< Back to Blog

April 2019

,

Luma Bendini

What is a High-Fidelity CFD simulation?

Computer aided engineering is helping many industries to optimise process and products. But with an enormous number of new tools and technologies coming up to the market every year, the challenge now is to know how to choose the proper solution. In this article we will explore the concept of high-fidelity CFD simulations.

It's all about the method: RANS X LES

These are images from research conducted in 2016. The goal was to analyse the dispersion of pollutants associated with traffic emissions on a busy street in downtown Rio de Janeiro.

When using computational tools to simulate traffic emissions, the researchers applied two different methods of simulations: Reynolds-averaged Navier-Stokes (RANS), low-fidelity, and Large Eddy Simulation (LES), high-fidelity simulations. Results showed that small vortices carrying the pollutants are captured only by the LES model. While the RANS model only shows the main characteristics of the flows, providing less reliable information.

That happens because LES methods produce a time series of snapshots of the actual flow field and RANS methods, on the other hand, compute a statistical mean field of the flow, with details of the transient flow field being averaged out. "As a consequence, LES provides a much richer data set from which transient effects can be analysed. RANS relies on a number of model parameters that have to be tuned to the problem at hand, which makes it unreliable in many cases. On the other hand, LES puts a much higher demand on the mesh resolution which makes it more expensive”, explains Professor Johan Hoffman, co-found at Ingrid Cloud.

Does it mean that high-fidelity simulations are out of reach? Or too expensive? Not anymore. New software-as-a-service platforms such as Ingrid Cloud allow anyone to access accurate information in an easy-to-use and affordable way. But how reliable is a high-fidelity flow simulation? 

The results of benchmark case studies conducted to compare Ingrid Cloud with a physical wind tunnel or other traditional simulation codes are outstanding. Find some benchmarks and other scientifically validate data here.

Professor Hoffman explains: Low-Fidelity x High-Fidelity 

johan-profile"The notion of high-fidelity CFD simulations refers to simulations where the actual fluid flow is recreated in the computer as a virtual experiment. Data from this virtual experiment can be processed to visualise the fluid flow in movies and snapshot images, and for quantitative measurement of forces on surfaces, and to probe velocities and pressures in the fluid volume. Low-fidelity CFD simulations have a lower computational cost, but do not provide as detailed or as reliable information. Instead only a statistical mean representation of the fluid flow is computed, with high uncertainty in the prediction of many features, such as flow separation" - Professor Johan Hoffman, R&D and co-founder at Ingrid Cloud