CFD Online Discussion Forums

CFD Online Discussion Forums (http://www.cfd-online.com/Forums/)
-   Main CFD Forum (http://www.cfd-online.com/Forums/main/)
-   -   Symmetric boundary (http://www.cfd-online.com/Forums/main/103507-symmetric-boundary.html)

Nereus June 19, 2012 12:53

Symmetric boundary
 
I'm working with a symmetric boundary and I'm trying to figure out what constraints this places on the system. I'm working from Blazek (2005), pg 290. In addition to requiring that the velocity normal to the axis of symmetry is zero, it states that the following gradients must vanish:
gradient of the scalar quantity normal to the boundary
gradient of the tangential velocity normal to the boundary
gradient of the normal velocity along the boundary (since \vec{v}\cdot \vec{n}=0)
The first and the last make perfect sense to me, but the second has me perplexed. What does this actually mean for the gradient of the velocity along the axis of symmetry? Should it be zero? Any help would be appreciated.

sbaffini June 19, 2012 14:06

I think that the general symmetry condition is:

- velocity normal to the boundary = 0

- gradient of any quantity in the direction normal to the boundary = 0

So, in a cell centered finite volume code, the symmetry boundary is not partecipating at all (all the fluxes trough it are 0). The convective flux is zero (first condition) like for walls, the boundary-normal derivative of the boundary-normal velocity component is zero (second condition) like for walls (incompressible flows at least) and the boundary-normal derivative of the boundary-parallel velocity component is zero (still second condition). The last condition is the difference with respect to wall boundary conditions as it implies a null wall shear stress.

Hence, as a matter of fact, the symmetry condition is like an inviscid wall condition. As you can see, no condition at all is placed on the boundary parallel derivatives which can be different from zero (think about the inviscid solution around a cylinder or the flow in an axisymmetric pipe with an expansion)

hhh August 30, 2012 09:10

boundary condition (symmetry vs periodic doubt)
 
Dear friends,
i am doing 3D wing analysis, i have doubt on boundary conditions in fluent ie both symmetry and periodic, anybody have the answer please let me know briefly,for what kind of purpose to use symmetry & periodic boundary conditions, please tell any example with problem.

sbaffini August 30, 2012 10:20

They are two fundamentally different boundary conditions whose applicability is dependent on the physics of the flow which is numerically simulated.

I already described the symmetry condition in the post above; the periodic boundary condition is fundamentally different as it treats the periodic boundaries as they were inside the domain with the neighbor cells being those next to the two periodic boundaries.

For example, in 1D with a finite difference approach and a grid index i going from 1 to nx, periodicity imply that the point i=1 also sees a fictitious point i=0 which actually is i=nx-1 (just traslated to the right position to the left of the point) and the point i=nx also sees a fictious point i=nx+1 which actually is i=2 (just traslated to the right position to the right of the point). Hence, periodicity means that on the boundary you don't actually use a b.c. (which would also require a modification of the discretization) but you use the fictious points.

From the physical point of view you have that a periodic condition is equivalent to infinitely replicate the computational domain in the periodic direction; also, you can expect a self-influence of the solution on itself which sometimes can lead to spurious effects.

Now, with this in mind, there are two different situations in which you can apply them: steady and unsteady simulations.

For steady simulations you can't really any more talk about a real dynamical systems, so the feasibility of one of the two b.c. is really to be determined according to the expected solution behaviour near the boundary. Moreover, symmetry from geometrical and external forcing parameters (including other b.c.) also imply symmetry in the solution (i'm not aware of any case where this is not correct). E.g., for a steady RANS simulation of a 3D wing in symmetric flight conditions you don't need to simulate the full wing and you can apply the symmetry condition on the root chord and simulate only half wing.

In this specific wing example, considering also that periodicity ALWAYS involves two boundaries at time (couples of boundaries),
i don't see how periodicity would apply. However there are a lot of steady cases where both are actually equivalent and some others where there are only slight differences. As always, physical insight is necessary to understand which one is better suited.

For an unsteady simulations (one in which unsteady fenomena are present) the situation is fundamentally different. The symmetry conditions is dynamically equivalent to an inviscid wall while the periodic boundary condition is like closing the computational domain on itself (any information leaving the domain from one periodic boundary will enter again from the other periodically coupled boundary).

Two very important cases where the differences between the two are very profound are in acoustics and DNS/LES computations. Immagine a plane indefinite channel flow where, of course, you have to truncate somehow the domain extension in the wall parallel directions. In acoustics as well as in DNS/LES a symmetry condition is extremely wrong and periodic boundary conditions are more suitable (with some restrictions).
Another case is the top boundary (relative to the external flow) in a plane boundary layer simulation. Of course, periodic boundary conditions do not apply any more and (still with restrictions) the symmetry (or inviscid wall) condition is more suitable.

hhh September 3, 2012 02:11

boundary condition
 
1 Attachment(s)
Dear sbaffini,
thanks a lot for your kind response, My problem specification is 3D wing, steady incompressible flow, velocity is 3 m/s, a=0.0102, length=0.065 for my 3D wing, here i attached my wing, in that domain i consider wall as wing one face is velocity inlet, opposite face is pressure outlet, remaining face i dont which boundary condition is suitable, please let me know

hhh September 3, 2012 02:16

boundary condition
 
1 Attachment(s)
Dear sbaffini,
thanks a lot for your kind response, My problem specification is 3D wing, steady incompressible flow, velocity is 3 m/s, a=0.0102, length=0.065 for my 3D wing, here i attached my wing, in that domain i consider wall as wing one face is velocity inlet(1), opposite face is pressure outlet(2), remaining face i dont which boundary condition is suitable, please let me know

truffaldino September 3, 2012 03:44

put velocity inlet also on top and bottom. Put symmetry on sides: one symmetry plane exactly at the wing root and another (paralllel) symmetry plane should be far enough from the wingtip.

mans October 21, 2012 01:26

Symmetry boundary - flotherm
 
FLOTHERM - When I do sanity check, it provides warning: more than one symmetry boundary defined on x and z direction - ignored for radiation. What is symmetry boundary. Do I need to do anything. or ignoe it.
FYI - I have placed a PCB assy in a channel and want to run Sim.
Thanks


All times are GMT -4. The time now is 19:48.