Fluidity

What kind of boundary condition would be used as default?

Asked by liu chin chi

Hi all,

 If I don't to take a BC option at a boundary in flml file,what kind of BC would be used at pressure,velocity,and scaler.If anyone knows or be mentioned somewhere in the manual,please give me a hint.

Thanks

Question information

Language:
English Edit question
Status:
Solved
For:
Fluidity Edit question
Assignee:
No assignee Edit question
Solved by:
James Robert Percival
Solved:
Last query:
Last reply:
Revision history for this message
liu chin chi (caite8001) said : 		#1

Default BC seems like below

Velocity field: Neumann with values in three ways [0 0 0]
Pressure field: Dirichlet with a value=0
other scalers field: Neumann with a value=0

Not sure.

Revision history for this message
Best James Robert Percival (j-percival) said : 		#2

If no boundary condition is specified, then the system which is solved is equivalent to the one generated by "natural boundary condition" problem, i.e. whatever will make the boundary surface surface in integrals in the weak formulation of the problem vanish.

This will depend a lot on your choice of spatial discretization, but would often amount to a zero Neumann condition for all variables, arising from the diffusive term in the momentum and scalar advection terms, and the straight Laplacian in the pressure correction term. Note that if the equations are being solved in conservative form, then this is akin to a Robin boundary condition (zero flux) instead.

Revision history for this message
James Robert Percival (j-percival) said : 		#3

I should probably note that specifying no boundary conditions what so ever on a set of faces is likely to lead to an ill-posed problem and failing numerics, You should normally be applying a Dirichlet boundary condition on either the normal component of velocity or on the pressure.

Revision history for this message
liu chin chi (caite8001) said : 		#4

Hi James,

Thank you for your explanation. that's helpful.

regards

Liu

Revision history for this message
liu chin chi (caite8001) said : 		#5

Thanks James Robert Percival, that solved my question.