Ansys 2 way FSI coupling
 

Hi,

I have recently started to model flow inside an artery using ANSYS Workbench through 2 way coupling.
After some run time my simulations crash due to excessive distortion in transient structural part. I have tried to refine the mesh but the model is still unstable and it crashes.

Any ideas you share will be appreciated.


Cheers!
Awais

Reducing the time-step as well may help here..

I actually tried reducing the time step here but it even makes it worse and the simulation crashes earlier than the former one.

In that case, you may need to provide some details (screenshots/ BCs etc.) to evoke any useful response..

Sorry for the delayed response, I was away and could not answer to the thread earlier.

I am currently using velocity-inlet and pressure-outlet (0 gauge pressure) boundary condition. The flow rate corresponding the velocity at inlet is 350ml/min. The Reynold's number is between 300-400 for the flow. The model set-up is transient laminar.

The tube is fixed at the end faces and there is no other boundary condition on the tube, except for the fluid structure interface (see attached image). The Young's modulus I am looking at is quite low (from 0.05 to 0.1MPa, with a poisson's ratio of 0.495). Linear Elastic properties are assumed in this model.

The time step I am using is 0.01 and when I run the simulation, it crashes after a while giving the error for too much mesh deformation in some elements.

I have tried to use the ramping function in workbench coupling environment, but when the solution progress to next step there is a sudden spike in deformation, which is where the simulation crashes.

Please do let me know if you require any further information about the model.

Attachment 47509

Attachment 47510

Attachment 47511

you may need to go to time-steps of the order of 10^-4 or less, which is a very blunt statement unless you really know what you intend to achieve, worth a try nevertheless... secondly, the point where the simulation crashes, open the results and try seeing the problematic region.. refine the mesh further in that region.. mesh deformation is tricky and needs careful meshing approach especially when it comes to massive deformations.. my 50 cents and what i could comprehend from your provided info..

Fresty,

I carried out the simulation with 10^-2 time step, from which I retrieved the distorted mesh (see attachments please).
As you can see from the images, the maximum distortion just before simulation crashed is around 0.7mm, where the wall thickness was 0.25mm and the internal diameter was 6.35mm. The element size that I have for my mesh is 0.0007m

I recall that I tried 10^-3 time step as well and simulation crash even earlier than the former one.


However, I will try the 10^-4 time step as you suggested.
I want to look at high amplitude, low frequency displacements as well as low amplitude, high frequency displacements.

The issue here I think might be the low compliance of the material. The young's modulus is as low as 0.1MPa and there is no boundary on the surrounding surface to contain the displacement. I thought it might be worth writing a UDF for a pressure boundary condition on the bounding solid surface where the pressure starts from certain value and tends to 0. Do you think it would work? The issue I think here might be that the simulation time would increase drastically.


I have also tried to run the case in a rigid tube and then import the final results (with developed flow) from fluent into a standalone case with a compliant tube so that it might not cause sudden deformation at the start but even that does not seem to be working.

is this deformation real or just due to some problem in numerical setup. FSI is very problematic when flow is in-compressible.

Far,

Thanks for your response.

I am not sure about this. By the way, the deformation you see in images has been magnified to aid visual appearance but I am looking at values in the legends and they don't seem to be very high.
However, I know that the displacement in real life situation (from experimental data) should be somewhere around this order of magnitude that simulation has given. But since the simulation has not stabilized, I can not say whether they are real or due to numerical errors? Is there a way that I can check this ? What possible issue can be there in the numerical set-up?

I am quite new to this area, but you are right that FSI with incompressible flow can be tricky. I read in the manual provided by Ansys that stabilization is tricky in these cases but there are ramping and stabilization factors that they refer to, but that did not help me (provided that I am applying them correctly).