Computational Fluid Dynamics (CFD) is a specific category of numerical analysis which is based on solving the Reynolds Averaged Navier Stokes (RANS) equations in a given volume of fluid. Variables that characterize the flow -velocity, pressure, temperature, concentration… - are resolved in the entire volume, resulting in a comprehensive 3D dataset which is especially suited for understanding flow physics and visualisation of complex flow patterns.
CFD as a research tool
By providing 3D data in the whole computational domain, CFD analysis is especially suited for analysing flow physics. It allows understanding why new technologies based on fluid mechanics do or do not work as expected, giving solid bases for further development of innovative solutions.
CFD as a design tool / performance analysis tool
Setting up a CFD simulation requires choices regarding time and space discretization methods as well as multiple numerical schemes. For some selected applications, our researchers have thoroughly validated those choices against large and trusted experimental dataset. Those validated CFD models are used by our team and by our customers to evaluate design performances in an effective way.
CFD as a database provider
In some cases, CFD is too CPU demanding (and therefore too expensive or too time consuming) to be used as a design tool and faster approaches such as empirical methods are preferred by designers. For empirical methods to be accurate, their underlying models must be derived from a large set of accurate data. Validated CFD models can be used as an alternative to experiments to generate such large database.
CFD versus model testing / experiments
CFD and model testing are sometime perceived as competing approaches. But numerical and experimental approaches each have their strengths which, once combined, lead to highly effective work processes. While CFD allows for full scale analysis of large test matrices and deep understanding of flow physics, model testing provide essential validation data and can capture complex flow phenomena that are beyond the range of applicability of numerical models. It is for that reason that we answer each of our customers request with a suggested scope of work that combines both approaches in a unique way.
Both commercial and open source CFD software are used in our projects. Here again, each software has its own advantages, and we always select the most suitable tool to provide high quality CFD analyses.
In addition, CFD is increasingly used in combination with other numerical methods such as optimization engines for improving performances of a studied system, structural analysis solvers for including fluid-structure interaction effects (such as body deformations), or even potential flow solvers when part of the solution can be solved externally by simpler and faster models than CFD.
Reliable hardware is an essential factor for conducting high-quality CFD analysis. Our CFD simulations are conducted on SINTEF's HPC cluster "Unity", established in 2017, which features over 3500 cores and InfiniBand connection. This facility grants us access to a stable computing resource with high performance computational nodes and little to no queuing time.