What would be the ill effects of putting a hook in the code that would make the boundary values correct? It would be a simple extrapolation using half the cell width and the gradient in the cell.
- - - - - - - - - - - - - - - - - L.Bryce Whitson Jr. [email protected] On Tue, Nov 18, 2014 at 1:09 PM, Daniel Wheeler <[email protected]> wrote: > On Tue, Nov 18, 2014 at 12:44 PM, L.Bryce Whitson Jr. <[email protected] > > wrote: > >> I recently noticed that the faceValue of a variable is incorrect when a >> Neumann (constant gradient) boundary condition is applied. For example, >> let's assume the following: >> >> dx = 1 >> nx = 5 >> dTr/dx = 4 for right face >> Tl = 300 for left face >> >> x = [0.5, 1.5, 2.5, 3.5, 4.5] >> >> The steady solution to this problem would be: >> T = [302, 306, 310, 314, 318] >> T.faceValue = [300, 304, 308, 312, 316, 320] >> >> However, in FiPy I get the following for the faceValues >> T.faceValue = [300, 304, 308, 312, 316, 318] >> >> I'm assuming this is a bug in the faceValue code. Would this be correct? >> > > Hi Bryce, > > It is certainly unintuitive and it causes results to be misinterpreted, > but it isn't a bug in the sense that the equations needs that value during > solve(). Basically, the "faceValue" property isn't aware of the faceGrad > constraint, but equations are aware. The "faceValue" is really something > that needs to be correct for the equation solution and wasn't intended to > be used to extract values, necessarily. However, it would be nice if it all > just worked intuitively. > > I hope that helps a bit. > > Cheers, > > Daniel > > > -- > Daniel Wheeler > > _______________________________________________ > fipy mailing list > [email protected] > http://www.ctcms.nist.gov/fipy > [ NIST internal ONLY: https://email.nist.gov/mailman/listinfo/fipy ] > >
_______________________________________________ fipy mailing list [email protected] http://www.ctcms.nist.gov/fipy [ NIST internal ONLY: https://email.nist.gov/mailman/listinfo/fipy ]
