SEARCH · Engineering Papers
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.
Pseudocompressibility, upwind differencing, and other techniques used to solve Navier-Stokes equations. Scheme for finite-difference numerical solution of two-dimensional Navier-Stokes equations of incompressible flow combines several recently developed methods, each developed to increase speed and/or accuracy of computations of this kind.
A multigrid acceleration technique developed for solving 3-D Navier-Stokes equations for subsonic/transonic flows was extended to supersonic/hypersonic flows. An explicit multistage Runge-Kutta type of time stepping scheme is used as the basic algorithm in conjunction with the multigrid scheme. Solutions were obtained for a blunt conical frustum at Mach 6 to demonstrate the applicability of the multigrid scheme to high speed flows. Computations were performed for a generic High Speed Civil Transport configuration designed to cruise at Mach 3. These solutions show both the efficiency and accuracy of the present scheme for computing high speed viscous flows over configurations of practical interest.
Numerical solutions of the Navier-Stokes equations governing shear flow in a trapezoidal enclosure have been obtained at fine grid resolutions using an efficient multigrid calculation procedure. The effects of flow Reynolds number and wall boundary conditions are studied in detail. Interesting flow patterns are observed to develop, especially at high Reynolds numbers.
Many aircraft employ ducts with centerline curvature or changing cross-sectional shape to join the engine with inlet and exhaust components. S-ducts convey air to the engine compressor from the intake and often decelerate the flow to achieve an acceptable Mach number at the engine compressor by increasing the cross-sectional area downstream. Circular-to-rectangular transition ducts are used on aircraft with rectangular exhaust nozzles to connect the engine and nozzle. To achieve maximum engine performance, the ducts should minimize flow total pressure loss and total pressure distortion at the duct exit. Changes in the curvature of the duct centerline or the duct cross-sectional shape give rise to streamline curvature which causes cross stream pressure gradients. Secondary flows can be caused by deflection of the transverse vorticity component of the boundary layer. This vortex tilting results in counter-rotating vortices. Additionally, the adverse streamwise pressure gradient caused by increasing cross-sectional area can lead to flow separation. Vortex pairs have been observed in the exit planes of both duct types. These vortices are due to secondary flows induced by pressure gradients resulting from streamline curvature. Regions of low total pressure are produced when the vortices convect boundary layer fluid into the main flow. The purpose of the present study is to predict the measured flow field in a diffusing S-duct and a circular-to-rectangular transition duct with a full Navier-Stokes computer program, PARC3D, and to compare the numerical predictions with new detailed experimental measurements. The work was undertaken to extend previous studies and to provide additional CFD validation data needed to help model flows with strong secondary flow and boundary layer separation. The S-duct computation extends the study of Smith et al, and Harloff et al, which concluded that the computation might be improved by using a finer grid and more advanced turbulence models. The present study compares results for both the Baldwin-Lomas and k-epsilon turbulence models and is conducted with a refined grid. For the transition duct, two inlet conditions were considered, the first with straight flow and the second with swirling flow. The first case permits examination of the effects of the geometric transition on the flow field, while the second case includes the rotational flow effect characteristic of a gas turbine engine.
The present paper deals with problems of two-dimensional laminar and turbulent flow of a viscous incompressible fluid near the trailing edge of a thin flat plate. The complete system of Navier-Stokes equations is solved by means of MacCormack's finite-difference scheme. In the laminar case, the results are compared with the triple deck solution of Daniels (1973). The comparisons indicate that the asymptotic triple deck theory for supersonic trailing-edge flow is accurate within five percent for Reynolds numbers greater than 1000. In the turbulent case, the Prandtl-Van Driest-Clausner algebraic eddy viscosity model is used, showing that the region of upstream influence is roughly of the order of the boundary layer thickness.
Acceleration techniques such as Wynn's (1986) epsilon algorithm and analysis techiques such as eigensystem analysis are used here to study numerically the convergence properties of an iterative scheme applied to the quasi-one-dimensional Euler and Navier-Stokes equations for flow through nozzles with shocks. The convergence and stability properties are studied by analyzing the dependence of convergence of the code on the discretization technique, boundary conditions, time-step, number of grid points, and the physics of the problem.
A compressible Navier-Stokes solution procedure is applied to the flow about an isolated airfoil. Two major problem areas were investigated. The first area is that of developing a coordinate system and an initial step in this direction has been taken. An airfoil coordinate system obtained from specification of discrete data points developed and the heat conduction equation has been solved in this system. Efforts required to allow the Navier-Stokes equations to be solved in this system are discussed. The second problem area is that of obtaining flow field solutions. Solutions for the flow about a circular cylinder and an isolated airfoil are presented. In the former case, the prediction is shown to be in good agreement with data.
This paper describes experimental measurements of secondary flow in a constant area, circular cross-section 30-30 deg S-duct, and compares the results obtained with the computations performed using the PEPSIG code, a parabolized Navier-Stokes code. The flow entering the duct was turbulent, with entrance Mach number of 0.6, and the boundary layer thickness at the duct entrance was 10 percent of the duct diameter. The duct mean radius of curvature to the duct diameter was 5.077. Flow parameters were measured at six stations along the length of the duct. These measurements were made using a five-port cone probe. At least ten radial traverses were made at each station on both sides of the symmetry plane. Wall static pressures along three azimuth angles of zero, 90, and 180 deg along the duct were measured. Plots presenting the secondary velocity field as well as contour plots of the total and static-pressure fields have been obtained. Strong secondary flows were observed in the first bend, and these continued into the second bend with the formation of new vorticity in the opposite sense in the second bend. The flow exiting the duct contained two pairs of counter-rotating vortices. The computational results are in general agreement with the experiments. However, it appears that the computations underestimate the extent of the pressure distortion, due to simplifications made in the pressure field calculations.
Finite element computations and laboratory experiments were carried out to study the mutual wake interaction behind two plates placed side by side. The computations are based on the vorticity-stream function formation of the two-dimensional incompressible Navier-Stokes equations. The flow visualization studies indicate independent vortex sheddings at large gap sizes and metastable biased flow at smaller gap sizes. The computational results exhibit a similar behavior.
Algorithm numerically integrates Navier-Stokes equations of time-dependent or steady flow of incompressible, viscous fluid. Simulates realistic three-dimensional flows bounded by stationary or moving surface(s) of complicated shape(s). Only two arbitrary parameters specified by user. Algorithm described in "Numerical Simulation Of Flow Through An Artificial Heart" (ARC-12478). Algorithm tested by application to number of steady and unsteady flows, including flow in artificial heart discussed in noted prior article.
Comparisons of Navier-Stokes solutions and particle simulations for a simple two-dimensional model problem at a succession of altitudes are performed in order to assess the importance of rarefaction effects on the base flow region. In addition, an attempt is made to include 'Burnett-type' extensions to the Navier-Stokes constitutive relations. The model geometry consists of a simple blunted wedge with a 0.425 meter nose radius, a 70 deg cone half angle, a 1.7 meter base length, and a rounded shoulder. The working gas is monatomic with a molecular weight and viscosity similar to air and was chosen to focus the study on the continuum and particle methodologies rather than the implementation of thermo-chemical modeling. Three cases are investigated, all at Mach 29, with densities corresponding to altitudes of 92 km, 99 km, and 105 km. At the lowest altitude, Navier-Stokes solutions agree well with particle simulations. At the higher altitudes, the Navier-Stokes equations become less accurate. In particular, the Navier-Stokes equations and particle method predict substantially different flow turning angle in the wake near the after body. Attempts to achieve steady continuum solutions including 'Burnett-type' terms failed. Further research is required to determine whether the boundary conditions, the equations themselves, or other unknown causes led to this failure.
Methods for determining the effects of mass injection from the trailing edge of a bluff body at low speeds and in transonic flow were numerically studied along with an unmodified blunt-based body to gain insight into the effects of vortex shedding on the base drag. The methodology used to obtain finite-difference solutions to the Navier-Stokes equations for subsonic compressible two-dimensional near-wake flows is presented. The effectiveness of an introduced outflow boundary condition which minimizes reflections back into the computational domain was demonstrated with the solution of a model vortex problem. Calculations of the near-wake flow past a circular cylinder were in excellent agreement with experimental data. Laminar-flow solutions for a blunt-based model with and without a base cavity and with mass injection into the wake agreed qualitatively with experimental observations. The drag reduction capability provided by such base modifications was demonstrated.
Papers are presented which deal with results of theoretical research on aerodynamic flow problems requiring the use of advanced computers. Topics discussed include: viscous flows, boundary layer equations, turbulence modeling and Navier-Stokes equations, and internal flows.
The present conference covers topics concerning the measurement and calculation of interactive flows, together with problems posed by subsonic and transonic wings, missiles, and ships. Discussions are presented on the time-dependent finite difference simulation of unsteady interactive flows, Navier-Stokes equation methods, numerical solutions for spatially periodic boundary layers, the application of unsteady laminar tripple deck theory to viscous-inviscid interaction, the coupling of boundary layer and Euler equation solutions, and viscous-inviscid flow interactions. Also discussed are leading and trailing edge flows, three-dimensional wing flows, small disturbance calculations including entropy corrections, an inviscid computational method for tactical missiles, and boundary layer and flow separation characteristics of bodies of revolution at incidence.
A three-dimensional Navier-Stokes code has been applied to the analysis of flow fields containing propulsive jets. Specifically, the application was made to a flow field containing a supersonic jet injected at an angle of 90 degrees to a subsonic free stream. Although wind tunnel data were available, the computational results were not readily comparable to the experimental data because of significant differences between the two plume trajectories. Reasons for the differences are suggested in the report and include: (1) incomplete convergence, (2) inadequate grid resolution in the high gradient regions, and (3) use of a low-order turbulence closure model.
A consistent and efficient set of boundary conditions is developed for the multi-sweep space-marching pressure-elliptic Reduced Navier-Stokes (RNS) scheme as applied for 3-D internal viscous flow problems. No-slip boundary conditions are directly imposed on the solid walls. There is no iteration procedure required in the cross plane to ensure mass conservation across each marching plane. The finite difference equations forming the coefficient matrix are ordered such that the surface normal velocity is specified on all the solid walls; unlike external flows, a pressure boundary condition in the cross plane is not required. Since continuity is directly satisfied at all points in the flow domain, the first order momentum equations can be solved directly for the pressure without the need for a Poisson pressure correction equation. The procedure developed herein can also be applied with periodic boundary conditions. The analysis is given for general compressible flows. Incompressible flow solutions are obtained, for straight and curved ducts of square cross section, to validate the procedure. These solutions are used to demonstrate the applicability of the RNS scheme, with the improved boundary conditions for internal flows with strong interaction, as would be encountered in ducts and turbomachinery geometries.
Two-dimensional Euler and Navier-Stokes solutions of the flow through three inlet/diffuser configurations with terminal shock systems are reported. Calculations without bleed indicate that the terminal shock location is very sensitive to the outflow back pressure. For cases where there are no available experimental results, it becomes difficult to estimate the back pressure that will result in a terminal shock. Estimates based on quasi-one-dimensional analysis are not found adequate for complex two-dimensional flows. It is found that since the flow downstream of the terminal shock is subsonic, and what happens at the outflow boundary affects the flow inside the inlet, enough of the subsonic diffuser must be modeled to accurately predict the terminal shock region. The diffuser portion should be fairly long with the outflow boundary occurring in a region of more or less uniform flow to be able to prescribe a uniform back pressure. The third configuration studied was investigated with and without incorporating bleed in the code. It is found that the use of bleed stabilizes the shock location and allows solutions which without bleed result in unstarting of the inlet. Comparisons are made with available experimental data.
A theory is presented for calculating the flow in the core of a separation-induced leading-edge vortex. The method is based on matching inner and outer representations of the vortex. The inner model of the vortex is based on the quasicylindrical Navier-Stokes equations; the flow is assumed to be steady, axially symmetric, and incompressible and in addition, gradients in the radial direction are assumed to be much larger then gradients in the axial direction. The outer model is based on the three-dimensional free-vortex-sheet theory, a higher-order panel method which solves the Prandtl-Glauert equation including nonlinear boundary conditions pertinent to the concentrated vorticity representation of the leading edge vortex. The resultant flow is evaluated a posteriori for evidence of incipient vortex breakdown and the critical helix angle concept, in conjunction with an adverse longitudinal pressure gradient, is found to correlate well with the occurrence of vortex breakdown at the trailing edge of delta, arrow, and diamond wings.