SEARCH · Search NASA
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.
The artificial viscosity method of Kuwahara and Takami (1973) is used to calculate the roll-up of trailing vortices behind a number of practical aerodynamic configurations. Where possible, the results are compared for core location with available experimental data.
Artificial viscosity is added either implicitly or explicitly in practically every numerical scheme for suppressing spurious oscillations in the solution of fluid dynamics equations. In the present central-difference scheme, artificial viscosity is added explicitly for suppressing high frequency oscillations and achieving good convergence properties. The amount of artificial viscosity added is controlled through the use of pre-selected coefficients. In the standard scheme, scalar coefficients based on the spectral radii of the Jacobian of convective fluxes are used. However, this can add too much viscosity to the slower waves. Hence, use the matrix-valued coefficients, which give appropriate viscosity for each wave component is suggested. With the matrix-valued coefficients, the central-difference scheme produces more accurate solutions on a given grid, particularly in the vicinity of shocks, while still maintaining good convergence properties.
Artificial viscosity is added either implicity or explicitly in practically every numerical scheme for suppressing spurious oscillations in the solution of fluid-dynamics equations. In the present central-difference scheme, artificial viscosity is added explicitly for suppressing high-frequency oscillations and achieving good convergence properties. The amount of artificial viscosity added is controlled through the use of preselected coefficients. In the standard scheme, scalar coefficients based on the spectral radii of the Jacobian of the convective fluxes are used. However, this can add too much viscosity to the slower waves. Hence, the use of matrix-valued coefficients, which give appropriate viscosity for each wave component, is suggested. With the matrix-valued coefficients, the central-difference scheme produces more accurate solutions on a given grid, particularly in the vicinity of shocks and boundary layers, while still maintaining good convergence properties.
A method is proposed to determine directly the amount of artificial viscosity needed for stability using an eigenvalue analysis for a finite difference representation of the Navier-Stokes equations. The stability and growth of small perturbations about a steady flow over the airfoils are analyzed for various amounts of artificial viscosity. The eigenvalues were determined for a small perturbation about a steady inviscid flow over a NACA 0012 airfoil at a Mach number of 0.8 and angle of attack of 0 degrees. The movement of the eigenvalue constellation with respect to the amount of artificial viscosity is studied. The stability boundries as a function of the amount of artificial viscosity from both the eigenvalue analysis and the time marching scheme are also presented. This procedure not only allows for determining the effect of varying amounts of artificial viscosity, but also for the effects of different forms of terms for artificial viscosity.
A method is proposed to determine directly the amount of artificial viscosity needed for stability using an eigenvalue analysis for a finite difference representation of the Navier-Stokes equations. The stability and growth of small perturbations about a steady flow over airfoils are analyzed for various amounts of artificial viscosity. The eigenvalues were determined for a small time-dependent perturbation about a steady inviscid flow over an NACA 0012 airfoil at a Mach number of 0.8 and angle of attack of 0 deg. The method has been applied to inviscid flows here, but as discussed is also applicable to viscous flows. The movement of the eigenvalue constellation with respect to the amount of artificial viscosity is studied. The stability boundaries as a function of the amount of artificial viscosity from both the eigenvalue analysis and the time-marching scheme are also presented. The eigenvalue procedure not only allows for determining the effect of varying amounts of artificial viscosity, but also for the effects of different forms of artificial viscosity.
To accurately model inviscid flow with shock waves using staggered-grid Lagrangian hydrodynamics, artificial viscosity is introduced to convert kinetic energy into internal energy, thereby providing a mechanism to generate the required entropy increase across shocks. In this paper, we propose a new method for constructing an adaptive, artificial viscosity in the context of one-dimensional, staggered-grid Lagrangian hydrodynamics. Our adaptive, artificial viscosity is defined in terms of two parameters that depend on density locally in a neighborhood around a shock, and hence, vary cell-by-cell. Our methodology is based on building a reference set of pre-computed optimal, globally constant artificial viscosity coefficients for a family of isolated shock test problems. For arbitrary flows, the evaluation of the unknown coefficients is automated by first estimating shock intensity locally, and second computing the corresponding adaptive parameter value via interpolation over the data from the pre-computed reference set of optimal, globally constant coefficient values. To illustrate the performance of our new approach, we compare our results against two existing methods. The first method is a limiter-based approach, which relies on estimating velocity gradients of the flow, and the second method is utilized in several commercial codes. Finally, we demonstrate that our new adaptive methodology produces more accurate results for a variety of tests with propagating shock waves, as well as for the aforementioned family of isolated shock problems.
Abstract Shocked particle-laden flows are important to many natural and industrial processes. When simulating these systems, artificial viscosity is often required to prevent numerical artifacts, such as ringing, from arising in the pressure and density fields. The linear and quadratic coefficients of the artificial viscosity determine the amount of smoothing that occurs in these fields. For particle-laden flows, however, many of the fluid–particle interaction forces, for example, the pressure gradient force and unsteady forces, depend on gradients in the fluid fields. Furthermore, while the shock passes over a particle, these forces can be more dominant than drag. This means that the artificial viscosity coefficients affect how a particle and fluid interact when simulating shocked particle systems. Here this effect is investigated for isolated particles and for a particle curtain using a staggered grid Lagrangian approach. The artificial viscosity coefficients have a significant impact on the maximum force that a fluid imparts to a particle, which is important for determining whether a particle will break up in response to the shock. Furthermore, it is found that the density ratio between the particle and the fluid is important in determining whether the artificial viscosity coefficients have a significant impact on the particle’s motion.
To accurately model flows with shock waves using staggered-grid Lagrangian hydrodynamics, artificial viscosity has to be introduced to convert kinetic energy into internal energy, thereby increasing the entropy across shocks. Determining the appropriate strength of the artificial viscosity is an art and strongly depends on the particular problem and experience of the researcher. The objective of this study is to pose the problem of finding the appropriate strength of artificial viscosity as an optimization problem and solve this problem using machine learning (ML) tools, specifically using surrogate models based on Gaussian Process regression and Bayesian analysis. We describe the optimization method and discuss various practical details of its implementation. The shock-containing problems for which we apply this method all have been implemented in the LANL code FLAG. First, we apply ML to find optimal values to isolated shock problems of different strengths. Second, we apply ML to optimize viscosity for a 1D propagating detonation problem based on Zel’dovich-von Neumann-Doring (ZND) detonation theory using a reactive burn model. We compare results for default (currently used values in FLAG) and optimized values of artificial viscosity for these problems demonstrating the potential for significant improvement in the accuracy of computations.
Standard techniques used to model chemically-reacting flows require an artificial viscosity for stability in the presence of strong shocks. The resulting shock is smeared over at least three computational cells, so that the thickness of the shock is dictated by the structure of the overall mesh and not the shock physics. A gas passing through a strong shock is thrown into a nonequilibrium state and subsequently relaxes down over some finite distance to an equilibrium end state. The artificial smearing of the shock envelops this relaxation zone which causes the chemical kinetics of the flow to be altered. A method is presented which can investigate these issues by following the chemical kinetics and flow kinetics of a gas passing through a fully resolved shock wave at hypersonic Mach numbers. A nonequilibrium chemistry model for air is incorporated into a spectral multidomain Navier-Stokes solution method. Since no artificial viscosity is needed for stability of the multidomain technique, the precise effect of this artifice on the chemical kinetics and relevant flow features can be determined.
Standard techniques used to model chemically-reacting flows require an artificial viscosity for stability in the presence of strong shocks. The resulting shock is smeared over at least three computational cells, so that the thickness of the shock is dictated by the structure of the overall mesh and not the shock physics. A gas passing through a strong shock is thrown into a nonequilibrium state and subsequently relaxes down over some finite distance to an equilibrium end state. The artificial smearing of the shock envelops this relaxation zone which causes the chemical kinetics of the flow to be altered. A method is presented which can investigate these issues by following the chemical kinetics and flow kinetics of a gas passing through a fully resolved shock wave at hypersonic Mach numbers. A nonequilibrium chemistry model for air is incorporated into a spectral multidomain Navier-Stokes solution method. Since no artificial viscosity is needed for stability of the multidomain technique, the precise effect of this artifice on the chemical kinetics and relevant flow features can be determined.
Lagrangian methods to solve the inviscid Euler equations produce numerical oscillations near shock waves. A common approach to reducing these oscillations is to add artificial viscosity (AV) to the discrete equations. The AV term acts as a dissipative mechanism that attenuates oscillations by smearing the shock across a finite number of computational cells. However, AV introduces several control parameters that are not determined by the underlying physical model, and hence, in practice are tuned to the characteristics of a given problem. We seek to improve the standard quadratic-linear AV form by replacing it with a learned neural function that reduces oscillations relative to exact solutions of the Euler equations, resulting in a hybrid numerical-neural hydrodynamic solver. Because AV is an artificial construct that exists solely to improve the numerical properties of a hydrodynamic code, there is no offline ‘viscosity data’ against which a neural network can be trained before inserting into a numerical simulation, thus requiring online training. We achieve this via differentiable programming, i.e. end-to-end backpropagation or adjoint solution through both the neural and differential equation code, using automatic differentiation of the hybrid code in the Julia programming language to calculate the necessary loss function gradients. A novel offline pre-training step accelerates training by initializing the neural network to the default numerical AV scheme, which can be learned rapidly by space-filling sampling over the AV input space. We find that online training over early time steps of simulation is sufficient to learn a neural AV function that reduces numerical oscillations in long-term hydrodynamic shock simulations. These results offer an early proof-of-principle that online differentiable training of hybrid numerical schemes with novel neural network components can improve certain performance aspects existing in purely numerical schemes.
A finite volume method is used to calculate compressible inviscid flows over blunt bodies using, in general, unstructured grids. Artificial viscosity forms are derived based on a simplified least squares procedure. The extra second order terms are consistent with the governing equations, hence a systematic treatment of the numerical boundary conditions can be easily implemented. A special treatment of blunt bodies may be required. The discrete equations are linearized and the resulting system is solved by a relaxation method. Preliminary results indicate that the effect of the numerical dissipation is minimal. For subsonic flows over smooth bodies, the solution is practically vorticity-free and the total pressure loss is of the same order as the truncation error. Finally, some extensions of the present method are briefly discussed.
The total variation diminishing (TVD) finite difference scheme can be interpreted as a Lax-Wendroff scheme plus an upwind weighted artificial dissipation term. If a particular flux limiter is chosen and the requirement for upwind weighting is removed, an artificial dissipation term which is based on the theory of TVD schemes is obtained which does not contain any problem dependent parameters and which can be added to existing MacCormack method codes. Numerical experiments to examine the performance of this new method are discussed.
The computational fluid dynamics code, PARC3D, is tested to see if its use of non-physical artificial dissipation affects the accuracy of its results. This is accomplished by simulating a shock-laminar boundary layer interaction and several hypersonic flight conditions of the Pegasus(TM) launch vehicle using full artificial dissipation, low artificial dissipation, and the Engquist filter. Before the filter is applied to the PARC3D code, it is validated in one-dimensional and two-dimensional form in a MacCormack scheme against the Riemann and convergent duct problem. For this explicit scheme, the filter shows great improvements in accuracy and computational time as opposed to the nonfiltered solutions. However, for the implicit PARC3D code it is found that the best estimate of the Pegasus experimental heat fluxes and surface pressures is the simulation utilizing low artificial dissipation and no filter. The filter does improve accuracy over the artificially dissipative case but at a computational expense greater than that achieved by the low artificial dissipation case which has no computational time penalty and shows better results. For the shock-boundary layer simulation, the filter does well in terms of accuracy for a strong impingement shock but not as well for weaker shock strengths. Furthermore, for the latter problem the filter reduces the required computational time to convergence by 18.7 percent.
Wall functions, as used in the typical high Reynolds number k-epsilon turbulence model, can be implemented in various ways. A least disruptive method (to the flow solver) is to directly solve for the flow variables at the grid point next to the wall while prescribing the values of k and epsilon. For the centrally-differenced finite-difference scheme employing artificial viscocity (AV) as a stabilizing mechanism, this methodology proved to be totally useless. This is because the AV gives rise to a large error at the wall due to too steep a velocity gradient resulting from the use of a coarse grid as required by the wall function methodology. This error can be eliminated simply by extrapolating velocities at the wall, instead of using the physical values of the no-slip velocities (i.e. the zero value). The applicability of the technique used in this paper is demonstrated by solving a flow over a flat plate and comparing the results with those of experiments. It was also observed that AV gives rise to a velocity overshoot (about 1 percent) near the edge of the boundary layer. This small velocity error, however, can yield as much as 10 percent error in the momentum thickness. A method which integrates the boundary layer up to only the edge of the boundary (instead of infinity) was proposed and demonstrated to give better results than the standard method.
Analytical and experimental techniques for modeling the aerothermodynamics of hypersonic flight are assessed, together with the problems which will be encountered in developing reusable hypersonic vehicles. Emphasis is placed on a numerical coupling between a nonequilibrium chemistry model and hypersonic flow kinematics. Finite difference and finite element descriptions of flow fields in which molecules encounter a shock wave and undergo various motion (and thereby energy) transformations are discussed. The effects of artificial smearing of the shock wave are considered in terms of the resulting effects on the distribution of the energies and chemical composition of the transition region. Results are provided from schlieren photographs of shock-tube experiments, Navier-Stokes calculations of axisymmetric flow over a conical body, and calculations using a spectral multidomain approach for chemically reacting flows.
We have developed new measures of errors for numerical shock. The new approach is based on analysis of the structure function, and separation of the errors related to oscillations and shock width, which also include error in the position of the ”center” of the numerical shock. We have demonstrated that those measures correctly characterize the numerical solution. We introduced an objective function in, which both types of errors are weighted, and presented optimal values of the coefficients of the linear and quadratic viscosity for different weights and different Mach numbers.
We are often faced with trying to capture the physics of compressible shocks, which are governed by the Euler equations. However, Euler shocks are formally discontinuous at the shock front (translating to a step-function behavior of rel evant flow variables). This poses a practical problem for codes with finite-sized grid elements. As a result, one must make a concession in simulating the be havior of shocks within a discrete framework. In particular, we must blur, or ‘regularize’ Euler shocks so that they may be captured on a finite grid.