What is Large Eddy Simulation (LES) in CFD?

shayan pakravan
Reading time: 10 min

Engineers use several methods for turbulent flow simulation. The Large Eddy Simulation (LES) is one of the most advanced of these methods. This article will explain the LES model and compare it to other methods like RANS. We will also look at how to use LES in ANSYS Fluent.

Contents

What is Large Eddy Simulation?

In CFD, we often need to simulate turbulent flow. Turbulent flow is complex and full of swirls of different sizes. These swirls are called “eddies”. Large Eddy Simulation, or LES, is a powerful turbulence model used to study these flows. There are three main ways to simulate turbulence:

Direct Numerical Simulation (DNS): This method simulates all eddies, big and small. It is the most accurate method. However, DNS needs enormous computer power and is too expensive for most engineering problems.
Reynolds-Averaged Navier-Stokes (RANS): This method does not simulate eddies directly. Instead, it models the effect of all eddies on the average flow. RANS is fast and uses less computer power. But it can miss important details.
Large Eddy Simulation (LES): The LES model is a smart balance between DNS and RANS. The main idea of an LES simulation is to directly calculate the large, energy-carrying eddies. At the same time, it uses a simpler model for the small eddies. Therefore, CFD LES gives much more accurate details than RANS, but it requires less computer power than DNS. The name Large Eddy Simulation comes from this key feature: the direct eddy simulation of large-scale turbulence. This makes the LES model a very useful tool for complex and unsteady flow problems

Figure 1: A great example of LES simulation performed by CFDLAND

What is the Difference Between LES vs RANS?

Both Large Eddy Simulation (LES) and RANS are methods used to simulate turbulent flows. However, they work in very different ways.

The RANS (Reynolds-Averaged Navier-Stokes) method is the most common approach in CFD. It is based on the idea that we can average the flow properties over time.

It uses models like k-omega and k-epsilon to calculate the effects of turbulence.
RANS is popular because it has a low computational cost.
This makes it great for many industrial problems where we need a good, quick answer.

The LES model was introduced by Joseph Smagorinsky. The main idea of an eddy simulation is different from RANS.

- LES directly simulates the large eddies in the flow.
- It models the effect of the small eddies using a subgrid-scale model.
- This gives a much more detailed and accurate picture of the turbulent flow.

Key Comparison: LES vs RANS

Choosing the right turbulence model depends on your project’s needs. LES simulation captures the time-dependent changes in a flow, which RANS models usually cannot. Because of this, CFD LES results often look more realistic for complex flows. However, this accuracy comes with a much higher computational cost.

Here is a simple guide to help you choose:

Check the Literature: First, see what models other validated studies in your field have used.
Start with RANS: If you are unsure, it is best to start with a RANS simulation. It is faster and requires less computing power.
Use LES for More Detail: If the RANS results do not match experimental data, you should then use an LES model. This will provide the higher accuracy you need for validation.

Figure 2: Flow streamlines over a backward-facing step using a RANS k-omega model. This RANS simulation shows the averaged behavior of the turbulent flow.

Figure 3: Flow streamlines over a backward-facing step with an LES model. The CFD LES simulation captures the complex, time-dependent eddy structures for a more realistic result.

What Are the Challenges of Using an LES Model?

While Large Eddy Simulation (LES) is a very powerful method, it also has some important challenges. The biggest challenge is the high computational cost. An LES simulation needs a lot of computer power and cannot usually be run on a personal computer. This high cost is due to several reasons. First, every LES model must be simulated in 3D to correctly capture the eddy structures. Second, the LES model is very sensitive to the quality of the mesh. The LES mesh requirements are very strict because the mesh must be fine enough to resolve the large eddies.

Another key challenge is choosing the correct subgrid-scale model (SGS). The SGS model accounts for the small eddies, and its performance has a big impact on the accuracy of the eddy simulation. It can be difficult to choose and validate the right SGS model for a specific problem. Furthermore, it is common to start an LES simulation using the results from a completed RANS simulation. This gives the LES a better, more realistic starting point but adds an extra step to the process.

Finally, all CFD LES simulations must be validated against real-world experimental data. This can be very difficult for turbulent flows. If the simulation results do not match the experimental results, the engineer must change the settings and run the entire, time-consuming LES simulation again. The combination of 3D domains, fine meshes, small LES time steps, and complex equations means LES produces a huge amount of data and takes a long time to complete.

Applications of Large Eddy Simulations

Because the Large Eddy Simulation (LES) is so good at capturing the details of turbulent flow, it is used in many important engineering fields. The LES model is perfect for complex problems where accuracy is critical.

Aerospace Engineering

The LES method is very useful in aerospace. Engineers use LES simulations to study the complex turbulent flow inside jet engines to understand combustion, or how fuel burns. A CFD LES can also predict the noise an airplane makes and calculate the exact drag and lift forces on the wings. These problems are time-dependent and complex, which makes the LES model the best choice for an accurate eddy simulation.

Figure 5: An image showing how the turbulent wakes and vortices from an airplane can interact with clouds. An LES simulation can accurately model these complex flow separation phenomena to better understand drag and lift forces.

You can get access to CFD Aerospace Projects.

Automotive Engineering

In the automotive industry, an LES simulation helps engineers see inside an engine to improve combustion and the mixing of fuel and air. Car companies also use the LES model to study the airflow around a car’s body. By understanding this flow, they can reduce drag and improve fuel efficiency. These are advanced simulations that require a high computational cost, so they are often done by large companies.

Energy Sector

A very common application for Large Eddy Simulation is studying wind turbines. When wind flows past a turbine, it creates a wake of turbulent flow. An LES simulation can accurately predict how this wake will affect other turbines in a wind farm. This helps engineers design better and more efficient energy systems.

Industrial Processes

Many industrial processes, like mixing chemicals or food products, involve turbulent flow. Engineers use CFD LES to optimize these mixing processes. LES is also used to design better HVAC (heating, ventilation, and air conditioning) systems by showing exactly how air moves and distributes in large rooms and buildings.

Large Eddy Simulation Equations

In LES, each field variable (ϕ), which is typically velocity or pressure, consists of two components: the Filtered Component (ϕ ̅) and the Subgrid Scale Component (ϕ ̀). So for each field variable such as velocity (u) and pressure (p):

$\phi=\overline{\phi} +\grave{\phi}$

$u=\overline{u} +\grave{u}$

$p=\overline{p} +\grave{p}$

The Filtered Component in LES represents the resolved, large-scale motions of the flow. It captures the dynamics of larger eddies and structures that are directly computed in the simulation, providing a detailed representation of the dominant, energy-containing turbulent features. The filtering operation in LES is typically expressed in the following integral form:

$\overline{\phi} \left( x,t \right) = \int_{- \infty}^{\infty} \phi \left( r,t \right) G\left( x-r \right) dr$

Where ϕ ̅ is the filtered variable, which usually is the velocity of pressure, ϕ is the original variable, x and r denote spatial coordinates and G is the filter kernel, which defines the characteristics of the filter, such as a Gaussian, box, or top-hat filter.
The Subgrid Scale (SGS) equations include the effect of that part of the turbulent flow that is smaller than the mesh cells into the simulation. There are various SGS models such as the Smagorinsky model, the Algebraic Dynamic model, Dynamic Global-Coefficient model, Localized Dynamic model, Wall-Adapting Local Eddy-viscosity (WALE) model, RNG-LES model, and various structural modeling approaches. Choosing the right option is done based on the user experience of the type of problem and the help of the previous simulation results. It should be noted that the use of SGS reduces the computational cost and the difference between LES and DNS is more in the use of SGS.
Finally, by selecting the desired SGS model and finding ϕ ̀, ϕ is calculated, and the desired field component variable is obtained and placed in different equations, such as continuity and Navier-Stokes.

Choice of the Subgrid-Scale (SGS) Modelling

To choose the right SGS model, different parameters should be considered. WALE or the Dynamic Smagorinsky model are suitable for starting simulation in many applications. The Dynamic TKE model is suitable for very non-equilibrium flows and reacting flows. In simulation, in many cases the fluid inlet boundary conditions are far from the experimental reality, in the Dynamic Smagorinsky and Dynamic TKE models, those simplified boundary conditions are less effective and the flow quickly approaches the real conditions. DDES or similar hybrid methods are recommended for streams with very high Re number. WMLES, a 0-equation algebraic Detached Eddy Simulation model, is suitable for scenarios with high near-wall cell aspect ratios. Each model has its advantages and disadvantages, and its choice depends on the type of problem and computational ability.

Click to access the CFD Heat Transfer Projects.

Large Eddy Simulation for Incompressible Flows

Whether fluid and flow are compressible or not, LES can be used to simulate them. The simulation of compressible flows is more complicated and has many difficulties in terms of convergence. It also requires a lot of computing power. Simulating such flows requires expertise and a lot of experience. It is very difficult to create and find experimental results to compare with simulation results in these cases.

Detached Eddy Simulation

Detached Eddy Simulation (DES) is a hybrid method that combines RANS and LES. Depending on the design and programming choices, RANS and LES methods, which have better performance, are used in different areas of the fluid flow. For example, in the area near the wall and the boundary layer, RANS is usually employed because of its ability to simulate small vortices. In the separation region, LES is used due to its capability to simulate the details of the turbulent flow.

Large Eddy Simulation by ANSYS Fluent

ANSYS Fluent is a CFD simulation software based on finite volume that has been used for several decades in various engineering fields and has shown brilliant performance. As can be seen in the figure below, Fleunt offers many methods for simulating turbulent flow, and it is possible to change the details and adjust each method. Rest assured that their algorithms and equations are the most advanced in this field. Whether you choose RANS or LES models, this software performs the simulation well.

In ANSYS Fluent, the viscous model setting window for LES includes extensive and detailed options for configuring the subgrid-scale (SGS) model, model constants, and user-defined constants. This setup is crucial for accurately capturing turbulent flows at varying scales.

Our experts at CFDLAND have done many simulations in the field of turbulent flows with RANS and LES and have a lot of experience. You can order your simulation CFD projects in the ORDER CFD PROJECT section with confidence in the quality of our work. Maybe one of them will help you.

Conclusion

In conclusion, LES is a powerful method for turbulent flow simulation, especially suitable for examining flow details. Compared to RANS, it simulates many more characteristics of the flow, but it has a much higher computational cost. For LES simulations, it is recommended to use ANSYS Fluent, this software has proven its capabilities in this field so far.