Detached-Eddy Simulation
Philippe R. Spalart
Boeing Commercial Airplanes, Seattle, Washington 98124; email:

Click here for quick links to Annual Reviews content online, including: • Other articles in this volume • Top cited articles • Top downloaded articles • Our comprehensive search

Annu. Rev. Fluid Mech. 2009.41:181-202. Downloaded from by b-on: Universidade da Beira Interior (UBI) on 09/20/11. For personal use only.

Annu. Rev. Fluid Mech. 2009. 41:181–202 First published online as a Review in Advance on August 4, 2008 The Annual Review of Fluid Mechanics is online at This article’s doi: 10.1146/annurev.fluid.010908.165130 Copyright c 2009 by Annual Reviews. All rights reserved 0066-4189/09/0115-0181$20.00

Key Words
turbulence, separation, boundary layer, modeling

Detached-eddy simulation (DES) was first proposed in 1997 and first used in 1999, so its full history can be surveyed. A DES community has formed, with adepts and critics, as well as new branches. The initial motivation of high– Reynolds number, massively separated flows remains, for which DES is convincingly more capable presently than either unsteady Reynolds-averaged Navier-Stokes (RANS) or large-eddy simulation (LES). This review discusses compelling examples, noting the visual and quantitative success of DES. Its principal weakness is its response to ambiguous grids, in which the wall-parallel grid spacing is of the order of the boundary-layer thickness. In some situations, DES on a given grid is then less accurate than RANS on the same grid or DES on a coarser grid. Partial remedies have been found, yet dealing with thickening boundary layers and shallow separation bubbles is a central challenge. The nonmonotonic response of DES to grid refinement is disturbing to most observers, as is the absence of a theoretical order of accuracy. These issues also affect LES in any nontrivial flow. This review also covers the numerical needs of DES, gridding practices, coupling with different RANS models, derivative uses such as wall modeling in LES, and extensions such as zonal DES and delayed DES.




Figure 1

Annu. Rev. Fluid Mech. 2009.41:181-202. Downloaded from by b-on: Universidade da Beira Interior (UBI) on 09/20/11. For personal use only.

(a) Vorticity isosurfaces colored with pressure over an F-15 jet at a 65◦ angle of attack (Forsythe et al. 2004). Figure courtesy of J. Forsythe. (b) Acoustic-source isosurface around a Ford Ka automobile (es turbo 3.1) (Mendonca et al. 2002). Figure courtesy of F. ¸ Mendonca and Ford Motor Co. ¸

Figure 1 illustrates the nature of detached-eddy simulation (DES). The aircraft geometry is complete (except for detailed surface and propulsion effects); the simulation is at flight Reynolds number; the large-eddy simulation (LES) content (resolved turbulence) in the separated region is rich; and the Reynolds-averaged Navier-Stokes (RANS) function plays a role on the aircraft’s nose. Furthermore, the forces and moments are accurate to within 6% (Forsythe et al. 2004). This approach must still be considered experimental as a prediction method, and the accuracy benefits from the thin edges on the wing; there is no marginal separation to challenge the model. In addition, grid refinement does not indicate grid independence on the smaller components, such as the tail surfaces. The automobile geometry is also complete, a feat of the grid generator and solver rather than of DES (Mendonca et al. 2002). The two regions of the DES are especially well visualized: ¸ steady attached boundary layers and striking LES content around the wheels and the important A pillar and outside mirror. The drag is dependent on the separation line near the end of the roof, and the accuracy of the RANS model matters. At the same time, the LES function is indispensable to predict the aerodynamic noise and in fact the drag. These two studies reflect the broad diffusion of DES.

1.1. Conceptual History
DES was created to address the challenge of high–Reynolds number, massively separated flows, which must be addressed in such fields as aerospace and ground transportation, as well as in atmospheric studies. It combined LES and RANS, spurred by the belief that each alone was powerless to solve the problem at hand (Spalart et al. 1997). This complaint can be revisited presently, assuming a working knowledge of LES and RANS (Rogallo & Moin 1984, Wilcox 1998). The objection to pure LES is simple and centers on computational cost. A pure LES of an airborne or ground vehicle would use well over 1011 grid points and close to 107 time steps, which is estimated to be possible in approximately 2045 (Spalart 2000). The boundary layer dominates this expense, which is necessary even if investigators solve the problem of wall modeling in LES. Regardless, the resolution needs in the outer region of the boundary layer are firm, with at the least 20 points per thickness δ in each direction. No unforeseen breakthrough has occurred in
182 Spalart

the model follows its own natural RANS history. as expected. although there were none for DES’s first application. This motivated a relatively deep change in its formulation with shielded DES and delayed DES (Menter & Kuntz 2002. and regions of massive separation are treated with LES. A second motivation for DES over RANS appears in situations that. The mixing length then can be limited by two constraints: the wall distance and the grid spacing. Spalart et al. 2006) as the DES length-scale limiter now depends on the solution. Fluid Mech. which does not imply that any geometry has allowed grid convergence. Types of Simulation for Massive Separation Simulation for massive separation is an important field in which the differences in approach are deep and deserve a detailed discussion. The computing cost of the RANS region is easily manageable. The original reasons to believe in DES can also be revisited. For personal use only. whether behind a sphere or past buildings. users quickly encountered grid spacings that disturbed the RANS model (see Section 3. of order 4/3 in the Kolmogorov situation. away from walls. and the principal difficulty may be to generate grids that cover all of the boundary layer well in terms of thickness.41:181-202. Refinement reduces the eddy viscosity. Rev. vibration and noise). may be problematic unless the separation is abrupt. The objection to pure RANS is not as limpid because it arises from a negative assessment of models and the relentless attempts to build into them first-principle content and rational ideas. The gray area drew complaints as soon as 2000 in an application to an overexpanded nozzle. When neither constraint is felt. RANS models can be adjusted to predict boundary layers and their separation well. which we refer to as DES97 here. Surprisingly. as almost all the models used in DES date back to • Detached-Eddy Simulation 183 . A working definition is that the boundary layer is treated by RANS. Downloaded from www.g. RANS development has been static. but not large separation regions. even if RANS were accurate. was defined as “a three-dimensional unsteady numerical solution using a single turbulence model. often fixed by the geometry.annualreviews. A single model. except in homogeneous turbulence. and so on. However. the Spalart-Allmaras model was used. since 1997. 2009. perfection cannot be reached. this is the case for free shear flows when they have a grid too coarse to use LES for that particular layer. 2000a).1). The capability of LES in free shear flows is not in question. Figure 2 illustrates possible contenders for the www. which functions as a subgrid-scale model in regions where the grid density is fine enough for a large-eddy simulation. 1999). In this view. the space between these areas. and grid refinement brings no improvement beyond the accuracy barrier of the model. and RANS is simply necessary for the large extent of thin boundary layers (the thicker parts are discussed below). rather than on the grid only. which was to a thin airfoil.LES since 1997. and a plausible view of LES is that the eddy viscosity is an error. There is only consensus that finer grids improve the physics and that grid refinement. in 1999 (Shur et al. provides the desired function in both the RANS and LES limits. 1. Reducing also reduces numerical errors because the cutoff is further down the spectrum. but DES now draws on several other models (Strelets 2001) (see Section 4. researchers have tended to shift their effort from RANS to LES and hybrid methods. would need unsteady information for engineering purposes (e. known as the gray area. with at best a factor of 2 in each direction. Initially. Nonetheless. has not created bad surprises. vehicles. In a natural DES. with a RANS origin but sensitized to grid spacing via a DES limiter.2). these methods are aimed at better fulfilling the original mission of DES. The original version of DES. and as a Reynolds-averaged model in regions where it is not” (Travin et al. Few groups have conducted grid by b-on: Universidade da Beira Interior (UBI) on 09/20/11.annualreviews. Observers are hopeful for a new perspective that could erase this objection soon. and velocity scales like 1/3 .. Annu. with RANS function extending to the separation line.2. in cavities.

41:181-202. Downloaded from by b-on: Universidade da Beira Interior (UBI) on 09/20/11.Annu.annualreviews. 2009. Rev. Fluid Mech. For personal use only. 184 Spalart f c e a d b .

even if 3D. Cd = 1. The three-dimensionality is much coarser than in DES and does not become finer on a finer grid. there are reasons to be familiar with URANS. URANS largely suppresses three-dimensionality. in a complex geometry. coarse grid. In spite of its failings. Besides the cylinder. simulation of flow past a circular cylinder and similar cases. URANS near the smaller components. in unsteady mode and with periodic spanwise conditions..24. Recent findings have revealed that under fairly general conditions. which it does in DES. Nishino et al. a point to which we return. Downloaded from www. Fluid Mech.25.16. 2. ←−−−−−−−−−−−−−−−−−−−−−−−−−−−−−−−−−−− −−−−−−−−−−−−−−−−−−−−−−−−−−−−−−−−−−−− Figure 2 Vorticity isosurfaces by a circular cylinder: Re D = 5 × 104 . Figure courtesy of A.g.Annu.” making 3D URANS with standard models a weak contender for this simulation. The situation is not as simple as it appeared in 1997. Second.73. The boundary layers being laminar. Figure 2 illustrates the classic steady RANS (an unstable solution) and 2D URANS and includes the newer 3D URANS.annualreviews. but not completely. URANS is vulnerable to the criticism that its partial differential equations are known. For personal use only. More effective RANS models could be devised. Figure 2 also vividly illustrates the response of DES to grid refinement in its LES region. 2003). is less accurate than DES and (when applicable) LES. some researchers do believe in its capabilities and would dispute our conclusions from Figure 2. www. and it had been found that drag and lift fluctuations were overpredicted by URANS. even with the knowledge that the geometric detail ideally would be granted LES content on its length scales and timescales through a finer grid and a shorter time step. Travin. effectively. sometimes the DES grid and time step only allow.28. Cd = 1. which strongly supports the idea that URANS. Cd = 0. 2005a). It was then considered obvious that unsteady RANS (URANS) solutions suppressed three-dimensionality over two-dimensional (2D) geometries. (e) SA DES. Shur et al. these simulations in fact sustain threedimensionality and are more accurate than 2D URANS (Shur et al. fine grid. Examples include the wiper blade on a car and the active-flow-control slot on an aircraft (Spalart et al.41:181-202. Figure 2 does not reflect DES’s value in treating turbulent boundary layers in a manner LES cannot. in the absence of a spectral gap. Experimental drag coefficient Cd = 1. 2009. but the (Reynolds?) averaging it actually represents is not known. The term URANS here means running an unmodified (grid-insensitive) transportequation turbulence • Detached-Eddy Simulation 185 . ( f ) SST DES.26. Rev. a backward-facing step) and is a valuable feature. (b) SST 2D unsteady RANS. Cd = 1.78. (a) Shear-stress transport (SST) turbulence model steady Reynolds-averaged Navier-Stokes (RANS). (c) SST 3D unsteady RANS. First. Cd = 1. It is desirable for hybrid methods to handle such situations gracefully. A somewhat similar challenge can be directed at DES. particularly if separation occurs). Cd = 1. Still. (2005a) also cite and demonstrate “a troublesome sensitivity to the spanwise period and to the turbulence model. it confirms that DES solutions with different base RANS models are not sensitive to model choice in the LES region (as opposed to the RANS region. but subsequent figures do.annualreviews. fine grid. There is no evidence that the lateral length scales in the 3D URANS field are physical. (2008) present a thorough URANS and DES study of a cylinder near a wall. Finally. STRENGTHS This section aims to verify the soundness of DES quantitatively in the important respects of comparison with experiment and response to grid refinement.15–1. although the shedding frequency was accurate. (d ) Spalart-Allmaras (SA) detached-eddy simulation (DES). these authors treated an airfoil and a rounded by b-on: Universidade da Beira Interior (UBI) on 09/20/11. This has been verified quantitatively in many cases (e. laminar separation.

and the pressure distributions are extremely favorable both when compared with experiment and when comparing baseline and fine grids. The study featured in Figure by b-on: Universidade da Beira Interior (UBI) on 09/20/11. approximately 0. Rev.2 0. (2002) simulated the flow past a sphere with approximately 600. Simple Geometries Above it was mentioned that grid refinement on the jet aircraft had nontrivial effects on the smaller components. 2009.3 G3 (6. A scale-adaptive simulation also produced resolved turbulence in this flow (Egorov & Menter 2008).25 0.1. but it is far superior to letting an untrained subgrid-scale model handle the boundary layer. The simpler geometry helped.000 points in the baseline grid and controlled the model in the boundary layer so that it produced laminar separation at a diameter of Re = 105 and turbulent separation at Re = 1. 2.5 M cells) 0. It is tempting to extend this approach to golf balls.annualreviews.a 0. With this approach. 2000a).56 × 106 (Morton 2003).’s (2000a) tripless approach is needed. Quite a few recent cylinder computational fluid dynamics (CFD) studies even failed to select laminar separation at subcritical Reynolds numbers. compared with 0. effectively in RANS mode.41:181-202. Travin et al. DES predicts a drag coefficient of 0.2 M cells) 0 0 0. shown visually in Figure 3a and quantitatively in Figure 3b.084 and experiments give 0. X/c = 0. and location.5 G4 (10. Figure courtesy of S. The latter is somewhat simplistic because in the real flow.1 Annu. Downloaded from www.6 M cells) G2 (2.41. on Morton’s (2003) delta wings. at the higher Reynolds number.4 Resolved TKE (k/U2) 0. G1 (1.00 X/c Figure 3 (a) Flow visualizations and (b) resolved turbulent kinetic energy (TKE) for a sharp-edged delta wing at a 27◦ angle of attack.65 ± 0.12. chord Reynolds number 1. DES gives 0. (2008) adopted it successfully. The drag crisis caused by dimples can be captured in a gross sense. For personal use only.5.05 (Mitchell et al. Morton.1 × 106 . The results are rewarding. Grid-refinement effects were more predictable. transition and separation are not segregated (Travin et al. as it suggests near-grid convergence of the resolved turbulent kinetic energy to a level that agrees with experiment both for energy level. Finer grids introduce vortex shedding at the trailing edge and much finer structures in the vortex.50 0. Fluid Mech. simply by imposing turbulent separation with a smooth geometry.40–0. At the lower Reynolds number. and Nishino et al. Figure 3b is especially favorable.7 M cells) 0. however. but the phenomenon of vortex breakdown is a subtle one. However.75 1. no 186 Spalart . Constantinescu et al. the prediction of a drag crisis is striking. The front half of the vortex is also quite different: The helical striations switch direction from a coarse to a fine grid.b also reflects the quantitative success of DES.5 b Experimental peak approximately 0.51 in experiments. 2000).

This is part of a general challenge stemming from the range of scales in fluid mechanics. (2008). Rev.1 × 106 Annu.50 –0. Color gradations denote DES conducted by Mockett et al.50 1 2 x/D 0 30 60 90 120 150 180 θ Figure 4 Simple bluff bodies.annualreviews. if they become dependent on geometry. Squires. LES addresses the Kolmogorov viscous scale limitation. Fluid Mech. (c) Phase-averaged vorticity contours for a cylinder. be it on the shape of a wiper blade or that of a dimple. For personal use only.and turbulentseparation cases and a surprise-free grid-refinement study. RANS model could reproduce the dimple effect accurately. which added confidence after Shur et al.’s (1999) initial thin-airfoil work.00 c 0. However. b 1. These eddies are numerous and fairly • Detached-Eddy Simulation 187 . LES treatment of their scales becomes necessary for high accuracy so that many problems.a Re = 105 Re = 1. and wall modeling addresses the similar viscous-sublayer scale.00 –1. In its RANS mode.5 Re = 105 0 –1. Open circles and diamonds denote experiments.41:181-202.annualreviews. DES in addition addresses the boundary-layer eddies of all sizes.’s (2000a) circular-cylinder study similarly included laminar. Figure 4c compares DES and experiment behind a www. 2009.1 × 106 . Downloaded from www.00 0. and this will require direct numerical simulation (DNS).5 y/D 0 Cp –0. Compared with DNS. and the solid line denotes experiments by the same authors.1 × 106 0. Re = 105 and 1. Travin et al. at least of the dimple flow proper. in particular active flow control.50 Re = by b-on: Universidade da Beira Interior (UBI) on 09/20/11. (a) Flow visualizations and (b) pressure distributions for a sphere. simply exceed even current grids in excess of 108 points. Panels a and b courtesy of K. whereas the dotted and dashed lines denote detached-eddy simulation (DES) on two grids.

2. DES did much better than RANS with simple models. the visualization with density gradient in Figure 5a vividly reveals much fine-scale turbulence and sound. 2006). the focus region is much larger. We note above the industrial importance of the turbulence adjacent to the driver’s window (Figure 1b). The agreement on the phase-averaged flow pattern is excellent. and afterbodies. Downloaded from www. which causes wall-pressure fluctuations. Langtry & Spalart 2007. vibrating cylinders with strakes (Constantinides & Oakley 2006). ¸ ground vehicles (Kapadia et al. high-lift devices (Cummings et al. the DES visualizations are close to those shown in Figure 2e. the agreement on the shape of the spectrum. which is positive. Grid adaptation in DES and LES is a future challenge. Dassault. 2006). Peng & Haase (2008) report on many promising applications at various stages of maturity: wing high-lift systems. invariably quite narrow (in contrast. 2004). Fluid Mech. 2004. Another promising direction is taken by Mockett et al. aircraft in a spin (Forsythe et al. 2003.annualreviews. Mendonca et al. This case is more onerous because the turbulence needs to be carried all the way to the airfoil. 2004). combustors. Applications DES has been applied often with good results to cavities over a range of Mach numbers (Allen et al. is the dominant noise mechanism. space launchers (Deck & Thorigny 2007. Shieh & Morris 2001). however. (2008) studied the flow in the slat cove of an airfoil in landing configuration. Spalart et al. At low Mach numbers. Second. 2002). appropriately. (2007) reveal a vigorous line of work in another aircraft company. Such studies will contribute both to interior noise in vehicles and aircraft and to community noise (airframe noise to the airline industry). no adjustment was needed for the spectra inside the turbulence region).org by b-on: Universidade da Beira Interior (UBI) on 09/20/11. Greschner et al. 10 diameters downstream. may be of help. 2006). active flow control by suction/blowing (Krishnan et al. (2008) and Greschner et al. Publications aimed at educating users and code writers have. Roy et al. 2003). 2003. jet-fighter tail buffet (Morton et al. Squires 2004). and wing-wall junctions (Fu et al.2. Maddox et al. An adjustment was made in the vertical direction: In 2D geometries. focused on grid generation (Spalart 2001) and on thorough testing of the codes (Bunge et al. 188 Spalart . 2003. (2006) in jets and by Ziefle & Kleiser (2008) in a supersonic channel with hills.circular cylinder (Mockett & Thiele 2007). Hamed et al.41:181-202. 2008). Flow visualizations resemble those in Figure 2. there is an unsolved problem when comparing an experiment of finite length (with some end conditions) to a simulation with periodic boundary conditions. the simulation is simultaneously free enough of numerical dissipation to welcome LES content and robust enough to capture shocks. 2009. Spalart & Squires 2004. Various Ffowcs-Williams-Hawkings surfaces are used to extract far-field noise. (2008) provide sound predictions for the flow past a cylinder. over five octaves. The terminology Euler region. Forsythe et al. Chalot et al. placed ahead of an airfoil so that its turbulent wake impinges on it (see Figure 5b). These Annu. Once this correction is accepted. and departure region. without as fine a level of resolution. 2006). This result has also been achieved by Shur et al. f. focus region. introduced by Spalart. which may not be expected to capture curvature effects. Figure 5c compares the sound spectrum with experiment. Slimon (2003) obtained positive results with (zonal) DES in a turn-around duct.4). helicopters. Sreenivas et al. a simplified landing-gear truck (Hedges et al. For personal use only. Mockett et al. 2004. 2007. the LES-content development in the mixing layer is nearly immediate. 2007) is of interest for two reasons. although it may be excessively 2D (see Section 3. this impingement. 2007). 2005. First. 2004. air inlets (Trapier et al. 2002). buildings (Wilson et al. cavitation in jets (Edge et al. Actual sound predictions are not included. Figure 6 (Chauvet et al. (2008): aerodynamic noise. Rev. RANS region. is quite amazing.

Greschner. Δ 100 St = f × D/u∞ www. Downloaded from www. power spectral density.a | p'| 0. (c) Far-field spectrum. Panel a courtesy of C. Panels b and c courtesy of B. (b) Vorticity isosurfaces for a rod-cylinder case. b Y X Z c PSD (dB) 100 80 60 40 20 0 θ = 90° Experiment DES + FWH 10–1 Figure 5 Complex bluff bodies.01 0. (a) Schlieren picture near a • Detached-Eddy Simulation 189 . PSD.001 by b-on: Universidade da Beira Interior (UBI) on 09/20/11. Mockett.41:181-202.1 1 10 Annu. Fluid Mech.annualreviews. For personal use only. 2009. Rev.annualreviews.

especially in view of the wide variety of anisotropies possible for the grid cell and time step. but the strength of DES (and WMLES) is precisely that the split is different in different parts of the same solution. Clear statements are much more difficult to make. 2009. The new difficulty beyond those in LES is that. in the gray area. The criticism of URANS mentioned above (namely that the approximate PDE that is solved is known. In that situation. Conceptual Issues The need to predict turbulence numerically is far-reaching.a y/D 1 b1 0 0 –1 1 2 3 –1 0 1 2 3 X/D Figure 6 X/D (a) Experimental schlieren (view through flow) and (b) numerical schlieren (contours in center plane) for a supersonic jet. simply put. A filter can be linked to the grid cell and to the integration implied by the CFD solver. The energy split can be adjusted in different regions. This is fine in simple flows. but the exact PDE it is meant to approximate is not) does not truly apply to DES. A similar problem is present even in simple LES. one can arbitrarily lessen the influence of the Smagorinsky constant and similar constants with grid refinement. but a convincing calibration is simply out of reach: There are far too few degrees of freedom (in DES97. which are strong especially in the all-important situation of a separating boundary layer (see Section 3. Below we first address these concerns and then delve into practical issues in the remaining subsections. Yet continuing concerns of a conceptual nature could categorize DES as a method that is intuitively correct and often successful but dissatisfying to the by b-on: Universidade da Beira Interior (UBI) on 09/20/11. 190 Spalart . 70% resolved and 30% modeled). models are adjusted based on results rather than explicit tests. studies remove the concern that LES might be barred from supersonic flows. Rev.4). Downloaded from www. in LES and DES practice. systematic studies use filtered versions of DNS fields to steer subgrid-scale model development. In LES.. Several hybrid proposals rest on the idea of splitting the turbulent energy in a specified ratio (e. WEAKNESSES 3. one would adjust the single Smagorinsky constant to ensure that all six subgrid stresses are correct. 2007. therefore widening the range expected for DES. given a powerful numerical method. but this increases the decision load for the user. This is known as the Clark test or a priori study and could be performed with DES but has not. WMLES has been exposed to this issue less than DES. only CDES ). possibly because it sometimes seems unable to escape channel flow. and also because of history effects. The literature reflects a desire for an approach that is somehow more justified and mathematically defined than DES. The problem is more severe in wall-modeled LES (WMLES) and more severe again in DES.1. Figure taken from Chauvet et al. For personal use only.annualreviews. The essential difficulty is that the model has much more impact on WMLES and DES than it does on the notional LES situation. namely away from walls and with a grid spacing in the inertial range.41:181-202. 3. the model has a strong impact. Fluid Mech.g. Annu.

. As a result. in which the differential equation does not depend on the grid. 2008). the drag’s drift is not mastered to better than several percent and went unnoticed at the time. A typical observation after analyzing a grid-refinement study even in a simple geometry is the honest but vague statement that the findings are “suggesting a certain degree of grid convergence” (Nishino et al. Recent efforts at organizing the quality control of CFD in the RANS field. 3. would be defeated by precisely this dependence in LES and DES. the issue of an order of accuracy is clear. visualizations over a cylinder look just like those in Figure 2e. Travin et al. Equally active are controversies over the definition of in noncubic grid cells (see Section 4. the SAS and turbulence-resolving RANS work is motivated by the disruptive effects of in DES with ambiguous grids (see Section 3. careful users are justified in asking for one because it is.’s (2004) turbulence-resolving RANS approach has similar features but uses the ratio of strain to vorticity rather than a high derivative. WMLES does not deal with this problem much better than DES does. with a thin wire). the difficulty in demonstrating grid convergence is compounded by the residual variations owing to finite time samples. which allows full RANS function. For personal use only. Wall distance can be expensive to calculate and has unphysical effects ( • Detached-Eddy Simulation 191 . the sustained wide use of these two models suggests that it is manageable and has a substantial accuracy payoff. The www. Downloaded from www. which is highly active on short scales.2). with the wall-parallel spacing in excess of the boundary-layer thickness δ.g. of course has been standard. LES.annualreviews. after an initial transient. but 1. With RANS-LES hybrids. In a pure LES. For instance. a key step in CFD quality control. This stimulating controversy is not over.annualreviews.2. in principle. as discussed in Section 3.033 over the second half. Finally. with three levels of grid density in a boundary layer. The time-averaged drag coefficient is 1. is not by b-on: Universidade da Beira Interior (UBI) on 09/20/11. Menter et al. This led to scale-adaptive simulation (SAS). or DES. this is related to the desire for monotonic grid convergence.2 and Section 4. because even the RANS or LES nature of the solution is in some cases dependent on initial or inflow conditions.’s (2000a) LS1 cylinder case. for instance. 1997). Rev. Modeled-Stress Depletion and Grid-Induced Separation Modeled-stress depletion (MSD) and grid-induced separation have been the most significant practical issues and have been worse to deal with than initially anticipated (Spalart et al. although it has been proposed to dissociate the filter size and grid spacing. it is a boundary-layer grid. Menter et al. Fluid Mech. There is no theory that would extrapolate to infinite sample length. some flows have severe modulations and drift. this order exists but depends on the quantity in question. Figure 8a shows the roots of these problems. however.083 over the first half of the sample. Whether in DNS.3. 2009.4). In LES. f. for example. Nonuniqueness issues are most intense with delayed DES (DDES). Figure 7 uses Travin et al. searching for grid convergence to 1%. (2003) use an SAS model that appears to have a pure RANS nature but achieves LES behavior unlike any traditional RANS model.A separate line of critical thought regards the use of the grid spacing in the model. it has even been proposed to dispose with any length scale of the nature of a filter width or grid scale. An order of accuracy has not even been proposed for a simulation using both modes within DES. Annu.’s (2003) model differs from traditional ones in its use of a higher derivative of the velocity field. The first level matches the initial vision of DES. the resolved or total turbulent kinetic energy or the dissipation. the simulation covered a generous 40 cycles of shedding.41:181-202. Besides a philosophical interest in the true nature of turbulence models. It echoes the one in RANS modeling over the use of the wall distance [as in the Spalart-Allmaras and shear-stress transport (SST) models]. Although the sample is sufficient to capture the modulations of the lift signal. the lift excursions are also noticeably less intense over the second half.

Figure 8b is an early example of gradual grid refinement degrading a solution that was rather good when the RANS model was fully active (S.20 Instantaneous Cd. see also Caruelle & Ducros 2003).41:181-202. For personal use only. (2006) introduced by b-on: Universidade da Beira Interior (UBI) on 09/20/11. Cι 1. Secondary effects are covered in Section 4. which 192 Spalart . However. Spalart et al.30 1.6 Annu.00 –0.05 Time-averaged –0.6 0. DES users promptly explored the effects of grid spacing and sought high accuracy. using the SST F2 function. but with a model that has become grid dependent in an obscure and unintended manner. (1999) fully defined DES. A new feature of DES is that the entire boundary layer can be handled by RANS. The flow field then obeys the URANS equations.10 1. Either modification successfully prevents GIS by extending the RANS region. even in this case of incipient separation.3. (2006) coined the term MSD. but in a shedding mode rather than in a sound turbulence-resolving mode. Fluid Mech. with disturbing outcomes. Travin. The second level is the troublesome one: small enough for the eddy viscosity to be affected by the DES limiter but not small enough to support accurate LES content (slow LES development adds to this difficulty. this time on an airfoil (Menter & Kuntz 2002). Re = 5 × 104 . 2009. Logarithmic-Layer Mismatch Simulations with an LES nature in one region and a RANS nature in another were conducted long before DES. Spalart et al.4). well after the issue was detected by S. Figure 9 is a visualization of GIS. who pointed out a consequence of MSD called grid-induced separation (GIS). 3.2 0 Instantaneous Time-averaged b 1. which is applicable to most models. wall modeling near the walls of an LES draws on RANS technology. Separation in a nozzle is premature and induces unsteadiness.annualreviews.2 –0.3.25 1.95 0 40 80 120 160 200 tU/D Figure 7 tU/D Instantaneous (solid line) and time-averaged (dashed line) values of force coefficients on a cylinder: (a) lift and (b) drag. third level matches the needs of LES in the outer layer and thus of the extended use of DES as a wall model (see Section 3.4 0.4 1. see Section 3. Cd Cι. Whereas the RANS solution is steady and quite accurate.15 1. Deck (personal communication) and by Menter & Kuntz (2002). exploiting a history effect. Deck.a 0. It also is unsteady. Rev. Created only one year after Shur et al. Figure courtesy of A. and early channel LES studies even used wall functions. DES also naturally provides a simple wall model.3): The grid spacing in all directions is much smaller than δ. Menter & Kuntz (2002) proposed a solution applicable to the SST model called shielding. personal communication. in which the DES limiter is disabled as long as the flow is recognized as a boundary layer. 0 40 80 120 160 200 0. Downloaded from www. the DES solution suffers from early separation.

5 2. which blends into the resolved log layer ( y + roughly from 3000 to 15.0 0. but the study was successful in key respects.5 1.0 y/δ 0. with no need for averaging or danger of negative values.0 x/δ 1. LES content was sustained even with coarse grids.000). Very high Reynolds numbers were reached at little additional cost.1. detached-eddy simulation.5 0 3. conversely.’s method to Reynolds number and grid spacing. LEA.annualreviews.5 1. (2000) attempted.5 1.41:181-202. 2009.05 x/δ b Normalized wall pressure (PW/PC) 0.0 Annu.5 0 0 0.03 DES computation PR40 0. Deck.0 1.5 0 0 0. Laboratoire d’Etudes A´ rodynamiques. Rev.3 0 0 10 20 X/rt Figure 8 (a) Types of grids in boundary layers. e Nikitin et al.5 1. The model was robust. (b) Pressure distribution in a supersonic nozzle. Downloaded from www. For personal use only.0 2. which trade places as the grid is varied (Figure 10b). where h is the half-width of the channel. which blends into the modeled log layer ( y + roughly from 70 to 700). Fluid Mech.0 1.5 2.0 1. The dashed line represents the velocity profile.04 0. An increase in Reynolds number on a fixed grid (same but refinement in y to retain a first y + near 1) lengthens the modeled part of the profile. because = h/10 in most runs. Figure courtesy of S. x/δ 0. The results were not • Detached-Eddy Simulation 193 . lengthens the resolved-turbulence part of the profile. www.5 4. The Reynolds shear stress comprises modeled stress and resolved stress. Figure 10a illustrates the response of Nikitin et al.02 SA-URANS computation PR40 0.5 2. DES.annualreviews.01 LEA steady experimental data PR41. SA-URANS.0 3. Spalart-Allmaras unsteady by b-on: Universidade da Beira Interior (UBI) on 09/20/11.0 a y/δ 0.0 0 0. Grid refinement.

000. (The study was also marked by deliberate constraints.2 0 102 103 104 105 0 0. For personal use only. In addition. wall-modeled large-eddy simulation. The imperfection is that the two log layers are misaligned.2 0. by almost three wall units of velocity U + .000.annualreviews.4 0. –5.00 –10. Downloaded from www. Each profile is shifted by five U + units. The probability that this log-layer mismatch would be zero was nil because this study used the pure DES97 model. This is different from MSD in a near-RANS boundary layer. this is highly inaccurate. The lower two curves use approximately 140. The dashed line represents the log law. Nikitin et al. to ensure the findings would translate into practice. and the upper curve uses approximately 1.000 grid points. (2000) mentioned the ensuing error of the order of 15% for the skin-friction coefficient but did not mention that the slope dU/d y is too high by 65% at y = .a 0 b Velocity by b-on: Universidade da Beira Interior (UBI) on 09/20/11.6 0. Rev. Arrows indicate separation.00 Figure 9 Vorticity contours over an airfoil: (a) Reynolds-averaged Navier-Stokes and (b) detached-eddy simulation.) All other wall-modeling approaches have required adjustments to align their log layers. adjusted for other purposes. 194 Spalart .000 points.0 25 20 15 10 5 101 y Figure 10 + y/h Channel-flow.6 U+ τ+ 0. Figure adapted from Nikitin et al. such as equal grid spacing in the wall-parallel directions. 2009.4 0.0 0. Fluid Mech.8 1. Reτ = 20. (a) Velocity: Reτ = 2000 and 20. 2000. grid refinement merely moves the same amount of mismatch closer to the wall.8 40 35 30 a b 0. Figure taken from Menter & Kuntz 2002. (b) Modeled and resolved shear stress: coarser grid (dashed line) and finer grid (solid line).000. which 1. Locally.

with 64 points around) and the standard definition of is used (see Section • Detached-Eddy Simulation 195 . 3.4. Recent work includes. Travin et al. much smaller than the lateral grid spacing. Unfortunately. 2009. This is the case with the book-shaped grid cells typical of such regions. with one dimension much smaller than the other two. RECENT PROPOSALS 4. By consensus.4). but in a manner dependent on the alignment and shape (book or pencil) of the grid cells. a zonal approach that disables the model in the mixing layer and activates implicit LES is visually far more successful (Shur et al.annualreviews. it is remediable with grid refinement. Fluid Mech. www. eddy viscosity disabled. making the DES limiter inoperative. and may be a perverse effect of the careful adaptation of the grid to the shear layer. relatively thick and curved boundary layers could make using complex models worthwhile.a 1 0 1 0 b y –1 0 Figure 11 2 4 x 6 8 y –1 0 2 4 6 8 x Vorticity in a jet: (a) standard detached-eddy simulation and (b) implicit large-eddy simulation. Rev.’s (2008) cubic explicit algebraic stress models. which has been smooth. Greschner et al. The motivation for complex models is debated because the RANS region normally comprises thin shear layers. Annu. with the expectation that the boundary layer is treated with RANS and is quasi-steady. The model defaults to RANS until the layer thickness reaches approximately 40% of D because the mixing length in a RANS-treated mixing layer is approximately one-tenth the vorticity thickness. standard DES on typical grids does not achieve this switch very fast at all (Figure 11). This problem has received and deserved attention. Figure courtesy of M. the sooner this takes place.annualreviews. 2006). Travin et al. motivate efforts to resolve log-layer mismatch (Piomelli & Balaras 2002.1. Alternate Reynolds-Averaged Navier-Stokes Models The original formulation of DES rested on the simple Spalart-Allmaras model. 10% of the diameter D. in particular the SST model. 2005b. becomes more severe as the grid is refined. Shur. and no CFD system should ever be confined to one model. but the free shear layer it feeds develops LES content. The practical advantages of wall modeling by DES. the better.4). The DES model fails to sense the opportunity because the lateral grid spacing is loose ( by b-on: Universidade da Beira Interior (UBI) on 09/20/11. for instance. but unlike the two problems discussed in the preceding subsections. For personal use only. Other definitions are then more successful (see Section 4. and the understanding that in practice thick wall-bounded layers lead to LES grids in the sense of Figure 8a. 4.41:181-202.c). Slow Large-Eddy Simulation Development in Mixing Layers Separation is the essential flow feature motivating DES. Follow-on work by Piomelli and his group also showed that the near-wall solution has poor LES content. Downloaded from www. (2000b) pioneered the adaptation to two-equation models.

Yet many groups have had success with IDDES in practice (Mockett & Thiele 2007). such as the jet featured in Figure 11. Downloaded from www. while addressing the MSD issue that affects DES97 (see Section 3. is made infinite (as opposed to zero in implicit LES). DDES was shown to resolve GIS.2. thus leaving it to the solution process to determine separation. some involving the cell Reynolds number. with high eddy viscosity and weak perturbations. it remains in that mode and finds a steady state.2) and to clarify the role of each region. Delayed Detached-Eddy Simulation and Improved Delayed Detached-Eddy Simulation A key motivation here is precisely to avoid zonal measures. Other components of IDDES include new empirical functions. it handled a backward-facing-step flow well. The approach is also nonzonal and aims at resolving log-layer mismatch in addition to MSD. Zonal Detached-Eddy Simulation In zonal DES. Annu. Following Menter & Kuntz (2002). which includes the wall distance and not only the local characteristics of the grid. Compared with DES. (2007) in jets. boosting the resolved Reynolds stress. 2006). + ). in which the use of a single but versatile equation set is central. the author may disagree. (2007) for a base flow. in RANS regions. depending on the initial condition. Simon et al. Deck (personal communication) is in favor of RANS as the default mode. 4. even with grids that would cause severe MSD both upstream of the step and all along the opposite wall. the solution has two branches. This is probably the strongest departure from the original concept of DES. which stimulates by b-on: Universidade da Beira Interior (UBI) on 09/20/11. The history effect introduced by shielding or by DDES has consequences in terms of the uniqueness of solutions.annualreviews. Which mode will be the default. in which decisions are needed for numerous regions (including the thickness of regions meant to contain RANS boundary layers). Fluid Mech. more importantly. DDES is likely to be the new standard version of DES. If the flow starts in an LES state with low eddy viscosity and sufficient LES content. 2009. the user explicitly marks different regions as RANS or as DES (Deck 2005). A fair question to propose to zonal DES proponents concerns complex flows. The motivation is to be fully safe from MSD and GIS (see Section 3. and which will be the exception? S. which address log-layer mismatch and the bridge between wall-resolved and wall-modeled DES (grids with moderate values of the spacing in wall units. 196 Spalart . In effect. The modification tends to depress near the wall and give it a steep variation. ZDES appears simultaneously more powerful and less self-sufficient.2). without impeding LES function after separation. and. If the flow is in a RANS state. even if the wall-parallel grid spacing would normally activate the DES limiter. it settles into a statistically steady ¨ LES.3. and Slimon (2003) in a duct. For instance. but this situation perplexes some observers (Frolich & von Terzi 2008). This is similar to issues with zonal control of laminar-turbulent transition. Chauvet et al. Zonal DES worked well for Brunet & Deck (2008) in the important problem of wing buffet. and creates most of the conceptual and practical challenges. there is the concern that smooth-wall separation is normally not known at the time the zones are set. This is a formal deviation from DES97 but not a different mission. DDES detects boundary layers and prolongs the full RANS mode. Both solutions are valid. These functions make the formulation less readable than that of DES97.4.41:181-202. The geometries in these studies were simple. For instance. so that the limiter now depends on the solution (Spalart et al. 2008). 2000 and Figure 10). This detection device depends on the eddy viscosity. One basis is a new definition of . For personal use only. in a channel flow with periodicity and a grid and time step capable of LES (as in Nikitin et al. Rev. Improved delayed DES (IDDES) is more ambitious yet (Shur et al.

and shock capturing. Other proposals relate instead to transition. the code used in Figure 1a is unstructured and uniformly based on second-order upwind differencing.g. 5. (2000b) and is now widely used (e.4. Downloaded from www. has been neglected. Mockett & Thiele 2007. 2007) have tested with some success definitions radically different from the standard one in DES. more precisely the growth of LES content. the RANS mode creates no sound. More striking is the behavior of models in the tripless mode (Travin et al. Temmerman & Hirsch 2008).’s (2007) length scale ≡ Nx2 y z + Ny2 x z + Nz2 x y. The mature solution depends on the level of the turbulence variables in the initial field. Mockett & Thiele 2007). however. such as airfoils near maximum lift. Several groups (Breuer et al. Modified Δ Length Scales The IDDES length scale’s principal motivation is in a fully turbulent wall layer in the LES mode. which is an essential tool for capturing the drag crisis of smooth bluff bodies. The unsteady capability.annualreviews. without implicit time integration or multiblock capabilities. but it displays generous LES content. complex-geometry compatibility. Fluid Mech. it is more common to start from a RANS code. not only in their treatment of turbulence. The addition of a transport-equation turbulence model is not trivial. with resolution of high frequencies and short waves. The debate is whether promoting the 2D Kelvin-Helmholtz instability. is not unknown in RANS practice. This hybridization was introduced by Travin et al.4) with the purpose of allowing the KelvinHelmholtz instability to by b-on: Universidade da Beira Interior (UBI) on 09/20/11. 2009. dissipation. but the near-2D LES mode could create too much. and few codes have shockcapturing capability (Hou & Mahesh 2004). namely the maximum dimension of the grid cell.Nonuniqueness. 2003. Conversely. NUMERICAL REQUIREMENTS DES codes need qualities that are absent in many RANS codes and others that are absent in many LES codes. In contrast with the DDES modification (which raises eddy viscosity in specific situations). The reduced length scales have an advantage over the implicit LES approach shown in Figure 11 as they are not zonal and can reverse to the normal scale when the grid is not strongly anisotropic.41:181-202. Chauvet et al. which of course reduces .org • Detached-Eddy Simulation 197 . They all appear to be responses to the problem of LES development in mixing layers (see Section 3. These codes often have placed a high priority on convergence to a steady state. and time-integration errors are key (Caruelle & Ducros 2003. knowing that the true switch to 3D turbulence occurs only once the mixing-layer thickness has caught up with the lateral grid spacing. Some flows. then = max( x. if it is aligned with the axes. Rev. For instance. therefore worsening the MSD tendencies. Chauvet et al. y. is aimed at the situation in which the vorticity is closely aligned with one of the grid lines. Yan et al. 2007. Strelets 2001. www. 2000a). it is best to avoid blanket statements. Considering the filiation of the model. For personal use only. If the starting code is an LES code. Therefore.. and the other demands all benefit from numerical dissipation. As a result. The differencing scheme is centered (nondissipative) or nearly so in the LES region and is more strongly upwind in the Euler and RANS regions. 4. let alone unstructured grids. The most effective schemes are structured and hybrid. Some use the time-honored definition in LES = ( x y z)1/3 . is far superior to letting the mixing layer thicken in the RANS mode. Annu. The priority was given to high orders of accuracy.annualreviews. but also in their numerics. have hysteresis both in real-world situations and in CFD. common obstacles include the limitation to simple geometries. an extensive testing campaign and modifications to reduce dispersion. z). where N is the unit vector aligned with vorticity. all these definitions tend to reduce it. but its physical justification is thin.

Even in DES97. transient RANS. flow simulation methodology. An unfortunate trend is that models have moved away from the simplicity of DES97 in terms of the equations and nonuniqueness of solutions (in DDES and IDDES) and in terms of the user decision load and need to mark regions (in ZDES).41:181-202. semideterministic method. even in the key region. and the self-adapting model) (some are found in ¨ Sagaut et al.An advantage of DES is the ease of programming and application. limited numerical scales. It does not detract from their value. What may be an ideal of CFD. but this could change in the future. Rev. The principal concerns are GIS and in general the potentially poor knowledge of the nature of the simulation in each region of a complex flow: driven URANS. nonlinear disturbance equations. . a form of a RANS-LES hybrid that is capable of full RANS function in boundary layers will be in use for the foreseeable future in many industries. For personal use only. Owing to space limitations. Such comments are encountered more often in conversations and anonymous reviews than in publications. Fluid Mech. Downloaded from www. resolution. time sample. LES-content creation in attached flows will flourish.g. partially integrated transport model. There are signs that a productive DES user community has formed. Greschner et al. lattice Boltzmann method. and the numerical quality of the codes will receive sustained attention. the genuine LES studies struggled with all the issues of lateral domain size.. organized eddy simulation. The simulation was formally a DES. Potentially. Users by now have identified situations in which DES gives too little eddy viscosity and others in which it gives too much. 6. and the results were fine. especially now that the dangers of ambiguous grids are known. and efforts toward more predictable behavior under grid variations and better wallmodeling performance will continue. (2008) deem that “DES is still in its infancy and undergoes continuing improvements. 2003).” Under one name or by b-on: Universidade da Beira Interior (UBI) on 09/20/11. However. it is activated directly from the menu of turbulence models in many of the vendor CFD codes. and initiation of LES content in attached flows. 2006. We must recognize. and so on. In contrast. or that the user community is being properly informed. OUTLOOK It is certain that DES has a future and therefore deserves a critique. Programs such as DESider and focused workshops will be most beneficial to the progress of DES and other hybrids (Peng & Haase 2008). partially averaged Navier-Stokes. large steps in the grid spacing can be used to steer the solution toward one mode or the other. very large eddy simulation. An early example of this was an entry in the LESfoil workshop (Mellen et al. except in the simplest of flows. The nominally universal character of DES makes these observers justifiably dubious that a sufficiently error-proof approach results. 2009. be monotonic) and follow a known power of the grid size. so that grid design can become involved. there is every indication from the grid that the simulation was actually in RANS mode. namely that grid refinement will do no harm (in other words. or LES. time step. This nature can change under grid refinement and become ambiguous. extra-large eddy simulation. a school of thought that considers DES to be a somewhat unsafe activity. grid resolution. A clear need in practice is to organize and facilitate grid generation and to set guidelines for systematic refinement.annualreviews. It will also remain conceptually difficult. spontaneous URANS. Frolich & von Terzi 2008). as users not invested in turbulence and/or too trusting of the experts could accept results without verifying LES content. will remain elusive in DES and LES (without explicit filtering). and therefore it is not the case that any grid refinement improves the solution. This is a doubleedged advantage. and the task of resolving them is an inspiring one. however. this review does not discuss hybrid RANS-LES methods besides DES and SAS (e. I do not believe that any of these methods provides a clear remedy to the difficulties discussed here. Locally ambiguous grids may be a permanent feature of 198 Spalart Annu.

Comparison of DES. Deck SF. Sci. Presented at AIAA Aerosp. OMAE2006-92334. Pap. Presented at AIAA Aerosp.. by b-on: Universidade da Beira Interior (UBI) on 09/20/11. 2008. Int. Exhib. Downloaded from www. FUTURE ISSUES 1. LITERATURE CITED Allen R. Deck. Shur. The choice between zonal and nonzonal treatments of the turbulence needs to be addressed. Pap. Detached-eddy simulation of flow over a sphere.annualreviews. For personal use only. 42nd.practical DES. Rev. Fluid Mech. Reno. pp. AIAA-2002-0425 Constantinides Y. No. Methods Fluids 41:357–88 Brunet V. The link between the DES flow field and the exact or DNS flow field should be established. Allmaras. Levasseur • Detached-Eddy Simulation 199 . Cummings RM. Mallet M. 2007. Morton SA. 45th. 2005. Kirkham D. Meet. Sci. Thiele F. Zonal-detached eddy simulation of transonic buffet on a civil aircraft type configuration.. 2002. No. New York: ASME Int. Jacquin L. AIAA J. LES and DES simulations for aircraft design. Mendonca F. RANS and DES turbulence model predictions of noise on the M219 ¸ cavity at M = 0. Zonal detached-eddy simulation of the flow around a high-lift configuration with deployed slat and flap. Strelets. Aerosp. Pacheco R. Guidelines for implementing detached-eddy simulation using different models. ultimately with grid adaptation. Numerical prediction of bare and straked cylinder VIV. Forsythe JR. Mazaev K. Proc. Reno. The generation of resolved turbulence in attached boundary layers needs to become routine and efficient. Numer. 40th. DISCLOSURE STATEMENT The author is not aware of any biases that might be perceived as affecting the objectivity of this review. Detached-eddy simulations of attached and detached boundary layers. Zonal detached eddy simulation of a controlled propulsive jet. 2005. Int. Ducros F. is it justified to simulate the flow past a car. Conf. 4:135–51 Breuer M. 2006. Annu. The numerical resolution over relevant geometries needs improvement. No.85. Sci. CFD 17:433–51 Chalot F. Exhib. 43:2372–84 www. J. Detached-eddy simulation of slat and flap aerodynamics for a high-lift wing. Int. 3. 45:2458–73 Constantinescu GS. 2007. Meet. Offshore Mech. Petit G. 2003. Exhib. Mockett. when the wiper and door handle are not well resolved? The answer depends on the purpose of the simulation. ACKNOWLEDGMENTS I am grateful to Drs. Aeroacoust. Pap. Technol. Artic Eng. Pap. 2007. and Travin for their comments on this manuscript and their partnership over the years.annualreviews.. Reno. Deck S. Jovici´ N. 2004. J. No. 2. Oakley OH. AIAA J. AIAA-2004-1233 Deck S. AIAA-2007-0723 Chauvet N. 25th Int. Mockett C. J. See Peng & Haase 2008. RANS and LES for the separated flow around a c flat plate at high incidence. Reau N. Meet. Presented at AIAA Aerosp.41:181-202. 11:376–85 Caruelle B. Squires KD.. One might ask. Sci. 182–91 Bunge U. 2009. 4.

Exhib. Presented at Int. Huang J. Squires KD. 41:573–81 Mendonca F. 261–70 Forsythe JR. 41st. Detached-eddy simulation of a nose landing-gear cavity. No. de Charentenay J. 2002. de Charentenay J. 2004. 2004. Oscillation of vortex breakdown location and control of e the time-averaged location by blowing. Perrin R. Modeling of cavitation inception in high-Reynolds number circular jets using detached-eddy simulation. Chen H. 5th. 2008. Downloaded from www. Aerodyn. 40th. 222–31 Mockett C. Paterson EG. Pap. No. Shanghai. Workshop LES Acoust. IUTAM Symp. AIAA-2003-0857 Krishnan V. Presented at AIAA/CEAS Aeroacoust. Presented at AIAA Aerosp. Wurtzler K. Presented at Appl. 2004. Pap. Reno. von Terzi D. 2000. Sci. 2002. Pap. No. 44th. Naval Hydrodyn. Hoffmann KA. Presented at. Squires KD.85. 26th Symp. 36th. Phys. Pap. 2004. Wurtzler E. Frolich J. Prediction of separated flow characteristics over a hump using RANS and DES. Exhib. Sci. Pap. Forsythe JR.. J. 38:793–803 Mockett C.. Deck S. Hybrid LES/RANS methods for the simulation of turbulent flows. 2003.41:181-202. AIAA-2003-0549 Hedges LS. Presented at AIAA Aerosp. “Unsteady Separated Flows and Their Control. Reno. No. 2nd. AIAA-2003-4217 Spalart Annu. 2003. Gottingen Menter FR. AIAA-2003-0767 Mitchell AM.First application to a full aircraft in a stall. Fluids Eng. Presented at Int. Roy S. AIAA J. Presented at AIAA Aerosp. Int. eds. High Reynolds number DES simulations of vortex breakdown over a 70 degree delta wing. Prediction of sound generated by a rod-airfoil configuration using EASM DES and the generalised Lighthill/FW-H analogy. Squires KD. 89–104 McCallen R. Thorigny P. 2007.. Prog. Sci. Forsythe JR. Wurtzler KE. AIAA-2002-0586 Forsythe JR. and Trains. Ross J. South Carolina. Berberis D. Comp. Lewis M. Mahesh K. 124:413–23 Hou Y. Meet. Yan J. Detached-eddy simulation with compressibility corrections applied to a supersonic axisymmetric base flow. Exhib. Bender R. pp. See McCallen et al. Greece Maddox S. For personal use only. Proc. Hilton Head. 2002. Adaptation of eddy-viscosity turbulence models to unsteady separated flow behind vehicles. Detached-eddy simulations over a simplified landing gear. D´ lery J. Conf. the best-known alternative to DES for hybrid RANS-LES. Orlando. Travin A.. No. Buses. turbulent flows. Unsteadiness of an axisymmetric separating-reattaching flow. Pap. 2003.. Allen R. AIAA-2006-0858 Forsythe JR. pp. A robust. Reno. New York: Springer ¨ Mellen CP. Sci. Presents SAS. Rev. See Peng & Haase 2008.. Squires KD. Heat Fluid Flow 28:1379–90 Greschner B. China Morton SA. 2009. 2003. Orlando. Aircraft 41:193–200 ¨ Frolich J. Meet. Towards understanding LES and DES for industrial ¸ ¨ aeroacoustic predictions. Detached-eddy simulation of the ground transportation system. Fluid Mech. 21st. CFD prediction of narrowband and broadband cavity ¸ acoustics at M = 0. Xiao Z. 339–52 Menter FR.. No. AIAA-2003-3303 Mendonca F. 2004. Spalart PR. Molton P. Six degree of freedom computation of the F-15E entering a spin. AIAA J. 2007. Meet. Pap. Development and application of SST-SAS turbulence model in the DESider project. Zhang Y. No. Meet. Browand F. Basu D. implicit algorithm for direct numerical simulation of compressible. 2004. Jacob MC. See Peng & Haase 2008. Trujillo MF. See McCallen et al. Presented at AIAA Flow Control Conf. Exhib. Detached-eddy simulation of the F-15E at high alpha. Conf. Fluids 37:402–13 Hamed A. Fluids 19:065103 Edge BA. Squires KD. 2002. Kuntz M. Allen R. J. Rodi W. Phys. Pap. with good agreement. Spalart PR. Sci. 2006. Spalart PR. A scale-adaptive simulation model for turbulent flow predictions. Thiele F. AIAA-2004-2224 Langtry RB. Simulation of wing-body junction flows with hybrid RANS/LES methods. Fluid Mech.. 2007. pp. 2006. Das by b-on: Universidade da Beira Interior (UBI) on 09/20/11. 2008. Conf. J. Knacke T. Kirkham D. Clearest critique of grid-induced separation. 2007. Comp. 2008. Aerosp.annualreviews. The Aerodynamics of Heavy Vehicles: Trucks. Exhib. Reno. colocated. No. Rome Egorov Y. 205:205–21 Kapadia S. Overview of detached-eddy simulation for external and internal turbulent flow applications. Strang WZ. Kuntz M. 2004. J. Portland. Thiele F.” Corfu. Casalino D.. Detached eddy simulation of supersonic flow over cavity. Greschner B. 41st. 200 . Conf. Presented at Thermophys.. Lessons from LESFOIL project on large-eddy simulation of flow around an airfoil. 2003. 2003. Presented at AIAA Aerosp. Detached-eddy simulation over a reference Ahmed car model. Demonstration of improved DES methods for generic and industrial applications. Thiele F. 2008. pp.. 44:349–77 Fu S. Menter F.

org by b-on: Universidade da Beira Interior (UBI) on 09/20/11. Morris PJ. Moin P.Morton SA. Portland. 465–80. In Engineering Turbulence Modelling and Experiments 4. J. Squires KD. Presents a wide collection of recent work on DES and other hybrid approaches (but not all). 4:213–46 Shur ML. Rep. J. Travin A. Meet. Aeroacoust. Numerical simulation of turbulent flows. C Liu. Heat Fluid Flow 21:252–63 Spalart PR. In press Simon F. ASME Int. Wasistho B. 669–78. Noise prediction for increasingly complex jets. Va. 2005c. Reno. Unsteady turbulent flow simulations of the base of a generic tractor/trailer. Strelets M. Presented at AIAA Aerosp. Pankajakshan R. Jou W-H. Roberts GT. Detached eddy simulation of massively separated flows. Aerodynamic simulation of heavy trucks with rotating wheels. Rev. 29–45 Squires KD. Congr. 2001. Exhib. 2006. Squires KD. Meet. 71:361–73 Spalart PR. including models other than Spalart-Allmaras.. Annu. Int. Shur M. 1997. Fluid Mech. Guillen P. Rev. J. 44th. Palm Springs.annualreviews. 39th.. Strelets MKh. Spalart PR. Meet. Numerical simulation of the compressible mixing layer past an axisymmetric trailing edge. UK: Elsevier Sci. Presented at AIAA/ASME/ASCE/AHS/ASC Struct. DeChant LJ. Mitchell BCJ. Reno. Whitfield DL. Berlin: Springer Piomelli U. Balaras E. Fluid Dyn. Strelets MKh. Dyn. No. Comments on the feasibility of LES for wings. and on a hybrid RANS/LES approach. In Direct and Large-Eddy Simulation V. Computation of internal separated flows using a zonal detached eddy simulation approach. AIAA-2004-2255 Sagaut P. Hampton. Strelets MKh. Rev. Deck S. 2000. Eng. Travin A. Taylor LK. Terracol M. http://techreports. Proc. 2000. IMECE2003-43881. and two-dimensional examples. Young person’s guide to detached-eddy simulation grids. Garbaruk AV. ed W Rodi. Pap.. AIAA-2006-1394 Strelets M. Downloaded from www. Strelets M.. Reno..41:181-202. 201 . Squires • Detached-Eddy Simulation Concise exploration of DES for wall modeling inside LES. Exhib. 2007. Barone MF. Multiscale and Multiresolution Approaches to Turbulence. 20:181–95 Spalart PR. Spalart PR. Struct. Sci. Pap. Oxford. 2004. The status of detached-eddy simulation for bluff bodies.. 34:349–74 Rogallo R. 2008. and successful prediction of airfoil forces at all angles. Dordrecht: Kluwer e Sreenivas K. pp. Annu. 2001. Phys Fluids 12:1629–32 Nishino T. 44th. Spalart PR. AIAA-2004-1676 Nikitin NV. Comparison of two. Advances in Hybrid RANS-LES Modelling. AIAA-2001-0511 Shur ML. Pap.. 2009. Simulation of active flow control on a stalled airfoil. See McCallen et al. BJ Geurts. Comp. Deck S. Annu. Aeroacoust. Reno. Unsteady RANS and detached-eddy simulations of flow around a circular cylinder in ground effect.. 2001. Brown JC. 2008. New York: ASME Int. Travin A. Pap. Meet. No. pp. J. Further steps in LES-based noise prediction for complex jets. 2006. Exhib. Hedges L. Spalart PR. No. No. Sci. High resolution turbulence treatment of F/A-18 tail buffet. Deck S. Combust. Presents a wide range of applications. Strelets pp. Strelets MKh. 2003. Presented at AIAA Aerosp. Tech. Spalart PR. basic equations (with CDES constant undetermined). Allmaras SR. Coll. Three-dimensionality in unsteady Reynoldsaveraged Navier-Stokes simulations of two-dimensional geometries. London: Imp. No. Z Liu.larc. In Advances in DNS/LES. Detached-eddy simulation: current status and perspectives. For personal use only. 2006. Spalart PR. NASA CR-2001-211032. An approach to wall modeling in large-eddy simulations. 591:215–53 Slimon S. AIAA J. Fluid Mech. Part II: applications. Center. Exhib. 2005b.nasa. Mech. 2008. AIAA-2001-0879 www. Fluid Mech. A new version of detachededdy simulation. Sci. 45th. Spalart PR. 2004. 4:247–66 Shur ML. Zhang X. Conf.pdf Spalart PR. 1984. Strelets M. A hybrid RANS-LES model with delayed DES and wall-modeled LES capabilities. 1999. Introduced delayed DES to combat grid-induced separation. Presented at AIAA Aerosp. 2005a. 16:99–137 Roy CJ. Presented at AIAA Fluid Dyn. Mater. Exhib. Sagaut P. Shur ML. Pap. ed R Friedrich.annualreviews. Press Shieh CM. Int J. Kholodar DB. ed. Nicoud F. 34th. Theor. Wall-layer models for large-eddy simulations. Part I: methods and tests. resistant to ambiguous grid densities. Travin A. Pap. Sci. 24:18–33 Peng SH.and three-dimensional cavity flows. Flow Turbul. O M´ tais. 2006. OH: Greyden Press Spalart PR. 43:1230–42 Shur ML. eds. 137–47. pp. Strategies for turbulence modelling and simulations. Fluids Struct. Heat Fluid Flow. Squires KD. First true 3D application. 2004. Presented at AIAA Aerosp. Nichols DS. No. calibration of CDES . No AIAA-2006-0485 Shur ML. Noise prediction for increasingly complex jets. Int J. Int. Merlen A. Shur ML. 2003. 2002. 2004. 2004. Travin A. Calif. Cummings RM. Fluid Mech. Spalart PR. Conf. D Laurence. Columbus. Pap. Detached-eddy simulation of an airfoil at high angle of attack. Haase W. 39th. Langley Res. Motivation for DES.

New York: Kluwer Acad.S. Proc. Delft. Presented at Eur. 46:118–31 Travin A. W Rodi. New York: ASME Int. 2000a. 2008. See Peng & Haase 2008. Detached Eddy Simulation of a surface mounted cube at high Reynolds number. CFD. Finland a a Travin AK.. Wilson RP. Shur ML.annualreviews. La Canada. 239–54. ASME Joint U. Travin AK. Methods Appl. Deck by b-on: Universidade da Beira Interior (UBI) on 09/20/11. and refined definition of DES in appendix.41:181-202. Delayed detached-eddy simulation and analysis of supersonic inlet buzz. 1998. Compressibility effects on turbulent separated flow in streamwise-periodic hill channel. No. Haupt SE. Turbulence Modeling for CFD. Hirsch Ch. Shur ML. 63:293–313 Travin A. Cong. Spalart PR. Strelets MKh. Fluid Mech. Physical and numerical upgrades in the detached-eddy simulation of complex turbulent flows.First DES with grid refinement. Tawackolian K. Eng. 2008. Conf. Shur M. Strelets MKh. 232–41 Trapier S. Strelets M. Temmerman L. Towards a successful implementation of DES strategies in industrial RANS solvers. Presented at Eur. Jyv¨ skyl¨ . Spalart PR. 2000b. Thiele F. 2006. Shur M. Peltier LJ. Fluids Eng. Combust. ECCOMAS CDF 2006. Neth. Conf. See Peng & Haase 2008. Sci. In Advances in LES of Complex Flows. Pap. Presented at AIAA/CEAS Aeroacoust. CA: DCW Ind. Computation of jet noise using a hybrid approach. Comput. 316–25 202 Spalart . Annu. 13th. AIAA J. pp. R Friedrich. Kleiser L. AIAA-2007-3621 Ziefle J.. pp. 2009. Flow Turbul. 2007. 2006. Yan J. Spalart PR. ˜ Wilcox DC. 2004. Eur. Spalart PR. ed. Summer Meet. fair agreement on the drag crisis. Downloaded from www. Duveau P. Improvement of Delayed Detached-Eddy Simulation for LES with wall modelling. Michel U. Detached-eddy simulations past a circular cylinder. part 2. 2008. Rev. For personal use only. On URANS solutions with LES-like behaviour. Kunz RF. pp. Strelets M.

Williams.A.F. Weissburg p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p73 Hemodynamics of Cerebral Aneurysms Daniel M. Nemat-Nasser p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p 1 Optimal Vortex Formation as a Unifying Principle in Biological Propulsion John O. Sforza. Clercx p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p 143 Study of High–Reynolds Number Isotropic Turbulence by Direct Numerical Simulation Takashi Ishihara.J. P. Dabiri p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p17 Uncertainty Quantification and Polynomial Chaos Techniques in Computational Fluid Dynamics Habib N. 2009.A.annualreviews. James p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p 129 Laboratory Modeling of Geophysical Vortices G. and Yosuke Tanaka p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p53 The Hydrodynamics of Chemical Cues Among Aquatic Organisms D. Fluid Mech. Yasushi Okada. v . Najm p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p35 Fluid Dynamic Mechanism Responsible for Breaking the Left-Right Symmetry of the Human Body: The Nodal Flow Nobutaka Hirokawa. and by b-on: Universidade da Beira Interior (UBI) on 09/20/11.Contents Von K´ rm´ n’s Work: The Later Years (1952 to 1963) and Legacy a a S. F. Putman.J. and Juan Raul Cebral p p p p p p p p p p p p p p p p p p p p p p p91 The 3D Navier-Stokes Problem Charles R.R. Christopher M. Webster and M.J. Downloaded from www. and Yukio Kaneda p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p 165 Detached-Eddy Simulation Philippe R.E. van Heijst and H. 2009 Annu.H. Rev.S. Toshiyuki Gotoh. Penner.41:181-202. For personal use only. Spalart p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p 181 Morphodynamics of Tidal Inlet Systems H. Zimmerman p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p 203 Annual Review of Fluid Mechanics Volume 41.F. de Swart and J.T. Libby. Doering p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p 109 Boger Fluids David F.

For personal use only.shtml vi Contents . and Koji Fukagata p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p 231 Ocean Circulation Kinetic Energy: Reservoirs.C. Yuji Suzuki. Fluid Mech. Collins p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p 405 Rheology of the Cytoskeleton Mohammad R.L. Volumes 1–41 p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p 463 Errata An online log of corrections to Annual Review of Fluid Mechanics articles may be found at http://fluid. and Shear in Stellar Interiors Mark S.annualreviews. Mofrad p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p 433 Indexes Cumulative Index of Contributing Authors. Downloaded from www. Fluid and Solute Transport in Bone: Flow-Induced Mechanotransduction Susannah P.Microelectromechanical Systems–Based Feedback Control of Turbulence for Skin Friction Reduction Nobuhide Kasagi. Rev. and Sinks Raffaele Ferrari and Carl Wunsch p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p 253 Fluid Mechanics in Disks Around Young Stars Karim Shariff p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p 283 2009. by b-on: Universidade da Beira Interior (UBI) on 09/20/11. Magnetism. Miesch and Juri Toomre p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p 317 Annu. Volumes 1–41 p p p p p p p p p p p p p p p p p p p p p p p p p p p p p 455 Cumulative Index of Chapter Titles. Salazar and Lance R. Fritton and Sheldon Weinbaum p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p 347 Lagrangian Properties of Particles in Turbulence Federico Toschi and Eberhard Bodenschatz p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p 375 Two-Particle Dispersion in Isotropic Turbulent Flows Juan P.41:181-202.