Christopher L. Rumsey

Prediction of high lift: review of present CFD capability

Abstract A survey is conducted of CFD methods applied to the computation of high-lift multi-element configurations over the last 10–15 years. Both 2-D and 3-D configurations are covered. The review is organized by configuration, in an effort to glean useful insights with respect to particular successes or failings of CFD methods as a whole. In general, for both 2-D and 3-D flows, if certain guidelines regarding grid, transition, and turbulence model are followed, then surface pressures, skin friction, lift, and drag can be predicted with reasonably good accuracy at angles of attack below stall. Velocity profiles can generally be predicted in 2-D flow fields, with the exception of the slat wake, which tends to be predicted too deep by most CFD codes for a range of different configurations. CFD codes can usually predict trends due to Reynolds number in 2-D, but they are inconsistent in the prediction of trends due to configuration changes. On the whole, 2-D CFD is unreliable for predicting stall (maximum lift and the angle of attack at which it occurs); in most cases, maximum lift is overpredicted, but for some configurations the opposite occurs. However, there is some evidence that stall misprediction of nominally 2-D experiments may be caused by 3-D effects, which are obviously not modeled by 2-D CFD. In general, 3-D computations are also inconsistent with respect to computing stall, but there have been fewer of these applications to date. The paper concludes with a list of challenges that confront CFD at the start of the next decade, which should witness a dramatic increase in the number of CFD applications for 3-D high-lift configurations.

Effective Inflow Conditions for Turbulence Models in Aerodynamic Calculations

The selection of inflow values for one- and two-equation turbulence models at boundaries far upstream of an aircraft is considered. Inflow values are distinguished from the ambient values near the body, which may be much smaller because the long approach can allow a deep decay. Ambient values should be selected first, and inflow values that will lead to them after the decay should be second; this is not always possible, especially for the time scale. The decay of turbulence in two-equation models during the approach to the aircraft is shown; in computational fluid dynamics practice, the time scale has often been set too short for this decay to be calculated accurately on typical grids. A simple remedy for both issues is to arrest decay below the chosen ambient values, either by imposing floor values, or preferably by adding weak source terms. A physical justification for overriding the equations in this manner is proposed. Selecting laminar ambient values is easy if the boundary layers are to be tripped, but it is common to seek ambient values that will cause immediate transition in shear layers. This opens up a wide range of values, and selection criteria are discussed. The turbulent Reynolds number, or ratio of eddy viscosity to laminar viscosity has a huge dynamic range that makes it unwieldy; it has been widely misused, particularly by setting upper limits on it. The value of the complete turbulent kinetic energy in a wind tunnel or the atmosphere is also dubious as an input to the model, because its spectrum contains length scales irrelevant to the turbulence in the boundary and shear layers. Concretely, the ambient eddy viscosity must be small enough to preserve potential cores in small geometry features such as flap gaps. The ambient-frequency scale should also be small enough, compared with shear rates in the boundary layer. Specific ranges of values are recommended and demonstrated for airfoil flows.

Summary of the 2004 Computational Fluid Dynamics Validation Workshop on Synthetic Jets

A computational-fluid-dynamics (CFD) validation workshop for synthetic jets and turbulent separation control (CFDVAL2004) was held in Williamsburg, Virginia, in March 2004. Three cases were investigated: a synthetic jet into quiescent air, a synthetic jet into a turbulent boundary-layer crossflow, and the flow over a hump model with no-flow-control, steady suction, and oscillatory control. This is a summary of the CFD results from the workshop. Although some detailed results are shown, the CFD state of the art for predicting these types of flows is mostly evaluated from a general point of view. Overall, for synthetic jets, CFD can only qualitatively predict the flow physics, but there is some uncertainty regarding how to best model the unsteady boundary conditions from the experiment consistently. As a result, there is wide variation among CFD results. For the hump flow, CFD is capable of predicting many of the particulars of this flow, provided that it accounts for tunnel blockage, but it consistently overpredicts the length of the separated region compared to the experimental results.

Turbulence Model Predictions of Strongly Curved Flow in a U-Duct

The ability of three types of turbulence models to accurately predict the effects of curvature on the flow in a U-duct is studied. An explicit algebraic stress model performs slightly better than one- or two-equation linear eddy viscosity models, although it is necessary to fully account for the variation of the production-to-dissipation-rate ratio in the algebraic stress model formulation. In their original formulations, none of these turbulence models fully captures the suppressed turbulence near the convex wall, whereas a full Reynolds stress model does. Some of the underlying assumptions used in the development of algebraic stress models are investigated and compared with the computed flowfield from the full Reynolds stress model. Through this analysis, the assumption of Reynolds stress anisotropy equilibrium used in the algebraic stress model formulation is found to be incorrect in regions of strong curvature. By the accounting for the local variation of the principal axes of the strain rate tensor, the explicit algebraic stress model correctly predicts the suppressed turbulence in the outer part of the boundary layer near the convex wall.

Summary of the 2004 CFD Validation Workshop on Synthetic Jets and Turbulent Separation Control

Summary of the 2004 CFD ValidationWorkshop on Synthetic Jets and TurbulentSeparation Control C. L. Rumsey ∗, T. B. Gatski †, W. L. Sellers III ‡, V. N. Vatsa §, S. A. Viken ¶ NASA Langley Research Center, Hampton, VA 23681-2199, USA Submission to AIAA Journal A CFD validation workshop for synthetic jets and turbulent separation control (CFD-VAL2004) was held in Williamsburg, Virginia in March 2004. Three cases were inves-tigated: synthetic jet into quiescent air, synthetic jet into a turbulent boundary layercrossflow, and flow over a hump model with no-flow-control, steady suction, and oscilla-tory control. This paper is a summary of the CFD results from the workshop. Althoughsome detailed results are shown, mostly a broad viewpoint is taken, and the CFD state-of-the-art for predicting these types of flows is evaluated from a general point of view.Overall, for synthetic jets, CFD can only qualitatively predict the flow physics, but thereis some uncertainty regarding how to best model the unsteady boundary conditions fromthe experiment consistently. As a result, there is wide variation among CFD results. Forthe hump flow, CFD as a whole is capable of predicting many of the particulars of thisflow provided that tunnel blockage is accounted for, but the length of the separated regioncompared to experimental results is consistently overpredicted.

Summary of Data from the Fifth AIAA CFD Drag Prediction Workshop

Results from the 5, AIAA CFD Drag Prediction Workshop are presented. This workshop is focused on force/moment predictions for the NASA Common research wing-body configuration, including a grid refinement study and an optional buffet study. The article presents the summary of data of all participants.

Prediction of nonequilibrium turbulent flows with explicit algebraic stress models

An explicit algebraic stress equation, developed by Gatski and Speziale, is used in the framework of the K-UJ and K-e formulations to predict two-dimensional separated turbulent flows. The nonequilibrium effects are modeled through coefficients that depend nonlinearly on both rotational and irrotational strains. The proposed models were implemented in the Courant-Friedrichs-Lewy three-dimensional Navier-Stokes code. Comparisons with the experimental data are presented, which clearly demonstrate that explicit algebraic stress models improve the response of standard two-equation models to nonequilibrium effects. I. Introduction

Summary of Data from the Fifth Computational Fluid Dynamics Drag Prediction Workshop

Results from the Fifth AIAA Computational Fluid Dynamics Drag Prediction Workshop are presented. As with past workshops, numerical calculations are performed using industry-relevant geometry, methodology, and test cases. This workshop focused on force/moment predictions for the NASA Common Research Model wing-body configuration, including a grid refinement study and an optional buffet study. The grid refinement study used a common grid sequence derived from a multiblock topology structured grid. Six levels of refinement were created, resulting in grids ranging from 0.64×106 to 138×106 hexahedra, a much larger range than is typically seen. The grids were then transformed into structured overset and hexahedral, prismatic, tetrahedral, and hybrid unstructured formats all using the same basic cloud of points. This unique collection of grids was designed to isolate the effects of grid type and solution algorithm by using identical point distributions. This study showed reduced scatter and standard deviation fr...

Apparent transition behavior of widely-used turbulence models

Abstract The Spalart–Allmaras and the Menter k – ω SST turbulence models are shown to have the undesirable characteristic that, for fully turbulent external flow computations, a transition region can occur whose extent varies with grid density. Extremely fine two-dimensional grids over the front portion of an airfoil are used to demonstrate the effect. As the grid density is increased, the laminar region near the nose becomes larger. In the Spalart–Allmaras model this behavior is due to convergence to a laminar-behavior fixed point that occurs in practice when freestream turbulence is below some threshold. It is the result of a feature purposefully added to the original model in conjunction with a special trip function. This degenerate fixed point can also cause non-uniqueness regarding where transition initiates on a given grid. Consistent fully turbulent results can easily be achieved by either using a freestream turbulence level higher than the threshold or by making a simple change to one of the model constants. Near the area where turbulence initiates, the SST model exhibits sensitivity to numerical resolution, but its solutions are unique on a given grid. Inconsistent apparent transition behavior with grid refinement in this case does not stem from the presence of a degenerate fixed point. A nullcline analysis is used to visualize the local behavior of the model.

Towards Verification of Unstructured-Grid Solvers

New methodology for verification of finite-volume computational methods using unstructured grids is presented. The discretization order properties are studied in computational windows, easily constructed within a collection of grids or a single grid. Tests are performed within each window and address a combination of problem-, solution-, and discretization/grid-related features affecting discretization error convergence. The windows can be adjusted to isolate particular elements of the computational scheme, such as the interior discretization, the boundary discretization, or singularities. Studies can use traditional grid-refinement computations within a fixed window or downscaling, a recently-introduced technique in which computations are made within windows contracting toward a focal point of interest. Grids within the windows are constrained to be consistently refined, allowing a meaningful assessment of asymptotic error convergence on unstructured grids. Demonstrations of the method are shown, including a comparative accuracy assessment of commonly-used schemes on general mixed grids and the identification of local accuracy deterioration at boundary intersections. Recommendations to enable attainment of design-order discretization errors for large-scale computational simulations are given.