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 research broadly concerned the symbolic computation, mixed numeric-symbolic computation, and data base computation of trajectories of dynamical systems, especially control systems. It was determined that trees can be used to compute symbolically series which approximate solutions to differential equations.
This final report comprises the formal semi-annual status reports for this grant for the periods June 30-December 31, 1993, January 1-June 30, 1994, and June 1-December 31, 1994. The research supported by this grant is broadly concerned with the symbolic computation, mixed numeric-symbolic computation, and database computation of trajectories of dynamical systems, especially control systems. A review of work during the report period covers: trajectories and approximating series, the Cayley algebra of trees, actions of differential operators, geometrically stable integration algorithms, hybrid systems, trajectory stores, PTool, and other activities. A list of publications written during the report period is attached.
Numerical simulations are used to study laminar breakdown characteristics associated with stationary crossflow instability in the boundary-layer flow over a subsonic swept-wing configuration. Previous work involving the linear and nonlinear development of individual, fundamental modes of secondary instability waves is extended by considering the role of more complex, yet controlled, spectra of the secondary instability modes. Direct numerical simulations target a mixed mode transition scenario involving the simultaneous presence of Y and Z modes of secondary instability. For the initial amplitudes investigated in this paper, the Y modes are found to play an insignificant role during the onset of transition, in spite of achieving rather large, O(5%), amplitudes of RMS velocity fluctuation prior to transition. Analysis of the numerical simulations shows that this rather surprising finding can be attributed to the fact that the Y modes are concentrated near the top of the crossflow vortex and exert relatively small influence on the Z modes that reside closer to the surface and can lead to transition via nonlinear spreading that does not involve interactions with the Y mode. Finally, secondary instability calculations reveal that subharmonic modes of secondary instability have substantially lower growth rates than those of the fundamental modes, and hence, are less likely to play an important role during the breakdown process involving complex initial spectra.
The numerical simulation of incompressible spatially-developing shear flows poses a special challenge to computational fluid dynamicists. The Navier-Stokes equations are elliptic and boundary equations need to be specified at the inflow and outflow boundaries in order to compute the fluid properties within the region of interest. It is, however, difficult to choose inflow and outflow conditions corresponding to a given experimental situation. Furthermore the effects that changes in the boundary conditions or in the size of the computational domain may induce on the global dynamics of the flow are presently unknown. These issues are examined in light of recent developments in hydrodynamic stability theory. The particular flow considered is the spatial mixing layer but it was expected that similar phenomena were bound to occur in other cases such as channel flow, the boundary layer, etc. A short summary of local/global and absolute/convective instability concepts is given. The results of numerical simulations are presented which strongly suggest that global resonances may be triggered in domains of finite streamwise extent although the evolution of the perturbation vorticity field is everywhere locally convective. A relationship between finite domains and pressure sources which might help in devising a scheme to eliminate these difficulties is discussed.
In order to arrive at physical models that can adequately describe supersonic combustion, and develop accurate and efficient numerical techniques for the solution of such models' governing equations, a computer program has been developed for the study of reacting flows which considers the multicomponent diffusion and convection of important chemical species, as well as their finite state reaction and the interaction of the fluid mechanics and the chemistry that occurred. The code employs a hybrid Chebyshev pseudospectral technique for integration of the models' resulting governing equations; the program is here used to study a spatially developing and reacting mixing layer.
A review is presented of some recently developed numerical methods for the solution of nonlinear equations of mixed type. The methods considered use finite difference approximations to the differential equation. Central difference formulas are employed in the subsonic zone and upwind difference formulas are used in the supersonic zone. The relaxation method for the small disturbance equation is discussed and a description is given of difference schemes for the potential flow equation in quasi-linear form. Attention is also given to difference schemes for the potential flow equation in conservation form, the analysis of relaxation schemes by the time dependent analogy, the accelerated iterative method, and three-dimensional calculations.
Several direct numerical simulation realizations of supercritical, three-dimensional, temporal mixing layers are used to investigate aspects of interest to turbulent combustion.
The injection of fully-developed turbulent heated air from a tube into a cooler turbulent duct flow is examined, as an analogy to film cooled turbine blades. A LES numerical model is developed and applied in which tube and duct turbulence inflow effects are emulated using a divergence-free synthetic eddy method (SEM). For direct comparison, a hot-wire experiment is conducted within the ERB test cell SW-6 at NASA Glenn Research Center. Results related to velocity, temperature, and heat flux are obtained numerically and experimentally for a blowing ratio of 1.2, involving a 36 K temperature difference. The relative effect on the solutions of tube and duct inflow turbulence is systematically evaluated. The impact of inherent low-pass filtering of temperature measurements and probe wire offset on the experimental results are addressed.
Two-dimensional numerical simulations of a temporally evolving mixing layer with an exothermic infinitely fast diffusion flame between two unmixed reactants have been carried out in the limit of zero Mach number to study the effect of the heat release on the early stages of the evolution of the flow. Attention has been directed to relatively large values of the oxidizer-to-fuel mass stoichiometric ratio typical of hydrocarbon flames, and initial vorticity distributions thicker than the temperature and species distributions have been chosen to mimic the situation at the outlet of a jet. The results show that, during the stages of the evolution covered by the present simulations, enhancement of combustion occurs by local stretching of the flame without much augmentation of its area. The rate of product generation depends strongly on the initial conditions, which suggests the possibility of controlling the combustion by acting on the flow. Rollup and vortex amalgamation still occur in these reacting flows but are very much affected by the production of new vorticity by baroclinic torques. These torques lead to counter rotating vortex pairs around the flame and, more importantly, in thin layers of light fluid that leave the vicinity of the flame when the Kelvin-Helmholtz instability begins to develop. Propelled by the vortex pairs, these layers wind around, split on reaching high pressure regions, and originate new vortex pairs in a process that ends up building large-scale vortices with a vorticity distribution more complex than for a constant density fluid.
This report describes a study of databases produced by direct numerical simulation of mixing layers developing between opposing flows of two fluids under supercritical conditions, the purpose of the study being to elucidate chemical-species-specific aspects of turbulence, with emphasis on helicity. The simulations were performed for two different fluid pairs -- O2/H2 and C7H16/N2 -- at similar values of reduced pressure.
A preliminary investigation of the geometry of flow patterns in numerically simulated compressible and incompressible mixing layers was carried out using 3-D critical point methodology. Motions characterized by high rates of kinetic energy dissipation and/or high enstrophy were of particular interest. In the approach the partial derivatives of the velocity field are determined at every point in the flow. These are used to construct the invariants of the velocity gradient tensor and the rate-of-strain tensor (P, Q, R, and P(sub s), Q(sub s), R(sub s) respectively). For incompressible flow the first invariant is zero. For the conditions of the compressible simulation, the first invariant is found to be everywhere small, relative to the second and third invariants, and so in both cases the local topology at a point is mainly determined by the second and third invariants. The data at every grid point is used to construct scatter plots of Q versus R and Q(sub s) versus R(sub s). Most points map to a cluster near the origin in Q-R space. However, fine scale motions, that is motions which are characterized by velocity derivatives which scale with the square root of R(sub delta), tend to map to regions which lie far from the origin. Definite trends are observed for motions characterized by high enstrophy and/or high dissipation. The observed trends suggest that, for these motions, the second and third invariants of the velocity gradient and rate-of-strain tensors are strongly correlated. The second and third invariants of the rate-of-strain tensor are related by K(-Q(sub s))(exp 3/2), which is consistent with the above scaling of velocity derivatives. The quantity K appears to depend on Reynolds number with an upper limit K = 2(the square root of 3)/9 corresponding to locally axisymmetric flow. For both the compressible and incompressible mixing layer, regions corresponding to high rates of dissipation are found to be characterized by comparable magnitudes of R(sub ij)R(sub ij) and S(sub ij)S(sub ij). For the incompressible mixing layer, regions characterized by the highest values of enstrophy are found to have relatively low strain rates.
The results of direct numerical simulations of chemically reacting mixing layers are presented. The reaction considered is a binary, irreversible reaction with no heat release, so that only the effect of the turbulence on the chemical reaction is investigated. The simulation results are shown to be consistent with similarity theory, and are found to be in approximate agreement with laboratory data, even though there are no adjustable constants in the method.
A hybrid numerical scheme, combined with a scalar convection scheme, is used to compute turbulent mixing layers. The hybrid scheme is used to solve the vorticity equation by dividing the domain into two regions, applying a Lagrangian method in one region and an Eulerian method in the other. The scalar convection scheme is used to track an interface which separates a dyed fluid from a dye-free fluid, thereby distinguishing the fluid that starts on the high speed side from the fluid that starts on the low speed side. A two-dimensional mixing layer has been simulated using this new method; the roll up of a vortex sheet into structures that interact and pair is observed. The growth of the momentum thickness and the development of the mean turbulence statistics are found to be consistent with experimental observations. Forcing of the layer at moderate and low frequencies shows a dramatic effect on the growth of the momentum thickness and the entrainment ratio. Low frequency forcing enhances the growth of the layer; moderate frequency forcing inhibits the growth.
Direct numerical simulations are performed of a confined three-dimentional, temporally developing, initially isothermal gas mixing layer with one stream laden with as many as 7.3 x 10***sup5*** evaporating hydrocarbon droplets, at moderate gas temperature and subsonic Mach number.
A parabolized, Navier-Stokes code, SHIP3D, is used to numerically investigate the mixing between air injection and hydrogen injection from a swept ramp injector configuration into either a mainstream low-enthalpy flow or a hypervelocity test flow. The mixing comparisons between air and hydrogen injection reveal the importance of matching injectant-to-mainstream mass flow ratios. In flows with the same injectant-to-mainstream dynamic pressure ratio, the mixing definition was altered for the air injection cases. Comparisons of the computed results indicate that the air injection cases overestimate the mixing performance associated with hydrogen injection simulation. A lifting length parameter, to account for the time a fluid particle transverses through the mixing region, is defined and used to establish a connection of injectant mixing in hypervelocity flows, based on nonreactive, low-enthalpy flows.
This work is a continuing effort to advance our understanding of the effects of polymer additives on the structures of the mixing layer. In anticipation of full nonlinear simulations of the non-Newtonian mixing layer, we examined in a first stage the linear stability of the non-Newtonian mixing layer. The results of this study show that, for a fluid described by the Oldroyd-B model, viscoelasticity reduces the instability of the inviscid mixing layer in a special limit where the ratio (We/Re) is of order 1 where We is the Weissenberg number, a measure of the elasticity of the flow, and Re is the Reynolds number. In the present study, we pursue this project with numerical simulations of the non-Newtonian mixing layer. Our primary objective is to determine the effects of viscoelasticity on the roll-up structure. We also examine the origin of the numerical instabilities usually encountered in the simulations of non-Newtonian fluids.
Several direct numerical simulations were performed and analyzed to study various aspects of the early development of mixing layers. Included are the phase jitter of the large-scale eddies, which was studied using a 2-D spatially-evolving mixing layer simulation; the response of a time developing mixing layer to various spanwise disturbances; and the sound radiation from a 2-D compressible time developing mixing layer.
A current research effort is underway at the NASA Langley Research Center to achieve a detailed understanding of important phenomena present when a supersonic flow undergoes chemical reaction. A computer program has been developed to study the details of such flows. The program has been constructed to consider the multicomponent diffusion and convection of important species, the finite-rate reaction of these species, and the resulting interaction between the fluid mechanics and chemistry. Code results from the analysis of a spatially developing and reacting mixing layer are presented, and conclusions are drawn regarding the structure of the evolving layer and its associated flame.