Re: [pooma-dev] Chevron Flow Scenario

[James Crotinger]

on 4/11/01 8:20 PM, James Crotinger at address@hidden wrote:

> I could scan the notes in and send around JPEGs if there is interest.

I went ahead and scanned the first two pages in and put them on my web site.
Those who are interested can get them at

  http://homepage.mac.com/jcrotinger/Lee1.jpe
  http://homepage.mac.com/jcrotinger/Lee2.jpe

These give the derivation in 2D of the 8 equations in 8 unknowns that are
solved in the neighborhood of every vertex on the mesh. I then rambled on
and derived the effective permeability given in the paper for the case of a
diagonal permeability, made a stab at expanding out some equations that
lagged the off diagonal terms (messy and not carefully done - didn't see
much insight coming), and finally applied the same technique in 1D and
compared to the centered difference I'd write down in 1D. (I think it would
be interesting to solve these numerically as I don't understand what really
makes this algorithm special/different. Indeed, this is my major criticism
of the paper - I get no feeling for the algorithm, either qualitatively or
from a more formal numerical analysis perspective. How does it converge?
They show that in simple cases it reduces to using an effective permeability
that is a harmonic average - is that good? Does it deal better with
discontinuities? Why is this algorithm better than some sort of normal
finite difference or other sort of boundary element method?)

    Jim