Detached Eddy Simulation
Also known as: DES, hybrid RANS-LES
Detached Eddy Simulation (DES) is a hybrid turbulence modeling approach introduced by Spalart in 1997 that combines the computational efficiency of RANS in attached boundary layers with the accuracy of LES in separated wake regions. By automatically switching between RANS and LES based on local grid spacing and turbulence length scales, DES provides superior predictions for flows with large separations, shear layers, and vortex shedding at a cost between pure RANS and pure LES. DES has become the standard method for complex aerospace applications involving separation and transient phenomena.
Read the full method
Sign in with a free account to read this section.
Method map
The neighbourhood of related methods — select a node to explore.
When to use it
DES is ideal for moderately complex separated flows at moderate to high Reynolds numbers (Re > 10,000) where full LES is too expensive but RANS is inadequate. Use DES for aerospace applications (landing gear, store separation, blunt bodies), automotive flows (A-pillars, side mirrors), and industrial configurations with backflow regions. DES is excellent when both steady mean flow and unsteady structures matter. Avoid DES for purely attached boundary layer flows (RANS is sufficient and cheaper) or extremely complex geometries (grid generation becomes prohibitive).
Strengths & limitations
- Automatic switching between RANS and LES based on grid resolution; no manual tuning needed
- Significantly more accurate than RANS for separated flows and vortex shedding
- Substantially cheaper than LES; typical cost is 5-10 times RANS instead of 100-1000 times
- Robust and stable compared to pure LES; grid independence issues of LES are reduced
- Works well with modern unstructured mesh generation; no strict requirements on boundary layer resolution like wall-resolved LES
- More expensive than RANS; requires unsteady time-stepping and statistical averaging
- Requires fine grid in separation regions; coarse separation zones remain in RANS mode with limited accuracy
- Depends on RANS model quality; Spalart-Allmaras is less robust than k-omega SST for some flows
- RANS-LES mismatch at transition regions can cause gray-area problems where neither RANS nor LES models are appropriate
- Initial transient behavior requires careful initialization to reach statistically steady state
Frequently asked
What is the difference between DES and LES?
DES uses RANS in boundary layers and LES in separated regions, decided automatically by grid resolution and length scales. LES uses LES everywhere if grid is fine enough globally. DES is cheaper (fine grid only where needed) but less accurate in boundary layers. LES is more accurate throughout but more expensive. DES is practical for industrial applications; LES is for research where budget permits.
What is the gray-area problem and how do I avoid it?
Gray-area occurs when the DES length-scale blending function operates in an intermediate range: neither full RANS nor true LES. The result is dissipation somewhere between the two—often worse than both. Avoid gray-area by ensuring separated regions have LES-grade resolution (Δ << turbulent length scale) and attached boundary layers have RANS resolution (y+ << 1). Refined mesh in wakes helps; check that DES operates unambiguously in each region.
How do I initialize a DES simulation?
Start from a RANS solution converged to steady state; add small turbulent perturbations to trigger instability and transition to unsteady LES behavior in separated regions. Alternatively, start from physical perturbations (vortex pair, shear layer instability). Simulation time: 10-50 eddy turnover times for detached regions to become statistically steady. For a bluff body, at least 10-20 shedding cycles before averaging.
What y+ should I use for DES mesh?
In attached boundary layers: y+ < 1 for wall-resolved DES (or y+ ~ 1 at wall, ~300-500 at layer edge). In separated wakes: y+ can be 30-100 (wall-modeled) since LES is dominant. DES doesn't use y+ as strictly as wall-resolved LES; relaxed y+ in separated regions reduces mesh cost. However, boundary layer resolution affects separation onset, so thin boundary layer meshes improve accuracy.
How long do I need to run a DES simulation?
Transient ramp-up: 20-50 flow-through times to reach periodic or statistically steady state. Statistical averaging: 50-200 shedding cycles for mean statistics, more for higher-order moments and spectra. Total time: 1000-10000 time steps typical. For a bluff body at Re=10^5, 10-50 shedding periods. Storage: save every 5-10 steps for post-processing; 3D solutions can be 1-10 GB per dump.
Sources
- Spalart, P. R., Jou, W. H., Strelets, M., & Allmaras, S. R. (1997). Comments on the feasibility of LES for wings, and on a hybrid RANS/LES approach. Advances in DNS/LES, 1, 4-8. link ↗
- Spalart, P. R., Deck, S., Shur, M. L., Squires, K. D., Strelets, M. Y., & Travin, A. (2006). A new version of detached-eddy simulation, resistant to ambiguous grid densities. Theoretical and Computational Fluid Dynamics, 20(3), 181-195. DOI: 10.1007/s00162-006-0015-0 ↗
- Gritskevich, M. S., Garbaruk, A. V., Shur, M. L., & Spalart, P. R. (2012). Development of DDES and IDDES formulations for the k-ω SST turbulence model. Flow, Turbulence and Combustion, 88(3), 431-449. link ↗
How to cite this page
ScholarGate. (2026, June 3). Detached Eddy Simulation. ScholarGate. https://scholargate.app/en/fluid-dynamics/detached-eddy-simulation
Which method?
Set this method beside its closest kin and read them side by side — the library lays the books on the table; the choice is yours.
- Boundary Layer TheoryFluid Dynamics↔ compare
- Direct Numerical SimulationFluid Dynamics↔ compare
- Large Eddy SimulationFluid Dynamics↔ compare
- Reynolds-Averaged Navier-StokesFluid Dynamics↔ compare
- Smoothed Particle HydrodynamicsFluid Dynamics↔ compare