In the last chapter a family of turbulence modeling approaches referred to as Scale-Resolving Simulations was introduced. These models aim to resolve some portion of the turbulent eddies in the flow rather than modeling all turbulence via closure models. The summary includes LES, wall-modeled LES, hybrid methods like DES and SAS, and notes on where and how each method can be applied

Motivation

Conventional RANS models simulate only the mean flow, modeling all turbulence effects through closure models. While this is computationally efficient, it limits accuracy in flows with strong separation, unsteadiness, swirl, or large coherent structures. In contrast, Scale-Resolving Simulations (SRS) aim to resolve the larger energy-carrying eddies while modeling only the smaller ones. This leads to more physically realistic unsteady flow fields, at the cost of increased computational effort.

SRS approaches are used where RANS fails to capture critical dynamics, such as wake behavior, noise prediction, vortex shedding, and massively separated flows. The goal is to strike a balance between resolution and computational cost.

Large Eddy Simulation (LES)

LES is the reference method for scale-resolving modeling. It resolves large-scale turbulent structures directly and uses a subgrid-scale (SGS) model for the smaller, less energetic eddies that the mesh cannot resolve.

Key features:

High spatial and temporal resolution required
Accurate for unsteady, complex flows
Assumes energy cascade from large to small eddies (inertial range)
Requires very fine meshes near walls (especially for wall-resolved LES)

Due to its high cost, wall-resolved LES is typically limited to academic studies or high-priority industrial cases.

Wall-Modeled LES (WMLES)

To reduce the near-wall resolution requirements of LES, WMLES introduces a model for the near-wall region and uses LES in the outer flow. This avoids resolving the entire boundary layer and reduces the total cell count.

Features:

Suitable for high Reynolds number flows
Requires accurate wall model (often based on logarithmic law)
Still needs good mesh quality in outer regions

WMLES enables LES in external flows like jets, airfoils, and vehicle aerodynamics without prohibitive mesh sizes.

Embedded LES (ELES)

ELES combines LES and RANS by using RANS in most of the domain and LES in a specific region of interest (e.g., a recirculation zone or wake). A RANS-LES interface is defined, and fluctuations are synthesized at the boundary to allow proper LES development.

Use cases:

Local unsteady phenomena (jet breakup, diffuser separation)
Reduces cost by avoiding full-domain LES
Requires careful interface setup and transition handling

ELES is particularly effective when only a part of the domain exhibits complex unsteady behavior.

Hybrid RANS-LES Models

Several hybrid models aim to blend RANS and LES dynamically within the domain, without needing a user-defined LES zone.

Detached Eddy Simulation (DES)

Uses RANS near the wall and switches to LES in separated regions
Originally developed for external aerodynamic flows
Based on a length-scale switch that limits turbulence model length scale in regions where the grid can resolve eddies

Delayed DES (DDES)

Improves DES by suppressing premature switch to LES near the wall
Adds a shielding function to preserve RANS in attached boundary layers

Improved DDES (IDDES)

Further improves shielding and stability
Incorporates wall-modeling capabilities
Better suited for complex geometries and general-purpose use

Stress-Blended Eddy Simulation (SBES)

Blends RANS and LES based on turbulence stress contributions instead of length scales
Offers smoother transitions and reduced dependence on mesh resolution

These hybrid models provide a compromise between LES accuracy and RANS robustness, and are increasingly used in industry.

Scale-Adaptive Simulation (SAS)

SAS is not a hybrid RANS-LES model but an unsteady RANS method that adapts its turbulence scale based on flow instabilities. It modifies the turbulence model source terms to become more responsive to resolved fluctuations when the grid allows it.

Key points:

Automatically adapts to resolve unsteadiness in separated flows
No explicit subgrid-scale model
Compatible with standard RANS meshes
Useful for vortex shedding and transient wakes

SAS is a practical alternative to DES for moderate unsteady flows when mesh is not fine enough for full LES.

Application Notes

All SRS methods require unsteady solvers, small time steps, and good temporal resolution
LES, WMLES, and hybrid models need fine meshes in at least some regions
RANS-to-LES interfaces must be treated carefully to avoid numerical artifacts
Use second-order time discretization and bounded central differencing for LES regions
Ensure sufficient time averaging for meaningful statistics

Summary

Scale-Resolving Simulation methods offer a range of options for capturing unsteady turbulent structures that RANS models cannot represent. Full LES resolves the large eddies and offers high fidelity, but comes at high computational cost. Hybrid models like DDES and SBES introduce localized resolution where needed, while SAS provides a more robust RANS-compatible alternative.

Choosing the right model depends on the flow features, available mesh, computing resources, and whether transient flow details are critical to the analysis.

Turbulence Chapter 5: Scale-Resolving Simulations (SRS)