Browsing by Subject "Compressible flows"

Now showing 1 - 3 of 3
  • Thumbnail Image
    Item
    LES of High-Re Reacting flows: Active Scalar Conservation and Boundedness
    (2018-01) Kartha, Cheranellore Anand Vijay
    We aim to bring predictive capabilities to reacting flow simulations at high Reynolds numbers. In particular, we are interested in non-premixed reacting flows with realistic inflow conditions and heat release. These flows find their application in Scramjet engines, used for hypersonic propulsion. Higher-order, low-dissipation simulations of non-premixed combustion flows are often subject to numerical errors due to the presence of sharp gradients in species mass-fractions. These dispersive errors lead to overshoots and undershoots in species mass-fractions, resulting in violation of conservation of mass. In reacting flows, these errors result in overshoots of temperature above that allowed by the adiabatic flame temperature rise, rendering the simulations unreliable. To overcome this issue, we develop a new switched, low-dissipation flux methodology that mitigates these errors. The new method is validated on a range of one, two and three-dimensional problems, showing its effectiveness and promise to provide reliable solutions. We use this newly developed method to simulate chemically reacting, spatially evolving subsonic and supersonic mixing layers at high Reynolds numbers. We investigate the effect of inflow conditions on subsonic reacting mixing layers, following the experiments of Slessor et al. [1], performed at the California Institute of Technology. Results from the simulations show close agreement to the experimentally measured velocity and temperature profiles, indicating that the entrainment and heat release is predicted with good accuracy. We also observe that varying the inflow conditions changes the nature of entrainment into the mixing layers, consistent with the past experimental observations. We also investigate the effect of heat release in supersonic reacting flows in an inclined ramp geometry, following the work of Bonanos et al. [2]. Probability density function plots and mass-fraction isosurfaces of `tracer' species reveal that heat release significantly alters the flow field.
  • Thumbnail Image
    Item
    LES of turbulent cavitating flows using the homogeneous mixture model
    (2020-09) Bhatt, Mrugank
    The objective of this dissertation is to develop LES (large-eddy simulation) capabilities to study cavitation in complex hydrodynamic geometries. A fully-compressible homogeneous mixture model with a finite rate mass transfer is used in the simulations. The ability of the homogeneous mixture approach to capture resolved small-scale vapor bubbles is evaluated by a vapor bubble collapse problem. The effects of physical length scale, surface tension, driving pressure, and dimensionality of the problem are assessed using the parametric study. The finite rate effects of the cavitation model are discussed using the non-dimensional parameters and compared to the flow advection time scales. The expression for the finite rate mixture speed of sound is derived. Partial cavitation over incipient, transitory, and periodic regimes in the experimental sharp wedge configuration of Ganesh et. al. (2016) is investigated. The vapor void fractions obtained from LES shows very good agreement with X-ray measurements in each of the regimes. Physical mechanisms of cavity transition, both re-entrant jet and bubbly shock waves are captured in the LES. Conditions favoring the formation of either the re-entrant jet or the bubbly shock waves are studied through a detailed analysis of streamline curvature, vapor production, and vorticity transport. Flow over a five-bladed marine propeller is studied at design conditions. The assessment of propeller shaft orientation, numerical dissipation, the pressure drop in vortex cores, free-stream nuclei, and grid resolution revealed that the propeller performance is sensitive to the free-stream nuclei content, lower values showing a better comparison to the experiments. A numerical approach based on the preconditioning and the DTS is proposed to address the acoustic stiffness; thereby, enabling the low free-stream nuclei calculations. The novelty of the method lies in the application of preconditioning to a fully-compressible cavitation solver; where the characteristic-based filtering is modified based on the all-speed Roe-type scheme in addition to the traditional time-derivative matrix. The results are demonstrated for the unsteady flow over a cylinder under wetted and cavitation inception conditions, and the LES of low over a propeller under wetted conditions.
  • Thumbnail Image
    Item
    Numerical Study of High-Speed Transition due to Passive and Active Trips
    (2019-04) Shrestha, Prakash
    Transitional hypersonic boundary layers due to passive and active trips on a flat plate are studied using direct numerical simulations (DNS). In the case of passive trips (diamond- shaped and cylindrical), three dynamically prominent flow structures are consistently observed in both their isolated and distributed configurations. These flow structures are the upstream vortex system, the shock system, and the shear layers and the counter-rotating streamwise vortices from the wake of the trips. Analysis of the power spectral density (PSD) reveals the dominant source of instability due to the diamond-shaped trips as a coupled system of the shear layers and the counter-rotating streamwise vortices irrespective of spanwise trip-spacing. However, the dominant source of instability due to an array of cylindrical trips (Williams et al. 2018) is observed to be the upstream vortex system similar to Subbareddy et al. 2014 who used an isolated cylindrical trip. Therefore, the shape of a roughness element plays an essential role in the instability mechanism. Furthermore, dynamic mode decomposition (DMD) of three-dimensional snapshots of pressure fluctuations unveil globally dominant modes consistent with the PSD analysis in all the trip configurations. Higher peak-amplitude frequencies and amplitudes characterize dominant instabilities in higher freestream Reynolds number flows. When the trip heights are reduced, the source of instability has been observed to be unchanged, while peak-amplitude frequencies, the mean upstream recirculation zone, the mean instability-onset location, and the maximum turbulent kinetic energy is found to be reduced. When the trip spacing is greater than three times the trip width, each trip of the trip array becomes isolated. In the case of active trips, a two-dimensional (2-D) sonic jet from a straight slot is injected into Mach-10 three-dimensional (3-D) laminar boundary layers (Berry et al. 2004). The dynamically dominant flow structures observed in the vicinity of the jet correspond to upstream and downstream separation bubbles, where the number and the size of these bubbles vary with the injector pressure. A higher injector pressure leads to the formation of larger bubbles that cause the flow to become more unstable, resulting in a sequence of three successive bifurcations: (1) steady 2-D bubble formation, (2) transition from 2-D steady to 3-D quasi-unsteady bubble, and (3) transition from 3-D quasi-unsteady to 3-D unsteady bubble. This finding indicates that specific injector pressures are required to control the onset of transition in the laminar boundary layers. Streamwise streaks with a dominant spanwise wavelength are observed in both 3-D quasi-steady and 3-D unsteady flows. DMD of spanwise velocity reveals that the streamwise streaks originate from the upstream bubbles. In particular, the streaks arise from the coupled undulation of a primary upstream bubble and the upstream secondary bubble, which causes the flow to bifurcate from 2-D steady to 3-D quasi-unsteady. It is proposed that the source of the unsteadiness observed is generated by high-pressure fluctuations present between the secondary bubble and the jet. The unsteady interaction between the secondary bubble and the jet selects a specific wavelength of the spanwise undulation of the secondary bubble, which then modulates the primary bubble across span with the same wavelength. These two bubbles emanate two flow structures that have opposite spanwise velocities. These flow structures then travel to the top of the downstream bubbles to form a streamwise streak. The spanwise wavelength of the dominant DMD mode agrees with that of the streaks observed in the DNS. The simulation data in all cases agree well with their corresponding experiment. No effect of real gas has been found in this current study. The source of instability is observed to be independent of the thermal nature of the wall (isothermal or adiabatic). The angle of injection is observed to play a significant role in flow unsteadiness downstream of the jet. The mean Mach-disk height and the mean upstream recirculation length are compared to existing models in order to access their accuracy under the present flow and jet configurations.