• Skip to main content
  • Skip to search
  • Skip to footer
Cadence Home
  • This search text may be transcribed, used, stored, or accessed by our third-party service providers per our Cookie Policy and Privacy Policy.

  1. Blogs
  2. Computational Fluid Dynamics
  3. Introducing Fidelity CharLES - Accelerated, Accurate Large…
JoshuaS
JoshuaS

Community Member

Blog Activity
Options
  • Subscribe by email
  • More
  • Cancel
CDNS - RequestDemo

Try Cadence Software for your next design!

Free Trials
Fidelity CharLES
CFD
large eddy simulation
Computational Fluid Dynamics
Fidelity CFD
Meshing

Introducing Fidelity CharLES - Accelerated, Accurate Large Eddy Simulation

30 May 2023 • 3 minute read

There’s an old joke in the CFD community: “Large eddy simulation is ten years away from being useful, just like it was in 1977.” Of course, that’s never been entirely true. Seventeen years ago, I did an LES analysis of a tiny wedge of an axisymmetric intake at an internship. It wasn’t a large mesh, and it took three weeks to run the simulation on my 6-core workstation, but it gave deep insights into the flow we were studying, particularly how and where transient vortices were forming and how they affected performance. My mentor remarked that even though the run took weeks, they’d been banging their head on the problem for almost a year, so the answer we got was well worth it.

Ever since that internship, I’ve been interested in LES. I even went on to do a dissertation in turbulence modeling for LES. The approach my adviser and I pursued then didn’t exactly catch the world on fire, but one really important conclusion has stuck with me over the years: the traditional eddy viscosity approach, which is based on physically unrealistic assumptions, introduces far too much artificial diffusion into the fluid domain for LES to work particularly well as an engineering tool. Often times, a well-calibrated RANS model will outperform a traditional LES model at predicting bulk engineering quantities, like lift and drag. Viable LES must be able to resolve length scales down to the order of the grid resolution itself.

When I first saw the Cascade Technologies team present their CharLES flow solver at NVIDIA’s GTC in 2021, I knew I was looking at something special. Crisp, highly resolved wakes trailing off an airplane wing looked like nothing I’d seen outside of academic experiments. It looked almost like DNS! None of the usual smearing and distorted flow topology that is the hallmark of eddy viscosity was visible. Whatever these people were doing, I knew it had to be something different.

Figure 1: This flow over a wing with high-lift devices illustrates that Fidelity CharLES is not your grandfather's LES

Soon after joining Cadence, I was excited to discover we were acquiring this fascinating technology. As I’ve gotten to know the Fidelity CharLES team and understand their software, it’s only confirmed my initial impression that this is the LES the industry has been waiting for. Instead of trying to stabilize the flow and model the effects of turbulence as nonphysical viscosity, they went back to the fundamental principles of physics and numerics. They carefully derived a scheme that doesn’t just guarantee stability and accuracy, even on irregular meshes, but also rigorously enforces conservation of entropy and kinetic energy, meaning no new entropy and no losses in energy are artificially introduced by the numerical scheme. Add to that a polyhedral meshing scheme that guarantees high-quality cells, and you have LES simulations that are unrivaled in accuracy.

Figure 2: Reacting jets with flux-limited central differencing (left) versus kinetic energy & entropy preserving scheme (right). Meshes are identical.

To top it off, Fidelity CharLES has supported GPU acceleration since early on. The software has been optimized to consume as little memory as possible and scales linearly to hundreds of GPUs across dozens of nodes. We’re launching on all platforms commonly available today – NVIDIA and AMD GPUs and x86 CPUs. With modern high-performance hardware, you can do a lot more than that small wedge I did almost two decades ago and get it done much faster than three weeks – you can get an industrial simulation done in less than a day on a typical HPC machine.

With the launch of Fidelity Charles, we at Cadence have made fast, high-resolution large eddy simulation accessible and available to design engineers who need to solve real problems and do a deep analysis of physically realistic flow structures. There’s nothing else like it, and I can’t wait to see the new things users will do with it. Already, Fideity CharLES is being used to design jet engine combustors, high-lift aerodynamics, and vehicle external aerodynamics. This is the real deal - fast, accurate LES, ready for design and optimization.


Live Webinar Thursday June 8th

One of the crucial technologies underpinning Fidelity CharLES is clipped-Voronoi Diagram mesh generation. This approach for CFD mesh generation was explicitly developed to produce high-quality meshes for scale-resolving simulations like new wall-modeled large eddy simulation (WMLES). Join this webinar to learn more:


Attachment:

  • 3426.jaxa-wing.png-1280x0.png
  • View
  • Hide

CDNS - RequestDemo

Have a question? Need more information?

Contact Us

© 2025 Cadence Design Systems, Inc. All Rights Reserved.

  • Terms of Use
  • Privacy
  • Cookie Policy
  • US Trademarks
  • Do Not Sell or Share My Personal Information