Transpiration Boundary Condition
 

Hi there,

I would like to implement a transpiration boundary condition in SU2 for simulating blowing/suction from a solid permeable surface of an airfoil, for instance.

What I exactly mean with transpiration (or permeability) boundary condition is practically a solid wall, where the velocity components tangential to the wall are zero, while a nonzero velocity is allowed in the wall-normal direction. In pratice, this should be something like the already available subsonic BC_Inlet boundary condition with specified flow direction and mass flow, with the only difference that the wall-parallel velocity components should be aware that there is a wall. This is at the moment not the case.

Has anyone experience on this or an idea on how to formulate this in SU2?

I have been trying to merge the inlet boundary condition and the (for instance) isothermal wall boundary condition but without success at the moment.

______________________
Davide Gatti
Postdoctoral Researcher
Karlsruhe Institute of Technology (Germany)

So what happens if you just copy the inlet boundary conditions into the isothermal wall bc function So basically what is inside the function BC_INLET. You throw away the PRESSURE_INLET sections. You hardcode the values for Flow_Dir.

Hi bigfootedrockmidget,
thank you for your reply! I'll try to do that asap and report the result.

The closest I've done so far is to simply use the BC_Inlet boundary condition to mimic blowing through a portion of the suction side of an airfoil in the subsonic regime. The result was that the imposed velocity was not exactly the one that I prescribed in the configuration file (most importantly, it varied along the chord while it should have been constant). In particular, a wall-parallel velocity component started to build up, like if there was no no-slip condition. This is probably due to the missing viscous correction in BC_Inlet (it is commented out in the code, due to severe convergence problems, a comment says). Probably I should try, when I copy the BC_Inlet in the CNSSolver, to try to reactivate the correction by using the CNSSolver::Viscous_Residual method. What do you think?

Hi bigfootedrockmidget,

here some tries I have done:

1) If I use BC_Inlet with the laminar compressible boundary layer case, it does not converge
2) If I use BC_Isothermal wall but I modify the SetVelocity_Old part as follows

/*--- Store the corrected velocity at the wall which will
be zero (v = 0), unless there is grid motion (v = u_wall)---*/
if (dynamic_grid) {
nodes->SetVelocity_Old(iPoint, geometry->nodes->GetGridVel(iPoint));
}
else {
nodes->SetVelocity_Old(iPoint, blowingVelocity);
}
for (auto iDim = 0u; iDim < nDim; iDim++)
LinSysRes(iPoint, iDim+1) = 0.0;
nodes->SetVel_ResTruncError_Zero(iPoint);

it does converge and the velocity at the wall is correct. The wall-normal velocity does not seem to affect the flow. I obtain a wall-normal velocity profile similar to the boundary layer without blowing, the only difference being the first few 3 to 5 grid points at the wall, where I see an oscillatory behaviour of the wall-normal velocity.

3) If I combine the isothermal and inlet boundary condition, the simulation does not converge, as for the inlet case.

Hi,

You should not forget that the inlet boundary conditions are imposed in a weak way, so they are not exact. You might want to impose them in a strong way instead.

Thank you bigfootedrockmidget,

I'll try to compute the inlet state as done in BC_Inlet for the MASS_FLOW inlet and impose it in a strong way as done for (some) variables in BC_Isothermal_Wall.

I guess, tough, that I will incur in the same problem as before, i.e. that I do achieve the specified values for the variables at the wall, however these values do not seem to affect the flow. It seems like the information that there is a mass flux is missing.

I'll try it and let you know how it goes. Thank you for your support!

I tried to impose the values of the Primitives computed by BC_Inlet in a strong way (i.e. as done in BC_isothermal_wall by setting Solution_old, setting LinSysRes to zero and making the respective rows of the Jacobian unitary diagonal).

If the velocity I set is actually zero, then everything looks fine and I get a result like the one for the standard boundary layer, as we should. If I set a positive velocity, which would correspond to wall blowing, then I do still get unphyical results. In particular, a positive wall-normal momentum can be observed at the wall and outside of the boundary layer, while within the boundary layer the wall-normal velocity is negative.

Maybe I am doing something wrong?

Do you see any big problems in my way of proceeding?