Guidelines on Refinement Ratio
Jonathan
Is there maximum refinement ratio for one step detonation simulations
that should not be exceeded? For example, in a simulation I ran of a
CD ramp, I used a refinement ratio of 5. The simulation did not crash,
however, Matei has expressed concern for the fidelity of the solution,
due to the large refinement ratio.
Thank you in advance,
Jonathan
James Quirk
At the script level, the commands r, rI, and rJ generate an error if
you try to specify a number greater than 10. This was put in when
one student at CALTECH tried using 2 levels of refinement with a
ratio of 100 and was surprised that he kept on running out of memory;
at least until I pointed out that in 2D each refined cell resulted in
10,000 cells!
As far as your question is concerned, there's no real reason why you
couldn't use a refinement ratio of 5, but depending on the details
of the flow you might be pushing your luck. You need to keep two
basic points in mind as you increase r: (i) the interpolation at
fine-coarse boundaries will get progressively less accurate; (ii)
the stable time step is updated less frequently. In practice (ii)
is likely to kill before (i). Hence in practice r is kept to be
2,3 or 4. But again that's not to say a value of 5 can't be
used if the circumstances are right.
James