How to Capture Dynamic Flow Features: URANS, DES & LES Methods
Author: 42CFDLab, proofread by AI
Date and Version: 2026.06.29; version 1
📋 Key Takeaways
URANS is the workhorse – Fast, robust, and well suited for flows dominated by large-scale unsteadiness, where turbulence is fully modelled.
DES offers the best compromise – Uses RANS near walls and LES in separated regions. Ideal for massively separated flows at high Re.
LES provides the highest fidelity – Resolves the energy-containing eddies. Requires fine isotropic grids and small time steps. Reserved for research and critical validation.
Method selection – The optimal choice depends on the flow nature, computational resources, and the level of detail required. Each method offers a different balance between computational cost and accuracy.
Many industrial flows are inherently unsteady. Vortex shedding behind bluff bodies, wake interactions, rotating machinery, and pulsating flows all exhibit dynamic features that steady RANS simulations smooth out or completely miss. To capture these transient phenomena, engineers must turn to scale-resolving and time-dependent methods.
Choosing between URANS (Unsteady Reynolds-Averaged Navier-Stokes), DES (Detached Eddy Simulation), and LES (Large Eddy Simulation) depends on the flow physics, available computational resources, and the required level of details. Each approach sits at a different point on the cost-accuracy spectrum.
URANS Unsteady Reynolds-Averaged Navier-Stokes
URANS extends the steady RANS approach by time-marching the Reynolds-averaged equations. While the turbulent fluctuations are still fully modelled (via a turbulence closure), the method resolves the large-scale, low-frequency unsteadiness of the mean flow. This makes it suitable for problems where the dominant dynamics occur at scales much larger than the turbulent eddies.
Strengths & Weaknesses
Practical Guidelines
- Grid Resolution: Follow standard RANS best practices. Near-wall y⁺ ≈ 1 for low-Re models, or y⁺ > 30 with wall functions. Streamwise and spanwise stretching should be moderate to avoid numerical diffusion of the large-scale unsteadiness.
- Time Step (Δt): Must resolve the characteristic frequency of the unsteady phenomenon (e.g., Strouhal number for vortex shedding). A general rule is Δt = T / 50 to T / 200, where T is the period of the dominant fluctuation. The Courant number (CFL = U·Δt/Δx) is often recommended to be below 5–10 for URANS to maintain accuracy, though lower is better.
- When to use: Early design scoping, parametric studies, and flows with a clear dominant frequency (e.g., wake behind a cylinder).
DES Detached Eddy Simulation
DES is a hybrid RANS-LES approach. It uses a RANS model (typically k-ω SST) in the near-wall region where the boundary layer is attached and the turbulent eddies are small and costly to resolve. Away from the wall, in separated and free-shear regions, the model switches to an LES-like mode, resolving the large turbulent structures. This makes DES exceptionally efficient for high-Reynolds-number flows with massive separation.
Strengths & Weaknesses
Practical Guidelines
- Variants: Variants such as Delayed DES (DDES) and Improved DDES (IDDES) improve robustness through shielding functions; in particular, IDDES reduces transition sensitivity and mitigates grey-area issues, though careful grid design and validation are still required.
- Grid Resolution: Near-wall grid must be fine enough for RANS (y⁺ ≈ 1). In the separated/LES region, the grid must be isotropic enough to support resolved eddies. A good starting point is Δx⁺ ≈ Δz⁺ ≈ 100–200 in the streamwise and spanwise directions, and Δy⁺ ≈ 1 at the wall. The LES region requires cells roughly cubic (Δx/Δy ≈ 1–3).
- Time Step (Δt): Must resolve the turbulent eddies in the LES region. Aim for CFL < 1 based on the local velocity and grid spacing in the wake. This ensures that the temporal resolution does not dampen the resolved eddy content.
- When to use: The default choice for bluff-body aerodynamics, wake interactions, and high-Re separated flows where URANS is insufficient but wall-resolved LES is too expensive.
LES Large Eddy Simulation
LES directly resolves the large, energy-containing turbulent eddies while modelling only the small, universal sub-grid scales (SGS). By resolving the anisotropic eddies that dominate momentum and heat transfer, LES provides the most detailed insight into transient flow physics, including pressure fluctuations, acoustic sources, and mixing.
Strengths & Weaknesses
Practical Guidelines
- Grid Resolution: Grid cells must be small enough to resolve the inertial subrange. A common rule of thumb: at least Δ ≈ 0.1–0.2 L where L is the integral length scale. For wall-resolved LES, near-wall cells require Δx⁺ ≈ Δz⁺ ≲ 50–100 and Δy⁺ ≈ 1. Cells should be as isotropic as possible (aspect ratio < 3) in the core flow.
- Time Step (Δt): Strictly constrained by the CFL condition. Use CFL < 0.5–1.0 based on the local convective velocity and grid spacing. This ensures that eddies are advected smoothly across the grid without numerical damping.
- When to use: Fundamental research, aeroacoustics, detailed validation studies, and problems where accurate prediction of unsteady loads, noise, or mixing is critical and budget permits.
Practical Comparison
| Feature | URANS | DES (Hybrid) | LES |
|---|---|---|---|
| Turbulence Modelling | Fully modelled | RANS near walls / LES away | Resolved large scales + SGS model |
| Grid Resolution | Moderate | Refinements in LES regions | Extremely fine, isotropic |
| Time Step (CFL) | |||
| Boundary Layer | Modelled by RANS | Modelled by RANS | Resolved (or modelled near-wall) |
| Computational Cost | Low-Moderate | Moderate–High | Very High |
| Industrial Readiness |
Conclusion & Practical Selection Criteria
There is no universally "best" method—only the most appropriate one for your particular scenario. The choice between URANS, DES, and LES is a strategic decision driven by the specific flow physics, the engineering objective, and the available resources.
URANS is the pragmatic choice for high-Reynolds-number flows where the primary interest is the time-averaged or phase-averaged mean field, and where the unsteadiness is dominated by large-scale periodic motions (e.g., vortex shedding behind a cylinder). It delivers acceptable accuracy at minimal cost, making it ideal for parametric studies, early design exploration, and applications where turnaround time is critical.
DES is the go-to method for bluff body turbulence, such as automotive aerodynamics. It offers a compelling compromise between cost and accuracy — resolving the large-scale wake structures while avoiding the prohibitive expense of wall-resolved LES. When URANS proves overly dissipative but LES is not feasible, DES is typically the most effective industrial solution.
LES is essential for accurately resolving the turbulent spectrum but is currently practical only for low- to moderate-Reynolds-number flows due to its high computational cost. It is the method of choice for aeroacoustics, fundamental turbulence research, and critical validation campaigns where the cost (both computational and human) is justified by the need for high-fidelity unsteady simulations.
Ultimately, selecting the right method requires weighing the required accuracy against computational cost, project timeline, and the inherent flow physics.
Need Help Capturing Dynamic Flow Features?
At 42 CFD Lab, we specialise in helping clients select and implement the right turbulence modelling strategy for their specific application. Whether you need a fast URANS setup, a robust DES grid, or a high-fidelity LES simulation, our team can guide you through the entire process.
Drop us a message →