Matching Items (14)
Filtering by
- All Subjects: Turbulence
- All Subjects: multiphase
- Creators: Herrmann, Marcus
Description
The evolution of single hairpin vortices and multiple interacting hairpin vortices are studied in direct numerical simulations of channel flow at Re-tau=395. The purpose of this study is to observe the effects of increased Reynolds number and varying initial conditions on the growth of hairpins and the conditions under which single hairpins autogenerate hairpin packets. The hairpin vortices are believed to provide a unified picture of wall turbulence and play an important role in the production of Reynolds shear stress which is directly related to turbulent drag. The structures of the initial three-dimensional vortices are extracted from the two-point spatial correlation of the fully turbulent direct numerical simulation of the velocity field by linear stochastic estimation and embedded in a mean flow having the profile of the fully turbulent flow. The Reynolds number of the present simulation is more than twice that of the Re-tau=180 flow from earlier literature and the conditional events used to define the stochastically estimated single vortex initial conditions include a number of new types of events such as quasi-streamwise vorticity and Q4 events. The effects of parameters like strength, asymmetry and position are evaluated and compared with existing results in the literature. This study then attempts to answer questions concerning how vortex mergers produce larger scale structures, a process that may contribute to the growth of length scale with increasing distance from the wall in turbulent wall flows. Multiple vortex interactions are studied in detail.
ContributorsParthasarathy, Praveen Kumar (Author) / Adrian, Ronald (Thesis advisor) / Huang, Huei-Ping (Committee member) / Herrmann, Marcus (Committee member) / Arizona State University (Publisher)
Created2011
Description
The atomization of a liquid jet by a high speed cross-flowing gas has many applications such as gas turbines and augmentors. The mechanisms by which the liquid jet initially breaks up, however, are not well understood. Experimental studies suggest the dependence of spray properties on operating conditions and nozzle geom- etry. Detailed numerical simulations can offer better understanding of the underlying physical mechanisms that lead to the breakup of the injected liquid jet. In this work, detailed numerical simulation results of turbulent liquid jets injected into turbulent gaseous cross flows for different density ratios is presented. A finite volume, balanced force fractional step flow solver to solve the Navier-Stokes equations is employed and coupled to a Refined Level Set Grid method to follow the phase interface. To enable the simulation of atomization of high density ratio fluids, we ensure discrete consistency between the solution of the conservative momentum equation and the level set based continuity equation by employing the Consistent Rescaled Momentum Transport (CRMT) method. The impact of different inflow jet boundary conditions on different jet properties including jet penetration is analyzed and results are compared to those obtained experimentally by Brown & McDonell(2006). In addition, instability analysis is performed to find the most dominant insta- bility mechanism that causes the liquid jet to breakup. Linear instability analysis is achieved using linear theories for Rayleigh-Taylor and Kelvin- Helmholtz instabilities and non-linear analysis is performed using our flow solver with different inflow jet boundary conditions.
ContributorsGhods, Sina (Author) / Herrmann, Marcus (Thesis advisor) / Squires, Kyle (Committee member) / Chen, Kangping (Committee member) / Huang, Huei-Ping (Committee member) / Tang, Wenbo (Committee member) / Arizona State University (Publisher)
Created2013
Description
Structural features of canonical wall-bounded turbulent flows are described using several techniques, including proper orthogonal decomposition (POD). The canonical wall-bounded turbulent flows of channels, pipes, and flat-plate boundary layers include physics important to a wide variety of practical fluid flows with a minimum of geometric complications. Yet, significant questions remain for their turbulent motions' form, organization to compose very long motions, and relationship to vortical structures. POD extracts highly energetic structures from flow fields and is one tool to further understand the turbulence physics. A variety of direct numerical simulations provide velocity fields suitable for detailed analysis. Since POD modes require significant interpretation, this study begins with wall-normal, one-dimensional POD for a set of turbulent channel flows. Important features of the modes and their scaling are interpreted in light of flow physics, also leading to a method of synthesizing one-dimensional POD modes. Properties of a pipe flow simulation are then studied via several methods. The presence of very long streamwise motions is assessed using a number of statistical quantities, including energy spectra, which are compared to experiments. Further properties of energy spectra, including their relation to fictitious forces associated with mean Reynolds stress, are considered in depth. After reviewing salient features of turbulent structures previously observed in relevant experiments, structures in the pipe flow are examined in greater detail. A variety of methods reveal organization patterns of structures in instantaneous fields and their associated vortical structures. Properties of POD modes for a boundary layer flow are considered. Finally, very wide modes that occur when computing POD modes in all three canonical flows are compared. The results demonstrate that POD extracts structures relevant to characterizing wall-bounded turbulent flows. However, significant care is necessary in interpreting POD results, for which modes can be categorized according to their self-similarity. Additional analysis techniques reveal the organization of smaller motions in characteristic patterns to compose very long motions in pipe flows. The very large scale motions are observed to contribute large fractions of turbulent kinetic energy and Reynolds stress. The associated vortical structures possess characteristics of hairpins, but are commonly distorted from pristine hairpin geometries.
ContributorsBaltzer, Jon Ronald (Author) / Adrian, Ronald J (Thesis advisor) / Calhoun, Ronald (Committee member) / Gelb, Anne (Committee member) / Herrmann, Marcus (Committee member) / Squires, Kyle D (Committee member) / Arizona State University (Publisher)
Created2012
Description
This dissertation presents a volume filtering framework to solve particle-laden flows. Particle-laden flows are studied, employing the well-established Euler-Lagrange method, using the point-particle approximation. This approach requires the filter width to be much larger than the particle diameter. The method assumes that the particle is smaller than the Kolmogorov length scale. This thesis investigates how inertial particles at semi-dilute volume fractions modulate the flow characteristics for particles smaller than 1 in wall units, when dispersed within wall-bounded channel flows at friction Reynolds number of 180. The simulations are performed with 4 way coupling in order to account for high local concentration of particles, to capture mechanisms such as turbophoresis and preferential concentration. We show that drag attenuation or augmentation is determined by the particle inertia. As particle size is increased greater than 1 in wall units, the regime becomes finite-sized, requiring an interface-resolved description. To do this a novel Immersed Boundaries (IB) framework based on the concept of volume-filtering called the Volume-Filtered Immersed Boundary (VF-IB) method is presented. Transport equations are obtained by volume-filtering the Navier-Stokes equation and accounting for the stresses at the solid-fluid interface. Boundary conditions are transformed into bodyforces that appear as surface integrals on the right hand side of the filtered equation. The approach requires the filter width to be much smaller than the particle diameter in order to accurately resolve the interfacial dynamics. Several canonical tests are conducted for both stationary and moving immersed solids and report comparable results to the experimental and/or body-fitted simulations. Keep in mind, the VF-IB method reverts back to the Euler-Lagrange formulation if the filter width is significantly greater than the particle diameter. An artifact of volume-filtering is the emergence of unclosed terms we define as the sub-filter scale term. In order to characterize the contribution of this term on the solution, a more simpler case of a 2-D varying coefficient hyperbolic equation that has an exact solution is looked into. It is observed that the sub-filter scale term scales inversely with the square of the filter width. For fine interface resolution (i.e. small filter width), this value can be ignored with negligible effect to the accuracy of the numerical solution. However for coarse interface resolution (i.e. large filter width), including the sub-filter scale term significantly increases the accuracy of the numerical solution
ContributorsDave, Himanshu (Author) / Kasbaoui, Mohamed Houssem (Thesis advisor) / Herrmann, Marcus (Thesis advisor) / Dahm, Werner (Committee member) / Kim, Jeonglae (Committee member) / Lopez, Juan (Committee member) / Arizona State University (Publisher)
Created2024
Description
Advancements to a dual scale Large Eddy Simulation (LES) modeling approach for immiscible turbulent phase interfaces are presented. In the dual scale LES approach, a high resolution auxiliary grid, used to capture a fully resolved interface geometry realization, is linked to an LES grid that solves the filtered Navier-Stokes equations. Exact closure of the sub-filter interface terms is provided by explicitly filtering the fully resolved quantities from the auxiliary grid. Reconstructing a fully resolved velocity field to advance the phase interface requires modeling several sub-filter effects, including shear and accelerational instabilities and phase change. Two sub-filter models were developed to generate these sub-filter hydrodynamic instabilities: an Orr-Sommerfeld model and a Volume-of-Fluid (VoF) vortex sheet method. The Orr-Sommerfeld sub-filter model was found to be incompatible with the dual scale approach, since it is unable to generate interface rollup and a process to separate filtered and sub-filter scales could not be established. A novel VoF vortex sheet method was therefore proposed, since prior vortex methods have demonstrated interface rollup and following the LES methodology, the vortex sheet strength could be decomposed into its filtered and sub-filter components. In the development of the VoF vortex sheet method, it was tested with a variety of classical hydrodynamic instability problems, compared against prior work and linear theory, and verified using Direct Numerical Simulations (DNS). An LES consistent approach to coupling the VoF vortex sheet with the LES filtered equations is presented and compared against DNS. Finally, a sub-filter phase change model is proposed and assessed in the dual scale LES framework with an evaporating interface subjected to decaying homogeneous isotropic turbulence. Results are compared against DNS and the interplay between surface tension forces and evaporation are discussed.
ContributorsGoodrich, Austin Chase (Author) / Herrmann, Marcus (Thesis advisor) / Dahm, Werner (Committee member) / Kim, Jeonglae (Committee member) / Huang, Huei-Ping (Committee member) / Kostelich, Eric (Committee member) / Arizona State University (Publisher)
Created2023
Description
Autonomic closure is a new general methodology for subgrid closures in large eddy simulations that circumvents the need to specify fixed closure models and instead allows a fully- adaptive self-optimizing closure. The closure is autonomic in the sense that the simulation itself determines the optimal relation at each point and time between any subgrid term and the variables in the simulation, through the solution of a local system identification problem. It is based on highly generalized representations of subgrid terms having degrees of freedom that are determined dynamically at each point and time in the simulation. This can be regarded as a very high-dimensional generalization of the dynamic approach used with some traditional prescribed closure models, or as a type of “data-driven” turbulence closure in which machine- learning methods are used with internal training data obtained at a test-filter scale at each point and time in the simulation to discover the local closure representation.
In this study, a priori tests were performed to develop accurate and efficient implementations of autonomic closure based on particular generalized representations and parameters associated with the local system identification of the turbulence state. These included the relative number of training points and bounding box size, which impact computational cost and generalizability of coefficients in the representation from the test scale to the LES scale. The focus was on studying impacts of these factors on the resulting accuracy and efficiency of autonomic closure for the subgrid stress. Particular attention was paid to the associated subgrid production field, including its structural features in which large forward and backward energy transfer are concentrated.
More than five orders of magnitude reduction in computational cost of autonomic closure was achieved in this study with essentially no loss of accuracy, primarily by using efficient frame-invariant forms for generalized representations that greatly reduce the number of degrees of freedom. The recommended form is a 28-coefficient representation that provides subgrid stress and production fields that are far more accurate in terms of structure and statistics than are traditional prescribed closure models.
In this study, a priori tests were performed to develop accurate and efficient implementations of autonomic closure based on particular generalized representations and parameters associated with the local system identification of the turbulence state. These included the relative number of training points and bounding box size, which impact computational cost and generalizability of coefficients in the representation from the test scale to the LES scale. The focus was on studying impacts of these factors on the resulting accuracy and efficiency of autonomic closure for the subgrid stress. Particular attention was paid to the associated subgrid production field, including its structural features in which large forward and backward energy transfer are concentrated.
More than five orders of magnitude reduction in computational cost of autonomic closure was achieved in this study with essentially no loss of accuracy, primarily by using efficient frame-invariant forms for generalized representations that greatly reduce the number of degrees of freedom. The recommended form is a 28-coefficient representation that provides subgrid stress and production fields that are far more accurate in terms of structure and statistics than are traditional prescribed closure models.
ContributorsKshitij, Abhinav (Author) / Dahm, Werner J.A. (Thesis advisor) / Herrmann, Marcus (Committee member) / Hamlington, Peter E (Committee member) / Peet, Yulia (Committee member) / Kim, Jeonglae (Committee member) / Arizona State University (Publisher)
Created2019
Description
Rapid expansion of dense beds of fine, spherical particles subjected to rapid depressurization is studied in a vertical shock tube. As the particle bed is unloaded, a high-speed video camera captures the dramatic evolution of the particle bed structure. Pressure transducers are used to measure the dynamic pressure changes during the particle bed expansion process. Image processing, signal processing, and Particle Image Velocimetry techniques, are used to examine the relationships between particle size, initial bed height, bed expansion rate, and gas velocities.
The gas-particle interface and the particle bed as a whole expand and evolve in stages. First, the bed swells nearly homogeneously for a very brief period of time (< 2ms). Shortly afterward, the interface begins to develop instabilities as it continues to rise, with particles nearest the wall rising more quickly. Meanwhile, the bed fractures into layers and then breaks down further into cellular-like structures. The rate at which the structural evolution occurs is shown to be dependent on particle size. Additionally, the rate of the overall bed expansion is shown to be dependent on particle size and initial bed height.
Taller particle beds and beds composed of smaller-diameter particles are found to be associated with faster bed-expansion rates, as measured by the velocity of the gas-particle interface. However, the expansion wave travels more slowly through these same beds. It was also found that higher gas velocities above the the gas-particle interface measured \textit{via} Particle Image Velocimetry or PIV, were associated with particle beds composed of larger-diameter particles. The gas dilation between the shocktube diaphragm and the particle bed interface is more dramatic when the distance between the gas-particle interface and the diaphragm is decreased-as is the case for taller beds.
To further elucidate the complexities of this multiphase compressible flow, simple OpenFOAM (Weller, 1998) simulations of the shocktube experiment were performed and compared to bed expansion rates, pressure fluctuations, and gas velocities. In all cases, the trends and relationships between bed height, particle diameter, with expansion rates, pressure fluctuations and gas velocities matched well between experiments and simulations. In most cases, the experimentally-measured bed rise rates and the simulated bed rise rates matched reasonably well in early times. The trends and overall values of the pressure fluctuations and gas velocities matched well between the experiments and simulations; shedding light on the effects each parameter has on the overall flow.
The gas-particle interface and the particle bed as a whole expand and evolve in stages. First, the bed swells nearly homogeneously for a very brief period of time (< 2ms). Shortly afterward, the interface begins to develop instabilities as it continues to rise, with particles nearest the wall rising more quickly. Meanwhile, the bed fractures into layers and then breaks down further into cellular-like structures. The rate at which the structural evolution occurs is shown to be dependent on particle size. Additionally, the rate of the overall bed expansion is shown to be dependent on particle size and initial bed height.
Taller particle beds and beds composed of smaller-diameter particles are found to be associated with faster bed-expansion rates, as measured by the velocity of the gas-particle interface. However, the expansion wave travels more slowly through these same beds. It was also found that higher gas velocities above the the gas-particle interface measured \textit{via} Particle Image Velocimetry or PIV, were associated with particle beds composed of larger-diameter particles. The gas dilation between the shocktube diaphragm and the particle bed interface is more dramatic when the distance between the gas-particle interface and the diaphragm is decreased-as is the case for taller beds.
To further elucidate the complexities of this multiphase compressible flow, simple OpenFOAM (Weller, 1998) simulations of the shocktube experiment were performed and compared to bed expansion rates, pressure fluctuations, and gas velocities. In all cases, the trends and relationships between bed height, particle diameter, with expansion rates, pressure fluctuations and gas velocities matched well between experiments and simulations. In most cases, the experimentally-measured bed rise rates and the simulated bed rise rates matched reasonably well in early times. The trends and overall values of the pressure fluctuations and gas velocities matched well between the experiments and simulations; shedding light on the effects each parameter has on the overall flow.
ContributorsZunino, Heather (Author) / Adrian, Ronald J (Thesis advisor) / Clarke, Amanda (Committee member) / Chen, Kangping (Committee member) / Herrmann, Marcus (Committee member) / Huang, Huei-Ping (Committee member) / Arizona State University (Publisher)
Created2019
Description
This work helps to explain the drag reduction mechanisms at low and moderate turbulent Reynolds numbers in pipe flows. Through direct numerical simulation, the effects of wall oscillations are observed on the turbulence in both the near wall and the bulk region. Analysis of the average Reynolds Stresses at various phases of the flow is provided along with probability density functions of the fluctuating components of velocity and vorticity. The flow is also visualized to observe, qualitatively, changes in the total and fluctuating field of velocity and vorticity. Linear Stochastic Estimation is used to create a conditional eddy (associated with stress production) in the flow and visualize the effects of transverse wall oscillations on hairpin growth, auto-generation and structure.
ContributorsCoxe, Daniel (Author) / Peet, Yulia (Thesis advisor) / Adrian, Ronald (Thesis advisor) / Herrmann, Marcus (Committee member) / Arizona State University (Publisher)
Created2019
Description
This dissertation describes a process for interface capturing via an arbitrary-order, nearly quadrature free, discontinuous Galerkin (DG) scheme for the conservative level set method (Olsson et al., 2005, 2008). The DG numerical method is utilized to solve both advection and reinitialization, and executed on a refined level set grid (Herrmann, 2008) for effective use of processing power. Computation is executed in parallel utilizing both CPU and GPU architectures to make the method feasible at high order. Finally, a sparse data structure is implemented to take full advantage of parallelism on the GPU, where performance relies on well-managed memory operations.
With solution variables projected into a kth order polynomial basis, a k+1 order convergence rate is found for both advection and reinitialization tests using the method of manufactured solutions. Other standard test cases, such as Zalesak's disk and deformation of columns and spheres in periodic vortices are also performed, showing several orders of magnitude improvement over traditional WENO level set methods. These tests also show the impact of reinitialization, which often increases shape and volume errors as a result of level set scalar trapping by normal vectors calculated from the local level set field.
Accelerating advection via GPU hardware is found to provide a 30x speedup factor comparing a 2.0GHz Intel Xeon E5-2620 CPU in serial vs. a Nvidia Tesla K20 GPU, with speedup factors increasing with polynomial degree until shared memory is filled. A similar algorithm is implemented for reinitialization, which relies on heavier use of shared and global memory and as a result fills them more quickly and produces smaller speedups of 18x.
With solution variables projected into a kth order polynomial basis, a k+1 order convergence rate is found for both advection and reinitialization tests using the method of manufactured solutions. Other standard test cases, such as Zalesak's disk and deformation of columns and spheres in periodic vortices are also performed, showing several orders of magnitude improvement over traditional WENO level set methods. These tests also show the impact of reinitialization, which often increases shape and volume errors as a result of level set scalar trapping by normal vectors calculated from the local level set field.
Accelerating advection via GPU hardware is found to provide a 30x speedup factor comparing a 2.0GHz Intel Xeon E5-2620 CPU in serial vs. a Nvidia Tesla K20 GPU, with speedup factors increasing with polynomial degree until shared memory is filled. A similar algorithm is implemented for reinitialization, which relies on heavier use of shared and global memory and as a result fills them more quickly and produces smaller speedups of 18x.
ContributorsJibben, Zechariah J (Author) / Herrmann, Marcus (Thesis advisor) / Squires, Kyle (Committee member) / Adrian, Ronald (Committee member) / Chen, Kangping (Committee member) / Treacy, Michael (Committee member) / Arizona State University (Publisher)
Created2015
Description
The Volume-of-Fluid method is a popular method for interface tracking in Multiphase applications within Computational Fluid Dynamics. To date there exists several algorithms for reconstruction of a geometric interface surface. Of these are the Finite Difference algorithm, Least Squares Volume-of-Fluid Interface Reconstruction Algorithm, LVIRA, and the Efficient Least Squares Volume-of-Fluid Interface Reconstruction Algorithm, ELVIRA. Along with these geometric interface reconstruction algorithms, there exist several volume-of-fluid transportation algorithms. This paper will discuss two operator-splitting advection algorithms and an unsplit advection algorithm. Using these three interface reconstruction algorithms, and three advection algorithms, a comparison will be drawn to see how different combinations of these algorithms perform with respect to accuracy as well as computational expense.
ContributorsKedelty, Dominic (Author) / Herrmann, Marcus (Thesis advisor) / Huang, Huei-Ping (Committee member) / Chen, Kangping (Committee member) / Arizona State University (Publisher)
Created2015