About CFD



British physicist Sir Horace Lamb is reported to have humorously said in 1932, shortly before his death: "I am an old man now, and when I die and go to heaven there are two matters on which I hope for enlightenment. One is quantum electrodynamics, and the other is the turbulent motion of fluids. And about the former I am rather optimistic."

The turbulent motion of fluids remains nowadays a scientific challenge. Yet turbulent fluid flow is the essence of many biological, environmental and industrial processes. To understand, control or improve such processes, computer simulation of fluid flow is gaining acceptance as an essential component in the engineer’s toolkit. Compared to experimental fluid dynamics, Computational Fluid Dynamics (or CFD) offers, in general, many advantages. Firstly, it is usually more economical and faster; secondly, it provides a complete set of results, even in circumstances where experimental measurements would be difficult (eg, the shear stress on the wall of a human artery); and, thirdly, it in principle allows to investigate situations for which experiments would be impossible (eg, catastrophic episodes).

However, there are virtually no limits to the number-crunching hunger of CFD. Taken to the extreme, disparity of time and length scales involved in turbulent flow means that a fully-resolved calculation of turbulence is prohibitively expensive, and will remain so for many years to come. Turbulence-research scientists estimate, that, at the current pace of computer-technology progress, a real-turbulence calculation for a simple problem may be doable by approximately year 2020, and last 5 years. For practical reasons, the need for resolving the full range of scales is usually overridden by use of models (such as turbulence models, combustion models, or multiphase models).

Yet ‘modelled’ fluid dynamics are also very computer-time demanding. Even if the smaller details of turbulence can be dispensed with, the problem to be investigated will often be very geometrically complex, thus introducing again a spread of length scales which requires the use of fine computational meshes and powerful computers. Regarding the hardware cost, it is well known that the computer processing power increases approximately a 100-fold each decade, and further that parallel-calculation advances have brought about a 1000-fold increase per decade. This is however of little comfort for the mainstream CFD practitioner, because they are nevertheless continually pressed, for competitiveness reasons, to perform the best simulation that money can buy at any given time. Recent technological trends, like the use of clusters of loosely-coupled Linux nodes (often called ‘Beowulfs’) are a reasonable price/performance compromise for high-performance computing in certain academic environments; but the effort of machine setup and maintenance are a deterrent for most SME and occasional users, as well as for small research groups.

Against this backdrop, the Grid offers itself as a unique opportunity where computing power, state-of-the-art software and human expertise can be found on demand, and hence very cost-effectively.


Copyright 2004 FlowGrid Consortium | Please send questions or comments to Norberto.Fueyo@posta.unizar.es, or to any other FlowGrid partner | Last modified on 02/04/2004