Considering that CFD (Computational Fluid Dynamics) is an advanced mathematical method for predicting fluid flow and heat transfer using a computational software approach I find it paradoxical that it is often as much an art as it is a science. A model by its very definition is a representation of something. A computer model is a virtual representation of a system (e.g. a laptop, a rack mounted server, a TQFP etc.) that may or may not exist already in reality. The representation must be sufficient in that it describes the system to the extent where when simulated, it behaves in the same way that it would/does in reality. Just in the same way that a painted portrait must be sufficiently life-like to ensure that the person can be recognised. Though try telling that to Picasso or Braque…
As CFD software gets easier to use, product by product, release by release, it will be used by a wider and more diverse group of engineers, maybe even industrial designers. Until such time as CFD vendors truly address the issues of automation, fully automating the art of creating a tractable CFD model that will simulate the reality to within a user specified tolerance (no pressure there then guys) then it’s critical that all CFD users understand what assumptions they are making when creating a computer model.
I’ll be taking a facetious approach in explaining why nearly all aspects of CFD modelling are assumptions to varying degrees. A good modeller will at least be aware of these, if not use them to his/her advantage. A about a year ago to the day I ran a series of blogs based on the quote “All models are wrong, but some are useful”. In the same way that art is never reality, just an interpretation of it, then a computer model will never be exactly true. Fret not though, even if you weren’t very good at art in school you’ll quickly find that as an engineer you are surprisingly artistic (and productive!) with the help of tools such as FloTHERM and FloVENT.
May 11th 2010, Ross-on-Wye