Engineering Papers⌕ Search

SEARCH · Engineering Papers

Results for “Conditional Normalizing Flow”

Search indexed NASA NTRS and DOE OSTI research on propulsion, heat transfer, battery materials and energy systems. Follow report and document links to the original sources.

Quote a phrase for an exact phrase match. Source license links do not imply unrestricted reuse.

At least 469 records · Page 26

Calculation of three-dimensional (3-D) internal flow by means of the velocity-vorticity formulation on a staggered grid

A method has been developed to accurately compute the viscous flow in three-dimensional (3-D) enclosures. This method is the 3-D extension of a two-dimensional (2-D) method developed for the calculation of flow over airfoils. The 2-D method has been tested extensively and has been shown to accurately reproduce experimental results. As in the 2-D method, the 3-D method provides for the non-iterative solution of the incompressible Navier-Stokes equations by means of a fully coupled implicit technique. The solution is calculated on a body fitted computational mesh incorporating a staggered grid methodology. In the staggered grid method, the three components of vorticity are defined at the centers of the computational cell sides, while the velocity components are defined as normal vectors at the centers of the computational cell faces. The staggered grid orientation provides for the accurate definition of the vorticity components at the vorticity locations, the divergence of vorticity at the mesh cell nodes and the conservation of mass at the mesh cell centers. The solution is obtained by utilizing a fractional step solution technique in the three coordinate directions. The boundary conditions for the vorticity and velocity are calculated implicitly as part of the solution. The method provides for the non-iterative solution of the flow field and satisfies the conservation of mass and divergence of vorticity to machine zero at each time step. To test the method, the calculation of simple driven cavity flows have been computed. The driven cavity flow is defined as the flow in an enclosure driven by a moving upper plate at the top of the enclosure. To demonstrate the ability of the method to predict the flow in arbitrary cavities, results will he shown for both cubic and curved cavities.

Stremel, Paul M.↗

Flammability Aspects of a Cotton-Fiberglass Fabric in Opposed and Concurrent Airflow in Microgravity

Microgravity combustion tests burning fabric samples were performed aboard the International Space Station. The cotton-fiberglass blend samples were mounted inside a small wind tunnel which could impose air flow speeds up to 40 cm/s. The wind tunnel was installed in the Microgravity Science Glovebox which supplied power, imaging, and a level of containment. The effects of air flow speed on flame appearance, flame growth, and spread rates were determined in both the opposed and concurrent flow configuration. For the opposed flow configuration, the flame quickly reached steady spread for each flow speed, and the spread rate was fastest at an intermediate value of flow speed. These tests show the enhanced flammability in microgravity for this geometry, since, in normal gravity air, a flame self-extinguishes in the opposed flow geometry (downward flame spread). In the concurrent flow configuration, flame size grew with time during the tests. A limiting length and steady spread rate were obtained only in low flow speeds ( 10 cm/s) for the short-length samples that fit in the small wind tunnel. For these conditions, flame spread rate increased linearly with increasing flow. This is the first time that detailed transient flame growth data was obtained in purely forced flows in microgravity. In addition, by decreasing flow speed to a very low value (around 1 cm/s), quenching extinction was observed. The valuable results from these long-duration experiments validate a number of theoretical predictions and also provide the data for a transient flame growth model under development.

Ferkul, Paul V.↗

An experimental study of the effect of tail configuration on the spinning characteristics of general aviation aircraft

The feasibility of using static wind tunnel tests to obtain information about spin damping characteristics of an isolated general aviation aircraft tail was investigated. A representative tail section was oriented to the tunnel free streamline at angles simulating an equilibrium spin. A full range of normally encountered spin conditions was employed. Results of parametric studies performed to determine the effect of spin damping on several tail design parameters show satisfactory agreement with NASA rotary balance tests. Wing and body interference effects are present in the NASA studies at steep spin attitudes, but agreement improves with increasing pitch angle and spin rate, suggesting that rotational flow effects are minimal. Vertical position of the horizontal stabilizer is found to be a primary parameter affecting yaw damping, and horizontal tail chordwise position induces a substantial effect on pitching moment.

Ballin, M. G.↗

Finite volume solultion of the compressible boundary layer equations

A box-type finite volume discretization is applied to the integral form of the compressible boundary layer equations. Boundary layer scaling is introduced through the grid construction: streamwise grid lines follow eta = y/h = const., where y is the normal coordinate and h(x) is a scale factor proportional to the boundary layer thickness. With this grid, similarity can be applied explicitly to calculate initial conditions. The finite volume method preserves the physical transparency of the integral equations in the discrete approximation. The resulting scheme is accurate, efficient, and very simple. Computations for similar and non-similar flows show excellent agreement with tabulated results, solutions computed with Keller's Box scheme, and experimental data.

Loyd, B.↗

Finite volume solution of the compressible boundary-layer equations

A box-type finite volume discretization is applied to the integral form of the compressible boundary layer equations. Boundary layer scaling is introduced through the grid construction: streamwise grid lines follow eta = y/h = const., where y is the normal coordinate and h(x) is a scale factor proportional to the boundary layer thickness. With this grid, similarity can be applied explicity to calculate initial conditions. The finite volume method preserves the physical transparency of the integral equations in the discrete approximation. The resulting scheme is accurate, efficient, and conceptually simple. Computations for similar and non-similar flows show excellent agreement with tabulated results, solutions computed with Keller's Box scheme, and experimental data.

Loyd, B.↗

A survey of oscillating flow in Stirling engine heat exchangers

Similarity parameters for characterizing the effect of flow oscillation on wall shear stress, viscous dissipation, pressure drop and heat transfer rates are proposed. They are based on physical agruments and are derived by normalizing the governing equations. The literature on oscillating duct flows, regenerator and porous media flows is surveyed. The operating characteristics of the heat exchanger of eleven Stirling engines are discribed in terms of the similarity parameters. Previous experimental and analytical results are discussed in terms of these parameters and used to estimate the nature of the oscillating flow under engine operating conditions. The operating points for many of the modern Stirling engines are in or near the laminar to turbulent transition region. In several engines, working fluid does not pass entirely through heat exchangers during a cycle. Questions that need to be addressed by further research are identified.

Simon, Terrence W.↗

Focused Schlieren flow visualization studies of multiple venturi fuel injectors in a high pressure combustor

Multiple venturi fuel injectors were used to obtain uniform fuel distributions, better atomization and vaporization in the premixing/prevaporizing section of a lean premixed/prevaporized flame tube combustor. A focused Schlieren system was used to investigate the fuel/air mixing effectiveness of various fuel injection configurations. The Schlieren system was focused to a plane within the flow field of a test section equipped with optical windows. The focused image plane was parallel to the axial direction of the flow and normal to the optical axis. Images from that focused plane, formed by refracted light due to density gradients within the flow field, were filmed with a high-speed movie camera at framing rates of 8,000 frames per second (fps). Three fuel injection concepts were investigated by taking high-speed movies of the mixture flows at various operating conditions. The inlet air temperature was varied from 600 F to 1000 F, and inlet pressures from 80 psia to 150 psia. Jet-A fuel was used typically at an equivalence ratio of 0.5. The intensity variations of the digitized Schlieren images were analytically correlated to spatial density gradients of the mixture flows. Qualitative measurements for degree of mixedness, intensity of mixing, and mixing completion time are shown. Various mixing performance patterns are presented with different configurations of fuel injection points and operating conditions.

Chun, K. S.↗

Aeroacoustic computation of gust-blade interaction

To better understand and address the challenges faced in computing the acoustics of flow fields, test problems must be considered. In the present study, the sound radiated by the interaction of a flat plate with an oncoming gust containing a two component, mean velocity is computed. The gust has a uniform mean flow in x with Mach number M(infinity) equal to 0.5. The gust's mean velocity in y is of smaller amplitude and is given by: v = 0.1 sin(pi/8(x/M(sub infinity) - t)). This problem has been posed for an upcoming ICASE/LaRC workshop on benchmark problems in computational aeroacoustics. A plate with a length of 30 units in x is used. The plate is assumed to be infinitesimally thin and is centered at the origin. All variables are made dimensionless using the scales specified. Acoustic quantities are obtained by numerically integrating the linearized Euler equations. Integration is performed on the computational domain -100.0 less than or equal to x less than or equal to 100.0, -100.0 less than or equal to y less than or equal to 100.0, using unit length grid spacing in x and in y. An integration scheme is sought which will provide accurate solution to the small quantities of interest at a minimal computational expense. Results indicate that with the given discretization a scheme of minimal fourth order accuracy might be adequate to approximate the waves within the given flow. Thus, a variation of the MacCormack scheme with fourth order accuracy in space and second order accuracy in time was chosen. A scheme with sixth order accuracy in space has also been implemented and results compared with those of the fourth order accurate scheme. To ensure no mass flux, zero normal velocity is assigned at the plate. This condition will induce a discontinuity in the pressure across the plate location. Values for the perturbation pressure p' along the surface of the plate are obtained using a one-sided, third order Taylor expansion, such that p'(sub y) = O. In accordance with the Kutta condition, perturbation pressure at the trailing edge is assigned to zero. In the far field, radiation boundary conditions have been implemented. The effectiveness of the far field conditions are validated by computing in a larger computational domain and comparing the results. Early in time, sound waves begin to radiate from the plate. A Doppler effect is observed. After the initial transients disappear, the strongest waves leave the trailing edge at an approximate 45 degree angle. The intensity pattern of pressure fluctuations shows five lobes (of increasing magnitude with increasing downstream direction) emerging from the plate. Undesirable short wave contaminants are observed in the computed pressure distribution along the plate surface. For a more accurate solution at small scales, a more refined discretization will be required.

Martin, James E.↗

Smoldering News From STS-77 Endeavour

The Microgravity Smoldering Combustion (MSC) experiment lifted off aboard the Space Shuttle Endeavour for its second flight in May 1996, as part of the STS-77 mission. This experiment is part of a series of studies focused on the smolder characteristics of porous combustible materials in a microgravity environment. Smoldering is a nonflaming form of combustion that takes place in the interior of combustible materials. Common examples of smoldering are nonflaming embers, charcoal briquettes, and cigarettes. The objective of this study is to provide a better understanding of the controlling mechanisms of smoldering in microgravity and normal Earth gravity (1g). As with other forms of combustion, gravity affects the availability of air and transport of heat, and therefore, the rate of combustion. The results of the microgravity experiments will be compared with identical ones carried out in 1g. In addition, they will be used to verify present theories of smolder combustion and will provide new insights into the process of smoldering combustion, enhancing our fundamental understanding of this frequently encountered combustion process and guiding improvements in fire safety practices. Two smoldering combustion tests with polyurethane foam were successfully accomplished during the STS-77 mission. The tests investigated smoldering combustion in a quiescent (no-flow) enriched oxygen environment, and in an air environment with a 2-mm/sec airflow through the fuel sample. The primary data from the tests are the ignition characteristics, spread rate, smolder reaction temperature, and products of combustion (solid and gas). On both the first mission on STS-69 and the second mission on STS-77, a smolder front propagated the length of the forced-flow samples, with the spread rate between the corresponding upward and downward 1g smolder rates. Neither of the quiescent cases propagated combustion (the first case was due in part to a problem with the experiment electronics). These results show a dramatic difference from the normal gravity results, where smolder propagation is very rapid and complete for both of these conditions. The experiment was conceived by Prof. A. Carlos Fernandez-Pello at the University of California-Berkeley. The MSC hardware was designed and built at the NASA Lewis Research Center by a team of civil servants and contractors from NYMA, Inc., and Aerospace Design & Fabrication, Inc. (ADF). The hardware consists of two sealed aluminum combustion chambers (each being a half a cylinder). The chambers hold the MSC test section, data acquisition electronics, power distribution electronics, and instrumentation. The hardware is fitted into a 5-ft(sup 3) Get-Away-Special (GAS) canister that is mounted in the shuttle cargo bay. The test section consists of a quartz cylinder that contains the polyurethane foam sample and an igniter. This igniter, which is an electrically heated wire sandwiched between two porous ceramic disks, is mounted in contact with the end of the foam sample. An array of 12 thermocouples placed axially and radially along the foam sample provide temperature histories, which are used to determine the rate of smolder propagation, and the characteristics of the reaction.

Koudelka, John M.↗

Combustion Instabilities Modeled

NASA Lewis Research Center's Advanced Controls and Dynamics Technology Branch is investigating active control strategies to mitigate or eliminate the combustion instabilities prevalent in lean-burning, low-emission combustors. These instabilities result from coupling between the heat-release mechanisms of the burning process and the acoustic flow field of the combustor. Control design and implementation require a simulation capability that is both fast and accurate. It must capture the essential physics of the system, yet be as simple as possible. A quasi-one-dimensional, computational fluid dynamics (CFD) based simulation has been developed which may meet these requirements. The Euler equations of mass, momentum, and energy have been used, along with a single reactive species transport equation to simulate coupled thermoacoustic oscillations. A very simple numerical integration scheme was chosen to reduce computing time. Robust boundary condition procedures were incorporated to simulate various flow conditions (e.g., valves, open ends, and choked inflow) as well as to accommodate flow reversals that may arise during large flow-field oscillations. The accompanying figure shows a sample simulation result. A combustor with an open inlet, a choked outlet, and a large constriction approximately two thirds of the way down the length is shown. The middle plot shows normalized, time-averaged distributions of the relevant flow quantities, and the bottom plot illustrates the acoustic mode shape of the resulting thermoacoustic oscillation. For this simulation, the limit cycle peak-to-peak pressure fluctuations were 13 percent of the mean. The simulation used 100 numerical cells. The total normalized simulation time was 50 units (approximately 15 oscillations), which took 26 sec on a Sun Ultra2.

Paxson, Daniel E.↗

Response of panels to turbulence-induced, surface-pressure fluctuations and resulting acoustic radiation to the flow field.

The response of finite rectangular panels, excited by turbulent boundary layers and the resulting acoustical radiation into the flow field is studied. The method of normal modes and the technique of spectral analysis are used to compute the power spectra of the panel amplitude and strain response. Computed results are compared with the corresponding response measurements for panels exposed to attached and separated flows at supersonic speeds and attached flow at subsonic speeds. The thin plate and the convected wave equations are coupled through the boundary condition on the panel to give a formula for the radiation pressure power spectra. A formula for the acoustical damping ratio of the panel is obtained through the principle of virtual work.-

Chyu, W. J.↗

Steady magnetic reconnection in three dimensions

The concept of magnetic reconnection, defined to occur when there is an electric field parallel to field lines which are potential reconnection locations and near which the field has an X-type topology in a plane normal to the field line, has been generalized to three-dimensional configurations. A continuum of neighboring potential singular lines is found to exist, one of which supports reconnection, depending upon the imposed flow or electric field. For the case of steady reconnection, the nearby flow and electric field are shown to be severely constrained in the ideal region by the condition that the electric field = 0 there. It is noted that reconnection may occur at singularities of the electric field where this constraint fails and there is singular plasma jetting.

Priest, E. R.↗

Preliminary analysis of measured sound propagation over various seasonal snow covers

Measurements of acoustic pulse propagation in the 5 to 500-Hz frequency band were conducted under various snow cover conditions during the 1989 to 1990 winter in New Hampshire. The objective was to determine the effect of snow cover thickness and other snow properties on the absorption of acoustic pulses. Blank pistol shots were used as the source of the acoustic waves, and geophones and microphones in an 80 m-long linear array served as receivers. Snow thicknesses ranged from 0.05 to 0.35 m, and densities varied from 100 to 350 kg m(sup -3) during the 10 separate measurement days. Preliminary analysis indicates that the peak pulse amplitude decayed in proportion to approx. gamma (sup -1.7) for most conditions and that the acoustic-to-seismic ratios varied from about 4 to 15 x 10(exp -6) m s(sup -1) Pa(sup -1). Theoretical waveforms were calculated for propagation in a homogeneous atmosphere using Attenborough's model of ground impedance. An automatic fitting procedure for the normalized experimental and theoretical waveforms was used to determine the effective flow resistivity of the snow covers, and gave values of 10 to 35 kN s m(sup -4), in agreement with earlier results.

Albert, Donald G.↗

An experimental/computational study of sharp fin induced shock wave/turbulent boundary layer interactions at Mach 5 - Experimental results

A combined experimental/computational study has been performed of sharp fin induced shock wave/turbulent boundary layer interactions at Mach 5. The current paper focuses on the experiments and analysis of the results. The experimental data include mean surface heat transfer, mean surface pressure distributions and surface flow visualization for fin angles of attack of 6, 8, 10, 12, 14 and 16-degrees at Mach 5 under a moderately cooled wall condition. Comparisons between the results and correlations developed earlier show that Scuderi's correlation for the upstream influence angle (recast in a conical form) is superior to other such correlations in predicting the current results, that normal Mach number based correlations for peak pressure heat transfer are adequate and that the initial heat transfer peak can be predicted using pressure-interaction theory.

Rodi, Patrick E.↗

Progressive wave expansions and open boundary problems

In this paper we construct progressive wave expansions and asymptotic boundary conditions for wave-like equations in exterior domains, including applications to electromagnetics, compressible flows and aero-acoustics. The development of the conditions will be discussed in two parts. The first part will include derivations of asymptotic conditions based on the well-known progressive wave expansions for the two-dimensional wave equations. A key feature in the derivations is that the resulting family of boundary conditions involves a single derivative in the direction normal to the open boundary. These conditions are easy to implement and an application in electromagnetics will be presented. The second part of the paper will discuss the theory for hyperbolic systems in two dimensions. Here, the focus will be to obtain the expansions in a general way and to use them to derive a class of boundary conditions that involve only time derivatives or time and tangential derivatives. Maxwell's equations and the compressible Euler equations are used as examples. Simulations with the linearized Euler equations are presented to validate the theory.

Hagstrom, T.↗

Instability Analysis of a Low-Density Gas Jet Injected into a High-Density Gas

The objective of this study was to determine the effects of buoyancy on the absolute instability of low-density gas jets injected into high-density gas mediums. Most of the existing analyses of low-density gas jets injected into a high-density ambient have been carried out neglecting effects of gravity. In order to investigate the influence of gravity on the near-injector development of the flow, a linear temporal stability analysis and a spatio-temporal stability analysis of a low-density round jet injected into a high-density ambient gas were performed. The flow was assumed to be isothermal and locally parallel; viscous and diffusive effects were ignored. The variables were represented as the sum of the mean value and a normal-mode small disturbance. An ordinary differential equation governing the amplitude of the pressure disturbance was derived. The velocity and density profiles in the shear layer, and the Froude number (signifying the effects of gravity) were the three important parameters in this equation. Together with the boundary conditions, an eigenvalue problem was formulated. Assuming that the velocity and density profiles in the shear layer to be represented by hyperbolic tangent functions, the eigenvalue problem was solved for various values of Froude number. The temporal growth rates and the phase velocity of the disturbances were obtained. It was found that the presence of variable density within the shear layer resulted in an increase in the temporal amplification rate of the disturbances and an increase in the range of unstable frequencies, accompanied by a reduction in the phase velocities of the disturbances. Also, the temporal growth rates of the disturbances were increased as the Froude number was reduced (i.e. gravitational effects increased), indicating the destabilizing role played by gravity. The spatio-temporal stability analysis was performed to determine the nature of the absolute instability of the jet. The roles of the density ratio, Froude number, Schmidt number, and the lateral shift between the density and velocity profiles on the jet s absolute instability were determined. Comparisons of the results with previous experimental studies show good agreement when the effects of these variables are combined together. Thus, the combination of these variables determines how absolutely unstable the jet will be. Experiments were carried out to observe the qualitative differences between a round low-density gas jet injected into a high-density gas (helium jet injected into air) and a round constant density jet (air jet injected into air). Flow visualizations and velocity measurements in the near-injector region of the helium jet show more mixing and spreading of the helium jet than the air jet. The vortex structures develop and contribute to the jet spreading causing the helium jet to oscillate.

Lawson, Anthony Layiwola↗

Comparison of theory and experiment for homogeneous turbulence with shear

Solutions for uniformly-sheared turbulence, in which the interaction of the turbulence with the mean shear dominates the turbulent self-interaction, are compared with experiment. An anisotropic spectral tensor, which appears general enough to represent the initial experimental turbulence, is used for the initial condition in the calculations. The evolution of one-point turbulence components and microscales, as well as two-point velocity correlations, are considered. In most cases the agreement with experiment is good. The theory correctly predicts the presence of a negative region for two-point longitudinal-velocity correlations only for point separations in the direction normal to the flow and the mean gradient.

Deissler, R. G.↗

Effect of Reynolds Number in Turbulent-Flow Range on Flame Speeds of Bunsen Burner Flames

The effect of flow conditions on the geometry of the turbulent Bunsen flame was investigated. Turbulent flame speed is defined in terms of flame geometry and data are presented showing the effect of Reynolds number of flow in the range of 3000 to 35,000 on flame speed for burner diameters from 1/4 to 1 1/8 inches and three fuels -- acetylene, ethylene, and propane. The normal flame speed of an explosive mixture was shown to be an important factor in determining its turbulent flame speed, and it was deduced from the data that turbulent flame speed is a function of both the Reynolds number of the turbulent flow in the burner tube and of the tube diameter.

Bollinger, Lowell M↗