Discussion Forum

Hi

not sure why, as I have an older release (had'nt noticed the latest patch 166 ;) not sure when it appeared as there are no dates on "update" the web site.

But I notice that your solid material is rather weak, probably too weak for the "linear elastic material" model even with large deformations on. You get a singularity of deformation t he end of the fixed contrained tube section. Try increasing it's Young modulus by a factor 5 to 10x (at least to see if it behaves better as then your model might not be correct anymore)

Then I notice that the "no-slip" condition for the inlet region is not respected, hence mesh and flow is not really laminar with "0" at border (and the laminar inflow in 2D-axi is still restraining v =0 on the axis when turning on restarin to "zero" at ends, at least in my .134)

You are getting also many inverted mesh elements, perhaps remeshing could help

Use the "plot while solving" to see what is happening during the solving process, or use probe plots.

It could be nteresting to monitor the inflow and outflow and flow difference too

--
Good luck
Ivar

Hi Ivar,

Thanks for your comments. I m Abbie's colleague, working on the same model.
1) I tried to increase the Young's modulus by 10x (into 100e3 Pa), however the results remained the same.
2) Thanks for ur comments on no-slip. I din't realize it before, until I use the probe plot as u suggested. I realized this is due to the "fixed constraint" I set on line 6 (the internal wall on the top of the geometry). If I remove this, then no-slip is obeyed and I got a laminar flow start from the inflow. However, I m not sure why. Do u ve an idea?
3) Regarding the mesh, I tried refining the mesh, and also tried with a quad mesh. The problem I mentioned before remains. May I know where can I see the "inverted mesh"? Why does that normally happen?
4) You mentioned about "a singularity of deformation at the end of the fixed contrained tube section" - may I know where can I see this?
In another similar practice, I removed the tube (I cut away the "tube" between z=4 and z=5). I ve now only one line of fixed constraint at the outlet section (the end horizontal line at the wall). It seems like the problem is fixed. However, I m not sure why. Any ideas?

Thanks.

Einly

Hi

Correction: COMSOL prescribes the mesh also on the axis, which means that the axial symmetry is not read by the ALE part.

The non slip condition is too some limitations of the physics tweaking of the multi-physics module here, I would call it slightly a weakness, but as user one shall always test all entries, we cannot blame COMSOL anyhow it's up to us users to check everything (I just got caught on some material properties, with a T and P dependence, I didn't check the domain of T, I was outside nominal range, hence I lost 1/2 a days work ;)

You can leave the fixed constraint, but then add a "wall node" flow "no-slip" on that same boundary/wall, then remesh, you will see COMSOL adds automatically a boundary layer along the all the no-slip walls

A general comment, it's worth to always resolve units issues in your case function calls use t[1/s] to get quit the orange warning colour. This helps for tricky unit cases, even if it's not a big issue here

Another comment, with laminar inflow, if you drive it with a pressure, take into account the pressure drop in the artificial entrance tube COMSOL is adding to develop the flow (1m is often far too long, take a good CFD textbook and estimate the flow length from there (with 100E3 as young modulus). Another way to see the balloon compress, as now you turn off your flow, is to use a pressure dependent out flow, i.e. 500Pa and set the initial condition to 500Pa everywhere then the balloon returns partly to its initial shape when the inflow stops

However, what remains tricky is the constrained in/outflow I too have problems to get it solving. One way to improve might be to start with smaller steps for the flow turn on. Another point is to remove the second seggregated step and add all variables to the first seg step, then I get a sort of solution

--
Good luck
Ivar

Thanks so much Ivar!

The problem has been solved by adding the wall with no-slip condition onto the "fixed constraint" line. Now I get a laminar flow at the inlet and outlet section and I ve what I expected in the results.

I ve 2 questions here:

1) What I understand is that by applying fixed constraint like what I did before, it overwrites the original FSI boundary setting which assumes no-slip. Because of that, I don't see a "laminar" flow at the inlet and outlet. However, since this is the requirement of my project, the fixed constraint has to be there. My question now is why with a fixed constraint I get a "wrong results"? The reason I m asking this is that, even without a laminar flow profile in (as before), I shd still get an expected volume change with what I set as inlet and outlet, but COMSOL is not giving me that. Is that a numerical error? What's the reason behind this?
2) I m using fully coupled solver, not a segregated solver, as I assumed that fully coupled solver is better suited for large deformation problem? Is it true?

Thanks again for ur time!

Hi

the "fixed" constraints are OK, you only need to add the no-slip wall for the fluids to get a true laminar flow.

On the other side you can also force directly a parabolic input shape with a function.
In 2D or 2D-axi you can use the "s" variable that goes from 0 to 1 along the edge direction of the arrow to input a velocity of the type vz=3/2*v0*(1-s^2) where v0 is the average velocity

your true problem is that your material is so soft that you get extreme deformation of your "solid" around the inlet fixed attachment region (turn on the deformations for the solid it's more apparent

If you have enough RAM the fully coupled solver is often more efficient w.r.t solving time (except if your problem is fully uncoupled, in which case the segregated solver (when correctly set-up) could be quicker)

--
Good luck
Ivar

Thanks so much Ivar! :) Now I m clearer.