dolfin team mailing list archive

["B. Emek Abali" <question155649@xxxxxxxxxxxxxxxxxxxxx>]

Question #155649 on DOLFIN changed:
https://answers.launchpad.net/dolfin/+question/155649

    Status: Open => Answered

B. Emek Abali proposed the following answer:
hi Kyle,

exter_B is for the exterior boudaries if you want to assign a Neumann
condition, which is not the case by you, so this is absolutely fine.

convection (hyperbolic pdes) is a highly sensible and unsolved phenomena
for the finite elements, your perturbation with upper lid drive is too
big for a stationary solution that Newton-Raphson may converge.

APicard (fix-point) iteration might be useful, or using high enough
viskosity for your Navier-Stokes material model,

but generally speaking do not try a steady state pde in fem. Therefore I
solved it as a time-dependent problem AND bring your lid not in a speed
in one step, start from zero and accelerate slow enough that you (Newton
with the first derivative, Galilei invariant) will find a solution, if
you want to use so high acceleration then leave Newton alone and invent
an Einstein (Lorentz invariant) procedere and please tell me also how to
do :)

So perturb your system slow enough that you can reach your state in a
time succeeded version, like in a laboratory, doing an experiment,
computing the reality.

-- 
You received this question notification because you are a member of
DOLFIN Team, which is an answer contact for DOLFIN.