Computational modelling and simulation has emerged as a crucial mode in engineering design and scientific investigation complementing theory and experiments. It involves mathematical and physical modelling, numerical analysis, data processing and visualisation.
The exponential growth of computer power, decreasing costs of hardware and optimization of numerical algorithms had a huge impact on the application of simulation to industrial processes and science.
In fluid dynamics this approach has been applied to the study of turbulent flows and transport phenomena associated with it (heat and mass transfer, combustion, particle-tracking).
The key issue for Computational Fluid Dynamics (CFD) is modelling: turbulence is an intrinsically non-linear phenomenon, which involves a multi-scale behaviour that in order to be simulated directly requires computational costs
far beyond the ones available in the next future – as a matter of fact Direct Numerical Simulation of turbulent flows (DNS) is nowadays used in academy as a high-level research tool and still confined to low-Reynolds application in simple geometries.
Large Eddy Simulation (LES) – a direct simulation of the large-scale motions – has become a new frontier of industrial research but it’s still confined to simple geometries and moderately high-Re flows, capable of revealing inner insight of physical phenomena and recover a broad part of turbulence spectrum.
The industrial approach for CFD is (U)-RANS-(Unsteady) Averaged Navier-Stokes – have been used in the last decades in order to recover information on the performances of a broad variety of engineering applications and physical processes.
(U)-RANS is based on modelling and modelling is crucial in order to cut computational and post-processing times, but it’s also tricky: industrial packages for CFD often contain over-simplified models of real physics, suitable for a broad variety of applications and computationally robust in order to be user-friendly handed by operators which have to be handled with care.
DNS, LES, URANS are currently used to study fluid flows as well as heat transfer phenomena, combustion, multi-phase flows, particle dispersion.
Advantages of CFD
- DNS and LES can be used to investigate the dynamics of physics phenomena that are not possible to explore in an experimental way – or too expensive to investigate in that way. A smart campaign with CFD can shine a light on small scale phenomena of turbulent flows in simple geometries
- URANS and 2nd generation URANS approaches (also known as hybrid LES/RANS models) can be successfully employed to implement virtual-prototyping procedures in industrial design, in order to drastically cut development time and costs
- Thermo-Fluid Dynamics is everywhere in modern technology! From turbomachinery to aerospace engineering, to cooling system for industry or laptops, to the study of environmental flows and bio-medics, correct design and extremely fine optimisation of heat and mass transfer phenomena is mandatory to implement efficiency
Drawbacks of CFD
- Direct Numerical Simulation of flow field is far beyond the computational power available in the next decades; even when DNS is possible (low-Reynolds flows, e.g. blood simulation) post-processing of data requires time and skills
- Large Eddy Simulation is the new frontier of high-resolution CFD, but still requires massive parallel facilities
- URANS computations require skilled personnel, able to choose the proper model among dozens
Aires Tech services for CFD
- In house developed and reliable codes with Finite Element and Finite Volume discretisation
- Long-experienced personnel who have been working of most of the branches of CFD as turbulence modelling, heat and mass transfer, combustion, super-sonic flows, particle-dispersion and particle-tracking, particle deposition, fouling
- A wide knowledge of mesh generation software, numerics, parallel algorithms, postprocessing techniques and software.