Mereja

blog post:”Sledgehammer CFD” – The Best Approach?

Written by

David Tatchell

in

Re-reading my last blog (Beginning at the Beginning …5 Parabolic or Elliptic, Or Somewhere In Between?) a slightly disturbing thought occurred to me. Does the distinction, highlighted in that blog, between parabolic, partially-parabolic, and elliptic solution methods have any real relevance to present day CFD?
On consideration, the honest answer is – only to a very limited extent.

In the 1970s when we were struggling to perform realistic 3D computations even on the most powerful computers of the day (at Imperial College we were using a CDC 6600, which was then a state-of-the-art super computer), the savings in computer time and storage from using a parabolic or partially-parabolic solution method rather than a fully-elliptic one, when it was valid to do so, could be crucial in making a computational study practicable or affordable.

Now, however, when we have virtually-free access to virtually-infinite computer resources (at least relative to the “austerity” of the 1970s), these distinctions are pretty much meaningless. One might as well use a fully-elliptic solution method for everything – the convenience in using one standard method for everything vastly outweighs any “wasted” computer resources.

To a purist it might seem unfortunate that the scientific elegance of the parabolic and partially-parabolic approaches have been lost. But it seems to me that, if the simplicity of a unified, standard approach (even when it might be “a sledgehammer to crack a nut“) has contributed to the widespread adoption of CFD that we see today, then it has been a price worth paying.

More posts