axisymmetric FSI problem

[hamed]

Dear Elmer group

I’m working on an axisymmetric steady-state FSI problem. But there are is a contradiction in the results that I’ll try to explain it in the following.

The problem initial state is shown in Fig. (1). As can be seen in this figure the fluid enters from the left inlet and exits from the outlet and meanwhile its velocity increases (to keep the volume flux constant). Thus the pressure in the inlet is larger than the pressure in the outlet which is prescribed as p=0.

In the first case we set inlet velocity equal to 0.1 m/s. The pressure distribution for this case is shown in Fig (2a) and the diaphragm deflection due to the fluid flow is shown in Fig (2b). Note that the fluid flow causes diaphragm to deflect in the upward direction (maximum deflection is: dy= +0.016).

In the second case we set the inlet velocity equal to zero and apply Force 2 = -20.0 on the bottom boundary of the diaphragm. The diaphragm deflects downwards as shown in Fig. (3) and the maximum deflection is equal to (dy=-0.127).

In the last case we combine the first and second cases. Namely, set the inlet velocity equal to 0.1 m/s and apply Force 2 = -20.0 on the bottom boundary of the diaphragm. Now we expect the deflection of the diaphragm to be something between its deflection in the first and second case (to move upward due to fluid flow and downward due to Force 2: dy = +0.016 - 0.127 = -0.111). But the result of the simulation gives the maximum deflection equal to (-0.146 = -0.016 - 0.127). This means both the fluid flow and Force 2 cause the diaphragm to move downward! Isn’t it a bug?

I also attached the simulation files to this post.
Regards,

[Juha]

Hi,

at least the elmerpost output from ElmerSolver concerning the mesh deformations in these
type of cases has been badly broken for a few weeks (since rev 4909). I've just commited
a fix to svn (rev 4976).

Regards, Juha

[Juha]

Hi,

your Elmerpost files are probably OK, by the way, judging from the pictures,
guess you have an ElmerSolver older than a few weeks...

I don't really know if the results you have are due to a bug, or to insufficient resolution,
for example, or just physics. Hard to predict nonlinear interactions like that with
changing geometry, at least for me...

Regards, Juha