Convective boundary condition

[Geoff Wright]

Hi,

I'm trying to use sfepy to model the temperature distribution at steady state, where theres a heat source, conduction through a solid, and convection across a solid/liquid interface.  Struggling with the convection part... I've been doing some reading and I think I need to enforce a convective boundary condition, sometimes known as a robin boundary condition.  This would add a term where you integrate the temperature of the liquid across the interface surface.  Does this sound right?  Are there any examples of doing something like this?

Thanks,

Geoff

[Robert Cimrman]

The Robin BC should be applicable using dw_surface_dot term. What particular
form of the BC do you need?

r.

[Geoff Wright]

Well, to be more specific, the interface over which convection is happening isn't an external surface.  Its an internal surface and I want to model the temperature of both the liquid and the solid parts.  I took a look at dw_surface_dot and BCNewtonTerm but I believe that they're intended only for external surfaces.  Is this correct?  How would you go about solving this?

Thanks,

Geoff

[Robert Cimrman]

On 08/19/2014 06:01 PM, Geoff Wright wrote:
> Well, to be more specific, the interface over which convection is happening
> isn't an external surface. Its an internal surface and I want to model the
> temperature of both the liquid and the solid parts. I took a look at
> dw_surface_dot and BCNewtonTerm but I believe that they're intended only
> for external surfaces. Is this correct? How would you go about solving
> this?

Yes, dw_bc_newton might be even better.

Those terms work on any surface, provided their region is correctly defined.
Say we have an internal boundary Gamma between Omega1 and Omega2. Then you can
define Gamma1, Gamma2 that behave like external surfaces to Omega1, Omega2
respectively as follows:

'Gamma' : ('r.Omega1 *v r.Omega2, 'facet'),
'Gamma1' : ('copy r.Gamma', 'facet', 'Omega1'),
'Gamma2' : ('copy r.Gamma', 'facet', 'Omega2'),

The last argument is the parent region. Then use Gamma1 for integration from
the "Omega1 point of view" etc.

Does that help?

r.

[Geoff Wright]

Yes that helps thanks! It's great that region definitions are so flexible. I think I have it working now with dw_surface_dot. I couldnt use bc_newton because the temperatures on each side are not constant.

G

[Robert Cimrman]

Hth!

r.