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.
Computed spectra resemble von Karman spectra of frequencies of interest. Numerical technique yields simulated atmospheric-turbulence spectra that closely approximates von Karman spectra within frequency ranges of interest for aircraft response. Suitable for computation in that they represent stable systems and roll off as f-2 at high frequencies outside range of interest.
Motivated by a recent survey of experimental data, we examine data on the Kolmogorov spectrum constant in numerical simulations of isotropic turbulence, using results both from previous studies and from new direct numerical simulations over a range of Reynolds numbers (up to 240 on the Taylor scale) at grid resolutions up to 512(exp 3). It is noted that in addition to k(exp -5/3) scaling, identification of a true inertial range requires spectral isotropy in the same wavenumber range. We found that a plateau in the compensated three-dimensional energy spectrum at k(eta) approx. = 0.1 - -0.2, commonly used to infer the Kolmogorov constant from the compensated three-dimensional energy spectrum, actually does not represent proper inertial range behavior. Rather, a proper, if still approximate, inertial range emerges at k(eta) approx. = 0.02 - 0.05 when R(sub lambda) increases beyond 140. The new simulations indicate proportionality constants C(sub 1) and C in the one- and three-dimensional energy spectra respectively about 0.60 and 1.62. If the turbulence were perfectly isotropic then use of isotropy relations in wavenumber space (C(sub 1) = 18/55 C) would imply that C(sub 1) approx. = 0.53 for C = 1.62, in excellent agreement with experiments. However the one- and three-dimensional estimates are not fully consistent, because of departures (due to numerical and statistical limitations) from isotropy of the computed spectra at low wavenumbers. The inertial scaling of structure functions in physical space is briefly addressed. Since DNS is still restricted to moderate Reynolds numbers, an accurate evaluation of the Kolmogorov constant is very difficult. We focus on providing new insights on the interpretation of Kolmogorov 1941 similarity in the DNS literature and do not consider issues pertaining to the refined similarity hypotheses of Kolmogorov (K62).
Numerical simulations of the weak beam-plasma instability were done in the turbulent regime where small-scale trapping is a dominant feature of the instability, a regime with behavior not predicted by quasi-linear theory. The results of the simulations were compared with those of a specific model of the turbulence, the so-called 'turbulent trapping' model, which gives precise formulas for the particle correlation functions, and predicts a growth rate well enhanced over the quasi-linear value. It was found that the model gives accurate predictions for the correlation functions. On the other hand, while growth rates were enhanced over the quasi-linear values, the enhancements observed are smaller than expected from the quantitative predictions of the model.
Numerical simulations of homogeneous turbulence subject to buoyant forcing were performed. The presence of a mean temperature gradient combined with a gravitational field results in a forcing term in the momentum equations. The development of the turbulence was studied and compared to the decay of similar fields in the absence of gravity. In the buoyantly driven field, the vorticity is preferentially aligned with the intermediate eigenvector of the strain-rate tensor and the local temperature gradient is more likely to be aligned with the most compressive eigenvector. These relationships are qualitatively similar to those observed in previous shear flow results studied by Ashurst (1987). A tensor diffusivity model for passive scalar transport developed from shear flow results in Rogers, Moin, and Reynolds (1986) also predicts this buoyant scalar transport, indicating that the relationship between the scalar flux and the Reynolds stress is similar in both flows.
The current computational methods for solving transport equations of turbulent reacting single-phase flows are critically reviewed, with primary attention given to those methods that lead to model-free simulations. In particular, consideration is given to direct numerical simulations using spectral (Galerkin) and pseudospectral (collocation) methods, spectral element methods, and Lagrangian methods. The discussion also covers large eddy simulations and turbulence modeling.
The numerical behaviour of transport-equation-based transition models, including both iterative and grid convergence, is influenced by the source terms. Transition models contain source terms that are large and highly nonlinear, and can be destabilizing in a strong implicit solver. Linearization strategies with varying levels of coupling are evaluated in conjunction with a source-term time step restriction to determine best-practices for solving the SA-sLM2015smooth local correlation-based transition model in an implicit Newton-Krylov flow solver. Achieving deep iterative convergence facilitates a detailed investigation of the grid convergence of these free-transition simulations, which are evaluated relative to fully-turbulent simulations performed using the Spalart-Allmaras turbulence model. Simulations of the NLF0416 general aviation airfoil, VA-2 supercritical airfoil, and NASA CRM-NLF wing-body geometry are performed over a range of grid levels. The results demonstrate that both a fully-coupled linearization strategy and a source-term time step restriction improve nonlinear convergence as the complexity of the free-transition simulations increases. In general, additional grid resolution is required for free-transition simulations relative to fully-turbulent simulations in order to achieve a similar level of accuracy, with the grid convergence of free-transition simulations sensitive to the streamwise grid spacings in the transition regions.
Turbulent plane Couette flow was numerically simulated at a Reynolds number (U(sub w)h/nu) of 6000, where U(sub w) is the relative wall speed and h is half the channel-height. Unlike in Poiseuille flow, where the mean shear rate changes its sign at the centerline, the sign of mean shear rate in plane Couette flow remains the same across the whole channel. This difference is expected to yield several differences between the two flows, especially in the core region. The most significant and dramatic difference observed was the existence of large-scale structures in the core region of the plane Couette flow. The large eddies are extremely long in the flow direction and fill the entire channel (i.e., their vertical extent is 2h). The large-scale structures have the largest contribution from the wavenumber (k(sub x)h,k(sub z)h) = (0, plus or minus 1.5), corresponding to a wavelength lambda(sub z)/h is approximately equal to 4. The secondary motion associated with the k(sub x)h = 0 mode consists of the large-scale vortices. The large eddies contribute about 30 percent of turbulent kinetic energy.
In this paper, we perform a large eddy simulation (LES) of turbulent flow in a channel containing a two-dimensional obstacle on one wall using a dynamic subgrid-scale model (DSGSM) at Re = 3210, based on bulk velocity above the obstacle and obstacle height; the wall layers are fully resolved. The low Re enables us to perform a DNS (Case 1) against which to validate the LES results. The LES with the DSGSM is designated Case 2. In addition, an LES with the conventional fixed model constant (Case 3) is conducted to allow identification of improvements due to the DSGSM. We also include LES at Re = 82,000 (Case 4) using conventional Smagorinsky subgrid-scale model and a wall-layer model. The results will be compared with the experiment of Dimaczek et al.
The capabilities of large eddy simulation in the prediction and analyses of wall-bounded turbulent shear flows are demonstrated. The dynamical equations for large scale field motions are derived. The computational grid network is described and its relation to the observed physical length scales in the flow are discussed. Some aspects of the mechanics and structure of the flow are examined both in the vicinity of the wall and in regions away from the wall. An attempt is made to correlate numerical results with laboratory observations. Other significant observations and conclusions are presented.
By using Fourier transforms for evaluating spatial derivatives, the accuracy of the large-eddy simulation of homogeneous isotropic turbulence is improved. Especially, the treatment of certain terms that arise in filtering the equations is considerably improved in both speed and accuracy. Use of vorticity as the principal variable is shown to be a viable and potentially useful alternative to the primitive variables. A method of deriving conservation properties of numerical schemes is given which is much simpler than previous methods and is widely applicable. The methods are applied to the computation of homogeneous isotropic turbulence, and it is found that the subgrid scale model is improved by using finite differences in place of 'exact' derivatives.
A high-order accurate finite-difference approach is presented for calculating incompressible turbulent flow. The methods used include a kinetic energy conserving central difference scheme and an upwind difference scheme. The methods are evaluated in test cases for the evolution of small-amplitude disturbances and fully developed turbulent channel flow. It is suggested that the finite-difference approach can be applied to complex geometries more easilty than highly accurate spectral methods. It is concluded that the upwind scheme is a good candidate for direct simulations of turbulent flows over complex geometries.
No abstract available
A code developed for simulating high Reynolds number transonic flow fields of arbitrary configuration is described. This code, in conjunction with laboratory experiments, is used to devise and test turbulence transport models which may be suitable in the prediction of such flow fields, with particular emphasis on regions of flow separation. The solutions describe the flow field, including both the shock-induced and trailing-edge separation regions, in sufficient detail to provide the profile and friction drag.
Understanding and modeling of turbulent combustion are key-problems in the computation of numerous practical systems. Because of the lack of analytical theories in this field and of the difficulty of performing precise experiments, direct simulation appears to be one of the most attractive tools to use in addressing this problem. The present work can be split into two parts: (1) Development and validation of a direct simulation method for turbulent combustion; (2) Applications of the method to premixed turbulent combustion problems. The goal of part 1 is to define and to test a numerical method for direct simulation of reacting flows. A high level of confidence should be attached to direct simulation results, and this can only be achieved through extensive validation tests. In part 2, direct simulation is used to address some of the many critical problems related to turbulent combustion. At the present time, I have limited this work to premixed combustion and considered only four basic issues: (1) The effect of pressure waves on flame propagation; (2) The interaction between flame fronts and vortices; (3) The influence of curvature on premixed flame fronts; and (4) The validation of flamelet models for premixed turbulent combustion.
During the last years the simulation of compressible viscous flows has received much attention. While the numerical methods were improved drastically, a satisfactory modeling of the Reynolds stresses is still missing. In this paper, after a short description of the numerical procedure used for solving the Reynolds equations, experiments with a promising simple turbulence model are discussed.
Simulation of turbulence as it relates to the flight training environment is discussed in general terms. Simulators that put random white noise into the system and simulators that put random motion into the equations of motion are discussed, as are simulators that incorporate pitch and roll moments into vertical turbulence. Wind shear models and simplified models of microburst phenomena are covered.
This paper presents the numerical simulations of confined three dimensional coaxial water jets. The objectives are to validate the newly proposed nonlinear turbulence models of momentum and scalar transport, and to evaluate the newly introduced scalar APDF and DWFDF equation along with its Eulerian implementation in the National Combustion Code (NCC). Simulations conducted include the steady RANS, the unsteady RANS (URANS), and the time-filtered Navier-Stokes (TFNS) with and without invoking the APDF or DWFDF equation. When the APDF or DWFDF equation is invoked, the simulations are of a hybrid nature, i.e., the transport equations of energy and species are replaced by the APDF or DWFDF equation. Results of simulations are compared with the available experimental data. Some positive impacts of the nonlinear turbulence models and the Eulerian scalar APDF and DWFDF approach are observed.
A distributed algorithm for a high-order-accurate finite-difference approach to the direct numerical simulation (DNS) of transition and turbulence in compressible flows is described. This work has two major objectives. The first objective is to demonstrate that parallel and distributed-memory machines can be successfully and efficiently used to solve computationally intensive and input/output intensive algorithms of the DNS class. The second objective is to show that the computational complexity involved in solving the tridiagonal systems inherent in the DNS algorithm can be reduced by algorithm innovations that obviate the need to use a parallelized tridiagonal solver.