approximately 500 times higher beam flux than previous pulsed-beam experiments. This data is interpreted to construct a new air-carbon ablation (ACA) model for use in modeling carbon heat shield ablation. The new model comprises 20 reaction mechanisms describing reactions between impinging O, N, and O2 species with carbon and producing scattered products including desorbed O and N, O2, and N2 formed by surface-catalyzed recombination, as well as CO, CO2, and CN. The new model includes surface-coverage-dependent reactions and exhibits non-Arrhenius reaction probability in agreement with experimental observations. All reaction mechanisms and rate coefficients are described in detail and each is supported by experimental evidence or theory. The model predicts pressure effects and is tested for a wide range of temperatures and pressures relevant to hypersonic flight. Model results are shown to agree well with available data and are shown to have significant differences compared to other models from the literature. A preliminary step towards validating the ACA model required simulating the plasma flows in the plasma chamber of von Karman Institute's (VKI's) Inductively Coupled Plasma (ICP) facility called Plasmatron using US3D, a 3D unstructured Navier-Stokes equations solver developed at the University of Minnesota. First, a parameter study of transport properties and the wall-catalycity of a catalytic probe used to characterize the plasma flow was conducted. It was found that the Gupta-Yos mixing rule with collision cross-section data performed better than Wilke's mixing rule with Blottner curve fits and Eucken relation to compute mixture viscosity and thermal conductivity. Also, the wall-catalycity had a strong effect on the boundary layer edge properties along the stagnation line for lower pressure flows in the Plasmatron. It was also found that the Self Consistent Effective Binary Diffusion (SCEBD) model predicted higher stagnation line enthalpy at the boundary layer edge for a flow over a non-catalytic wall when compared to the Fickian diffusion model that attributes a single diffusion coefficient to all the species in the mixture. Then, a series of iterative US3D simulations were performed to characterize the plasma flows over a fully-catalytic wall for seven air-plasma experiments in the Plasmatron. The simulations matched the experimentally measured cold wall heat flux and agreed relatively well with the boundary layer edge properties predicted by VKI's own analysis, giving confidence that the plasma freestream was well characterized in the seven experiments. Then, preliminary simulations of carbon ablation using the seven plasma flows were performed. Two ablation models called the ZA model and the MURI model gave comparable carbon mass loss rates with the experiments. However, the ZA model predicted lower surface heat flux than the MURI model due to the presence of spurious gas-surface reactions. Further experiments of carbon ablation measuring the surface heat flux are suggested in addition to mass-loss measurements. Finally, an analytical framework was developed that characterizes a flight mission or an experimental condition as reaction-limited or diffusion-limited with respect to carbon ablation. The framework uses the flight conditions such as the velocity, altitude, nose radius of the vehicle, and the surface temperature of the ablating heat shield to calculate time scales for diffusion and gas-surface chemical reactions. A new Damkohler number for ablation, defined as the ratio of diffusion time scale to the time scale for gas-surface chemical reactions was proposed. The framework was applied to several flight conditions in the existing literature. It was found that ablation of larger heat-shields like the Stardust re-entry capsule and Orion space-crew vehicle falls under a diffusion-limited regime, while the ablation of smaller objects like the nose tip of the Re-entry F vehicle falls under a reaction-limited regime. Future CFD simulations of ablation on various heat shields using the existing ablation models in the literature are recommended to establish a reference Damkohler number for the classification of the ablation regimes.
  • Thumbnail Image
    Item
    Analysis of Stationary Crossflow Instability on HIFiRE-5 Using Direct Numerical Simulation
    (2016-02) Dinzl, Derek
    Direct numerical simulation is performed on a 38.1% scale HIFiRE-5 forebody to study stationary crossflow instability. Computations use the US3D Navier-Stokes solver to simulate Mach 6 flow at Reynolds numbers of 8.1e6 /m and 11.8e6$ /m, which are conditions used by quiet tunnel experiments at Purdue University. Distributed roughness with point-to-point height variation on the computational grid and maximum heights of 0.5-4.0 microns is used with the intent to emulate smooth-body transition and excite the naturally-occurring most unstable disturbance wavenumber. Cases at the low Reynolds number condition use three grid sizes, and hence three different roughness patterns of varying wavelength, and demonstrate that the final flow solution is extremely dependent on the particular roughness pattern. The same roughness pattern is interpolated onto each grid which yields similar solutions, indicating grid convergence. At the high Reynolds number condition, a steady physical mechanism is introduced which explains sharp increases seen in the wall heat flux for both computations and experiment. Namely, the sharp increase is caused by large streamwise velocity disturbances impinging on the wall. Evolution of disturbance spanwise wavelength is computed, and it is found that this wavelength is more sensitive to Reynolds number than roughness, indicating that the disturbance wavelength is primarily flow--selected for these cases. The calculation of disturbance growth rates shows the region over which crossflow disturbances behave linearly and where nonlinear effects become important. The effect of roughness height and nose sharpness are considered, and both were found to have a large effect on the resulting heating pattern. Crossflow vortex coalescence is observed and a possible cause is discussed.
  • Thumbnail Image
    Item
    Consistent Chemical Kinetics and Continuum Models for High Temperature Air
    (2020-02) Singh, Narendra
    High-fidelity nonequilibrium reaction models for hypersonic air flow are developed. Hypersonic flows create shock waves, which compress and heat the surrounding gas to high-temperatures. Strong shock waves cause dissociation of nitrogen and oxygen molecules. Predicting the extent of dissociation and recombination of atomic species is important since the state of the gas near the vehicle surface determines heating rates and gas-surface chemistry that damages the heat shield. Since experimental data is difficult to obtain under such extreme conditions, numerical simulation plays an important role. Predictive numerical simulations require accurate reaction chemistry models. Computational models developed thus far range from simple empirical models fit to limited experimental data to models with millions of input parameters that track individual quantized energy state transitions. The level of model fidelity required for accurate engineering analysis remains an open question of active research. Models coupling internal energy and dissociation chemistry tend to be developed at either the kinetic scale or the continuum scale. In this dissertation, we develop new nonequilibrium models for shock heated flows that are analytically consistent between kinetic and continuum formulations, and are based on recent ab-initio data.
  • Thumbnail Image
    Item
    Design of Medical Devices Involving Multi-disciplinary Processes and Based on Fundamental Physical Principles
    (2016-04) Krautbauer, Kevin
    This dissertation focuses on the optimal design of medical devices through the use of numerical simulation and the utilization of first principles of the participating phenomena. Through three broadly ranging case studies, the dissertation explores a wide variety of physical phenomena found within medical devices and in other applications. Pressure drop and sound generation are the primary focii of the leading case study which constitutes the first-ever analysis of the fluid mechanics of a therapeutic device for the treatment of cystic fibrosis. The treatment utilizes a time varying pressure that acts on the abdomen of the patient in order to break up masses of mucus. The second study is the first known effort to design peristaltic pumps using the principles of fluid-structure interaction. The time-dependent mechanics of peristaltic pumping were utilized to determine the deformations and pressures in the flexible-walled plastic tubing. The change of volume of the tubing serves to propel a liquid contained within the tube. Finally, the third study investigates the fluid mechanics and heat transfer mechanisms found in an enhanced-surface fluid warming device. The key analysis and design tools used throughout the aforementioned case studies of this dissertation are physical model formulation adapted to computational fluid dynamics (CFD), the theory of turbulence-based sound generation, Ogden’s hyperelastic model of polymeric materials, and the theory of heat transfer. The fluid flow phenomena dealt with in this work include three-dimensional, unsteady, laminar and turbulent flows. Heat transfer concepts utilized include conduction within both fluids and solids, advection within interacting parallel flow regions, and the theory of heat transfer enhancement. Each chapter contains multiple results pertaining to the device in question. These results serve to expand the reader’s knowledge of the underlying physical processes which control the function and effectiveness of the medical device.
  • Thumbnail Image
    Item
    Development and Validation of a Turbulence Wall Model for Compressible Flows with Heat Transfer
    (2016-08) Komives, Jeffrey
    The computational cost to model high Reynolds number flows of engineering interest scales poorly with problem size and is excessively expensive. This fact motivates the development of turbulence wall models to lessen the computational burden. These models aim to provide accurate wall flux quantification on computational meshes that would otherwise be unable to accurately estimate these quantities. The benefit of using such an approximation is that the height of the wall-adjacent computational elements can be increased by one to two orders of magnitude, allowing for comparable increases in stable explicit timestep. This increase in timestep is critically necessary for the large eddy simulation of high Reynolds number turbulent flows. To date, most research in the application of wall models has focused on incompressible flows or flows with very weak compressibility. Very few studies examine the applicability of wall models to flows with significant compressibility and heat transfer. The present work details the derivation of a wall model appropriate for compressible flows with heat transfer. The model framework allows for the inclusion of non-equilibrium terms in the determination of wall shear and heat transfer. The model is applied to a variety of supersonic and hypersonic flows, and is studied in both Reynolds-averaged simulations and large eddy simulations. The impact of several modeling approaches and model terms is examined. The wall-modeled calculations show excellent agreement with wall-resolved calculations and experimental data. For time accurate calculations, the use of the wall model allows for explicit timesteps more than 20 times larger than that of the wall-resolved calculation, significantly reducing both the cost of the calculation and the time required converge the solution.
  • Thumbnail Image
    Item
    Development of Modal Analysis for the Study of Global Modes in High Speed Boundary Layer Flows
    (2017-05) Brock, Joseph
    Boundary layer transition for compressible flows remains a challenging and unsolved problem. In the context of high-speed compressible flow, transitional and turbulent boundary-layers produce significantly higher surface heating caused by an increase in skin-friction. The higher heating associated with transitional and turbulent boundary layers drives thermal protection systems (TPS) and mission trajectory bounds. Proper understanding of the mechanisms that drive transition is crucial to the successful design and operation of the next generation spacecraft. Currently, prediction of boundary-layer transition is based on experimental efforts and computational stability analysis. Computational analysis, anchored by experimen- tal correlations, offers an avenue to assess/predict stability at a reduced cost. Classi- cal methods of Linearized Stability Theory (LST) and Parabolized Stability Equations (PSE) have proven to be very useful for simple geometries/base flows. Under certain conditions the assumptions that are inherent to classical methods become invalid and the use of LST/PSE is inaccurate. In these situations, a global approach must be considered. A TriGlobal stability analysis code, Global Mode Analysis in US3D (GMAUS3D), has been developed and implemented into the unstructured solver US3D. A discussion of the methodology and implementation will be presented. Two flow configurations are presented in an effort to validate/verify the approach. First, stability analysis for a subsonic cylinder wake is performed and results compared to literature. Second, a supersonic blunt cone is considered to directly compare LST/PSE analysis and results generated by GMAUS3D.
  • Thumbnail Image
    Item
    Fluid-Structure Interaction Simulation of Complex Floating Structures and Waves
    (2015-11) Calderer Elias, Antoni
    A novel computational framework for simulating the coupled interaction of complex floating structures with large-scale ocean waves and atmospheric turbulent winds has been developed. This framework is based on a domain decomposition approach coupling a large-scale far-field domain, where realistic wind and wave conditions representative from offshore environments are developed, with a near-field domain, where wind-wave-body interactions can be investigated. The method applied in the near-field domain is based on a partitioned fluid-structure interaction (FSI) approach combining a sharp interface curvilinear immersed boundary (CURVIB) method with a two-phase flow level set formulation and is capable of solving free surface flows interacting non-linearly with complex real life floating structures. An aspect that was found critical in FSI applications when coupling the structural domain with the two-fluid domain is the approach used to calculate the force that the fluid exerts to the body. A new force calculation approach, based on projecting the pressure on the surface of the body using the momentum equation along the local normal to the body direction, was proposed. The new approach was shown, through extensive numerical tests, to greatly improve the ability of the method to correctly predict the dynamics of the floating structure motion. For the far-field domain, a large-scale wave and wind model based on the two-fluid approach of Yang and Shen (JCP 2011), which integrates a viscous Navier-Stokes solver with undulatory boundaries for the motion of the air and an efficient potential-flow based wave solver, was employed. For coupling the far-field and near-field domains, a wave generation method for incorporating complex wave fields into Navier-Stokes solvers has been proposed. The wave generation method was validated for a variety of wave cases including a broadband spectrum. The computational framework has been further validated for wave-body interactions by replicating an experiment of floating wind turbine model subject to different sinusoidal wave forces. The simulation results, which agree well with the experimental data, have been compared with other numerical results computed with available numerical codes based on lower order assumptions. Despite the higher computational cost of our method, it yields to results that are in overall better accuracy and it can capture many additional flow features neglected by lower order models. Finally, the full capabilities of the framework have been demonstrated by carrying out large eddy simulation (LES) of a floating wind turbine interacting with realistic ocean wind and wave conditions.
  • Thumbnail Image
    Item
    Modeling and Analysis of Chemical Kinetics for Hypersonic Flows in Air
    (2018-11) Chaudhry, Ross
    Gas-phase chemical kinetics are relevant for hypersonic flows, but they are currently modeled in CFD using empirical assumptions and decades-old experimental data. Recent advances in quantum chemistry have enabled the construction of accurate potential energy surfaces (PESs) for diatom-diatom interactions in air. Using these PESs, a database of simulated interactions is generated and analyzed; N2 + N2, N2 + N, N2 + O2, O2 + O2, and O2 + O reactions are considered. The conditions studied range from 4000 K to 30,000 K and include thermal equilibrium and nonequilibrium test sets. The nitrogen dissociation rate is found to be similar for collision partners N2, N, and O2. The oxygen dissociation rate, in contrast, is moderately dependent on partner species; O2 is approximately 2 to 3 times more effective than partner N2. Oxygen dissociation with partner N2 is therefore found to be substantially overpredicted by current CFD models, which is consistent with the limited experimental data available for this reaction. The presence of N is known from experiments to promote nitrogen dissociation; this augmentation is found to be due to increased vibrational relaxation, rather than an increased dissociation rate as described by current CFD models. Similar observations are made for oxygen dissociation with partner O, due to a combination of vibrational and electronic energy relaxation. Using only the shock tube data that informed popular CFD models, it was impossible to isolate the effect of increased relaxation from increased dissociation. The change in vibrational energy per dissociation, a necessary input to CFD, is found to be very sensitive to the degree of thermal nonequilibrium. This dependence is not well predicted by any existing chemical kinetics models; correctly describing this term fundamentally changes the thermal evolution of a gas in CFD. The mechanics of dissociation are similar for all reactions studied, so a series of aggregate analyses on all dissociation reactions is performed. Vibration is found to have a more pronounced effect on dissociation than rotation, due to rotation increasing the centrifugal barrier. The classic Marrone-Treanor preferential dissociation model is found to accurately describe all data in the nonequilibrium test sets, but it neglects the effect of rotational energy on dissociation. A modified model is proposed that describes rates to within 22% and vibrational energy changes to within 4% of the dissociation energy, for all dissociation reactions and conditions. For this work, we have considered Boltzmann or approximately Boltzmann distributions, but the population of high-energy molecules is known to be depleted in a dissociating gas ensemble. Various kinetics models based on Boltzmann distributions are implemented in US3D, a production CFD solver designed for hypersonic flows. As expected, the dissociation rate is overpredicted compared to the benchmark data. Work remains, therefore, to account for the non-Boltzmann distributions that exist in reality. These data and insights about dissociation can form the basis for next-generation chemical kinetics models for CFD.
  • Thumbnail Image
    Item
    Modeling and simulation of high-speed wake flows.
    (2009-08) Barnhardt, Michael Daniel
    High-speed, unsteady flows represent a unique challenge in computational hypersonics research. They are found in nearly all applications of interest, including the wakes of reentry vehicles, RCS jet interactions, and scramjet combustors. In each of these examples, accurate modeling of the flow dynamics plays a critical role in design performance. Nevertheless, literature surveys reveal that very little modern research effort has been made toward understanding these problems. The objective of this work is to synthesize current computational methods for high-speed flows with ideas commonly used to model low-speed, turbulent flows in order to create a framework by which we may reliably predict unsteady, hypersonic flows. We wish to validate the new methodology for the case of a turbulent wake flow at reentry conditions. Currently, heat shield designs incur significant mass penalties due to the large margins applied to vehicle afterbodies in lieu of a thorough understanding of the wake aerothermodynamics. Comprehensive validation studies are required to accurately quantify these modeling uncertainties. To this end, we select three candidate experiments against which we evaluate the accuracy of our methodology. The first set of experiments concern the Mars Science Laboratory (MSL) parachute system and serve to demonstrate that our implementation produces results consistent with prior studies at supersonic conditions. Second, we use the Reentry-F flight test to expand the application envelope to realistic flight conditions. Finally, in the last set of experiments, we examine a spherical capsule wind tunnel configuration in order to perform a more detailed analysis of a realistic flight geometry. In each case, we find that current 1st order in time, 2nd order in space upwind numerical methods are sufficiently accurate to predict statistical measurements: mean, RMS, standard deviation, and so forth. Further potential gains in numerical accuracy are demonstrated using a new class of flux evaluation schemes in combination with 2nd order dual-time stepping. For cases with transitional or turbulent Reynolds numbers, we show that the detached eddy simulation (DES) method holds clear advantage over heritage RANS methods. From this, we conclude that the current methodology is sufficient to predict heating of external, reentry-type applications within experimental uncertainty.
  • Thumbnail Image
    Item
    Numerical Simulation Of The Atmospheric Boundary Layer Over Complex Topography: A Modern Approach To A Classical Problem
    (2020-05) Andersen, Noah
    Numerical methods were developed and validated to simulate the atmospheric boundary layer (ABL) using large eddy simulation (LES). This framework captures the topography of the Earth’s surface rather than modeling it. To robustly simulate the ABL, four unique capabilities (temperature transport, topographic data, immersed boundary method with wall modeling, and turbulent inflow generation) were added to a traditional finite difference computational fluid dynamics code. The accuracy of each capability was analyzed individually using validation tests. Then, a full scale simulation of the ABL over a tidal inlet was conducted. It was found that the resolved topography of the Earth’s surface had a significant effect on the flow field. Furthermore, it was found that the results from LES are more accurate than mesoscale simulations. Lastly, it was found that the errors in the present simulation are a result of the roughness model used over the sea surface.
  • Thumbnail Image
    Item
    Parallel Adaptive Mesh Refinement for High-Order Finite-Volume Schemes in Computational Fluid Dynamics
    (2015-08) Schwing, Alan
    For computational fluid dynamics, the governing equations are solved on a discretized domain of nodes, faces, and cells. The quality of the grid or mesh can be a driving source for error in the results. While refinement studies can help guide the creation of a mesh, grid quality is largely determined by user expertise and understanding of the flow physics. Adaptive mesh refinement is a technique for enriching the mesh during a simulation based on metrics for error, impact on important parameters, or location of important flow features. This can offload from the user some of the difficult and ambiguous decisions necessary when discretizing the domain. This work explores the implementation of adaptive mesh refinement in an implicit, unstructured, finite-volume solver. Consideration is made for applying modern computational techniques in the presence of hanging nodes and refined cells. The approach is developed to be independent of the flow solver in order to provide a path for augmenting existing codes. It is designed to be applicable for unsteady simulations and refinement and coarsening of the grid does not impact the conservatism of the underlying numerics. The effect on high-order numerical fluxes of fourth- and sixth-order are explored. Provided the criteria for refinement is appropriately selected, solutions obtained using adapted meshes have no additional error when compared to results obtained on traditional, unadapted meshes. In order to leverage large-scale computational resources common today, the methods are parallelized using MPI. Parallel performance is considered for several test problems in order to assess scalability of both adapted and unadapted grids. Dynamic repartitioning of the mesh during refinement is crucial for load balancing an evolving grid. Development of the methods outlined here depend on a dual-memory approach that is described in detail. Validation of the solver developed here against a number of motivating problems shows favorable comparisons across a range of regimes. Unsteady and steady applications are considered in both subsonic and supersonic flows. Inviscid and viscous simulations achieve similar results at a much reduced cost when employing dynamic mesh adaptation. Several techniques for guiding adaptation are compared. Detailed analysis of statistics from the instrumented solver enable understanding of the costs associated with adaptation. Adaptive mesh refinement shows promise for the test cases presented here. It can be considerably faster than using conventional grids and provides accurate results. The procedures for adapting the grid are light-weight enough to not require significant computational time and yield significant reductions in grid size.
  • Thumbnail Image
    Item
    A strategy for high performance in computational fluid dynamics
    (2013-08) Jayaraj, Jagan
    Computational Fluid Dynamics is an important area in scientific computing. The weak scaling of codes is well understood with about two decades of experience using MPI. The recent proliferation of multi- and many-core processors have made the modern nodes compute rich, and the per-node performance has become very crucial for the overall machine performance. However, despite the use of thread programming, obtaining good performance at each core is extremely challenging. The challenges are primarily due to memory bandwidth limitations and difficulties in using the short SIMD engines effectively. This thesis is about the techniques, strategies, and a tool, to improve the in-core performance. Fundamental to the strategy is a hierarchical data layout made of small cubical structures of the problem state called the briquettes. The difficulties in computing the spatial derivatives (also called near neighbor computations in the literature) in a hierarchical data layout are well known, and data blocking is extremely unusual in finite difference codes. This work details how to simplify programming for the new data layout, the inefficiencies of the programming strategy, and how to overcome the inefficiencies.The transformation to eliminate the overheads is called pipeline-for-reuse. It is followed by a storage optimization called maximal array contraction. Both pipeline-for-reuse and maximal array contraction are highly tedious and error-prone. Therefore, we built a source-to-source translator called CFD Builder to automate the transformations using directives. The directive based approach we adopted to enable the transformations eliminates the need for complex analysis, and this work provides the linear time algorithms to perform the transformations under the stated assumptions. The benefits of the briquettes and CFD Builder are demonstrated individually with three different applications on two different architectures and two different compilers. We see up to 6.92x performance improvement with applying both the techniques. This strategy with briquettes and CFD Builder was evaluated against commonly known transformations for data locality and vectorization. Briquettes and pipeline-for-reuse transformations to eliminate the overheads outperforms even the best combination of canonical transformations, for data locality and vectorization, applied manually by up to 2.15x