Skip to Main Content

      

ANSYS BLOG

September 29, 2022

Ansys Fluent: A History of Innovations in CFD

In the early 1980s, from the buildings of Sheffield University with contributions from multiple personalities, Ansys Fluent became the first commercial computational fluid dynamics (CFD) software to have a graphical user interface and workflow rather than a command-line input. The popularity of Fluent increased year after year, along with its adoption in the industry.

In May 2006, Fluent Inc. was acquired by Ansys. Since joining the Ansys family, we have relentlessly pushed the envelope of performance, accuracy, and productivity of Fluent, bringing cutting-edge innovations that have helped engineers overcome the most challenging design obstacles imaginable when dealing with fluid dynamics problems.

Are you a new customer that is not yet familiar with all the Fluent innovations that are available to you? Did you make a different vendor choice and want to learn more about what you’ve been missing? Let’s travel back in time through some of the most innovative Fluent features.

2014: GPU-Accelerated Offloading and Adjoint Solver

While graphics processing unit (GPU) technologies are in the spotlight today, the concept of using GPUs as CFD accelerators was introduced in Fluent in 2014 with the NVIDIA AmgX solver. Bear with us in this time-travel scenario, because in 2022, Fluent will be the first commercial CFD software to introduce an unstructured finite volume, fully resident multi-GPU solver, overcoming the offloading limitations and disrupting the market for CFD simulations.

The Adjoint solver was first introduced in Fluent in 2014 to revolutionize insights from CFD simulations by using the adjoint sensitivities to drive intelligent design changes that are not intuitive to a designer. Since its inception, the reliability and usability of the Adjoint solver has consistently evolved into a comprehensive product optimization framework.

Optimize designs through automated shape optimization using the Adjoint solver, introduced in 2014.

2016: Fluent Breaks a Supercomputing Record of 170,000 Cores

Research and development on high-performance computing (HPC) to improve the parallel scalability of Fluent has been — and still is — a key focus area. In 2016, Cray Inc. and the High Performance Computing Center (HLRS) of the University of Stuttgart set a new supercomputing world record by scaling Fluent to more than 172,000 computer cores, enabling organizations to create innovative and groundbreaking complete virtual prototypes of their products faster and more efficiently than ever.

2017: PUMA Adaptation and SBES Turbulence Model

Another cutting-edge innovation for a commercial CFD software was the introduction of patented polyhedral unstructured mesh adaptation (PUMA) in 2017. This adaption technique automatically and dynamically refines the mesh to track fine details in the flow. As a result, engineers can get the accuracy they need, where they need it, to capture simulation details while leaving courser mesh elsewhere for faster solve times.

Accelerate solve times by automatically refining a mesh to resolve fine details while leaving coarser mesh in place using polyhedral unstructured mesh adaptation (PUMA), introduced in 2017.

The same year, the stress-blended eddy simulation (SBES) turbulence model overcame inherent problems with hybrid RANS-LES simulations. Large-eddy simulation (LES) is impractically expensive in the near-wall region while Reynolds-averaged Navier-Stokes (RANS) models are well-suited for wall boundary layers. Hybrid RANS-LES models provide efficient near-wall modeling using a RANS turbulence model and a high-fidelity solution away from walls using an LES approach, but switching between the two traditionally suffered from mesh dependency and false mesh-induced flow separation. The SBES model developed by Ansys introduced a unique and proprietary shielding function to overcome these problems, resulting in a reliable and efficient scale-resolving turbulence model that engineers can trust.

2018: Mosaic Meshing and Spray Break-up Model

Transitioning among varying types of mesh elements in complex geometries while retaining mesh quality has long been a major meshing challenge, particularly when transitioning from boundary layer prisms to hex elements away from walls. In 2018, Fluent introduced Mosaic meshing to address this challenge by automatically connecting different types of meshes with high-quality polyhedral elements. The resulting simulations were dramatically faster with greater solution accuracy while using less RAM.

em modeling

Ferrari Competizioni GT increases simulation productivity by 300% using Mosaic meshing technology, introduced in 2018.

The same year, the spray breakup model was the first commercial implementation of volume of fluid (VOF) to discrete particle method (DPM) transition for multiphase simulations, such as spray breakup. In this model, distinct droplets in the VOF simulation are detected and replaced by DPM (i.e., mass-point) particles, and the mesh is coarsened accordingly.

Efficiently simulate spray breakup using the volume of fluid (VOF) to discrete particle method (DPM) hybrid multiphase model, introduced in 2017.

2019: GEKO Turbulence Model

In 2019, the Ansys turbulence team introduced a generalized k-ω turbulence (GEKO) model with tunable coefficients. GEKO’s tunable coefficients can be adjusted over a wide range to match specific physical effects while maintaining the underlying model calibration. No other turbulence model has enough flexibility to match experimental data over a large set of test cases — even if tuned by a turbulence model expert — while GEKO’s coefficients provide that flexibility. The GEKO model can even be tuned to scale-resolving simulation results, such as those using the SBES turbulence model. The GEKO turbulence model is exclusively available in Ansys fluids applications.

2020: AIAD Transition Model and 3D Electrochemistry for Batteries

In 2020, Fluent was the first commercial software to showcase an Eulerian multiphase transition method based on the algebraic interfacial area density (AIAD) approach. This method is suitable for a range of applications such as loss of coolant scenarios in pressurized-water reactors, with strong agreement with experimental data compared to alternative methods.

Gas‒liquid flow with droplet entrainment and re-absorption using AIAD transition, introduced in 2020

The same year, Fluent introduced the transient simulation of lithium (Li)-ion transport during battery charging/discharging providing a complete commercial solution for 3D electrochemistry of Li-ion batteries.

2021: Bidirectional VOF-to-DPM-to-EWF and AI/ML Turbulence Tuning

As previously mentioned, the VOF to DPM transition was first introduced in Fluent in 2018 with the spray breakup model. In 2021, Fluent made this transition bidirectional, supporting the inverse DPM-to-VOF transition and completing it with the transition to Eulerian wall film (EWF). In the bidirectional VOF-to-DPM-to-EWF model, DPM particles falling onto a free liquid surface transition back into VOF formulation and the mass-point particles are replaced by mesh-resolved VOF liquid.

Efficiently simulate breakup, pooling, and thinning of multiphase flows using the bidirectional VOF-to-DPM-to-EWF feature, introduced in 2021.

The same year, Fluent introduced AI/ML artificial intelligence (AI)/machine learning (ML) turbulence tuning, in which the GEKO coefficients are tuned with ML algorithms rather than manually. This enables engineers to use a scale-resolving turbulence model such as SBES to generate a high-fidelity solution. They then can use ML to tune the GEKO coefficients throughout the 3D flow field so that subsequent design iterations can use the much-faster GEKO model while retaining accuracy that approaches the scale-resolving solution.

2022: Live-GX Multi-GPU Solver and PyFluent

As previously mentioned, Fluent has been a trailblazer in the use of GPU technology for simulation, and in 2022 it took it to a new level with the introduction of a native multi-GPU solver. This brand-new multi-GPU solver provides many benefits for steady-state and transient CFD simulations, including reducing simulation solve time, hardware costs, and power consumption with the same accuracy of the CPU solvers and without all the limitation of GPU offloading we previously mentioned.

Six high-end GPUs provide the same performance as more than 2,000 CPUs when using the native multi-GPU solver, introduced in 2022.

Finally, Fluent introduced PyFluent, an open-source library to access all Fluent commands from pre- to post-processing using Python. PyFluent is designed to combine a robust community of peers, a programming language such as Python that encourages reuse, and Ansys' state-of-the-art simulation stack to create endless possibilities.

What About Future Innovations?

Fluent has continuously introduced cutting-edge innovations that have changed the way CFD simulations are performed and have set the standard for the industry.

Our research and development commitment is to keep innovating so that our customers can employ state-of-the-art software to relentlessly push the envelope of performance, accuracy, productivity, and sustainability in an unprecedented way.

Check out our infographic highlighting the groundbreaking innovations introduced in Fluent over the last few decades, and don’t forget to keep an eye on our latest innovations.

Eager to test some of these innovative features Fluent has developed over the years? Request for free 30-day trial now.

See What Ansys Can Do For You

Contact us today

* = Required Field

Thank you for reaching out!

We’re here to answer your questions and look forward to speaking with you. A member of our Ansys sales team will contact you shortly.

Footer Image