Flow and deformation in an elastic tube

General discussion about Elmer
Post Reply
mrceresa
Posts: 18
Joined: 11 Jan 2013, 21:18
Antispam: Yes

Flow and deformation in an elastic tube

Post by mrceresa »

Dear all,
I was trying to repeat the FSI validation experiment [1] in which an elastic tube is deformed by a passing flow. I tried to build up the solution using both the tutorial on fluid-solid interaction and the suggestions to Emilie's post (viewtopic.php?f=3&t=324) but with no luck. After some iterations the solver says: "ERROR:: ElementMetric: Degenerate 2D element" and diverges. Maybe I have a bad mesh, or are the deformations too large for the current timestep/mesh size?

I attach both the solver log and the entire directory case.

Thanks for any hints you might provide,

Best,

Mario

[1] C. J. Greenshields, H. G. Weller: A unified formulation for continuum mechanics applied to fluid-structure interaction in flexible tubes, Int. Journal for Num. Methods in Engineering, 2005
Attachments
solverlog.txt
Solver log
(311.63 KiB) Downloaded 353 times
elasticTube.tar.gz
The entire case directory with mesh included
(113.2 KiB) Downloaded 374 times
raback
Site Admin
Posts: 4812
Joined: 22 Aug 2009, 11:57
Antispam: Yes
Location: Espoo, Finland
Contact:

Re: Flow and deformation in an elastic tube

Post by raback »

Hi

You need some trick to be able to use the weak coupling. Our trick is the arificial compressibility (AC) that mimic the elastic response of the structure [1,2]. Set the artificial compressibility to roughly 2*R/(E*h) and you'll have smooth convergence. Otherwise you have no way of having success when the pipe is long enough. The method of AC works beatifully for many challenging problems of internal flow FSI. There are many much more complicated ideas but I like this one because its simple and robust.

-Peter

[1] Peter Råback, Juha Ruokolainen, Mikko Lyly, Esko Järvinen. Fluid-structure interaction boundary conditions by artificial compressibility , A paper presented in ECCOMAS CFD 2001, Swansea 4-7 September. (available online)
[2] E. Järvinen, P. Råback, M. Lyly and J.-P. Salenius, A method for partitioned fluid-structure interaction computation of flow in arteries, Medical Engineering & Physics, 30 (2008), 917-923. doi:10.1016/j.medengphy.2007.12.008
mrceresa
Posts: 18
Joined: 11 Jan 2013, 21:18
Antispam: Yes

Re: Flow and deformation in an elastic tube

Post by mrceresa »

Thanks Peter!
I tried and indeed it worked beautifully. I post the complete solution in case someone else is looking for it.

I also used:
* non-linear iteration 1 and relaxation factor 0.5
* gradP discretization

I made a short animation in which pressure is the background color, wall displacement is the deformation of the mesh and fluid velocity is represented by arrows. It is available here:

https://www.youtube.com/watch?v=mwZw_9rpc2Q

However, I'm not sure if it is completely correct: in the first 10 seconds of the video, the pressure gradient goes backward as if the wave is reflected by the outlet. But velocity is always pointing in the same direction, so I'm not really sure what's going on... Maybe I should not fix both pressures at the beginning and end of the channel?

Best regards,
Mario
Attachments
elasticTube11.zip
The complete case dir
(290.23 KiB) Downloaded 403 times
raback
Site Admin
Posts: 4812
Joined: 22 Aug 2009, 11:57
Antispam: Yes
Location: Espoo, Finland
Contact:

Re: Flow and deformation in an elastic tube

Post by raback »

Hi Mario

Nice work! It is always nice when little advice has an impact.

The outlet conditions are really not that straight-forward. For that reason the community has been eager to use 1D outlet conditions. These has been used also with conjunction with Elmer (see some old tutorial). Unfortunately they are currently easility taken into use. There is some preliminary work to make this as a generic BC but this is waiting for better times currently....

You could also try with a velocity input. The AC method is so robust that it should have no problems either.

-Peter
mrceresa
Posts: 18
Joined: 11 Jan 2013, 21:18
Antispam: Yes

Re: Flow and deformation in an elastic tube

Post by mrceresa »

Thanks Peter! I'm really getting caught up in using Elmer :)
I'll do some more investigations...
shubh_shinde
Posts: 19
Joined: 13 Aug 2017, 23:58
Antispam: Yes

Re: Flow and deformation in an elastic tube

Post by shubh_shinde »

Hi Peter,
Thanks for the information on Artificial Compressibility and links to paper.
I have some doubts related to value of AC. I am working with an arterial geometry with stenosis. Thus, I consider the analytical solutions of AC do not apply here.
What is the method to set the AC value when the geometry is not regular.
In the paper there is a method called test load method but I do not see any such option in ElmerGUI. Is it such that it first asks some value and then the solver sets the value itself based on relative change in norm after iterations.
Please suggest on test load method and its implementation.
Thank you in advance,
Shubham
shubh_shinde
Posts: 19
Joined: 13 Aug 2017, 23:58
Antispam: Yes

Re: Flow and deformation in an elastic tube

Post by shubh_shinde »

Dear Peter,
I could figure out from the test case box how to use the test load method.
However, when I run the solver it says

Code: Select all

ERROR:: ListGetReal:  Can't find INDEPENDENT variable:[ac]for dependent variable:[Artificial Compressibility]
In the Artificial compressibility section under Material defination I put Equals ac and the ac.results file is generated as specified in the test case.
Also, I am not able to find out CompressibilitySolve module. Is it embedded in NavierStokes solver? If not please suggest path to the module.
I am attaching the .sif file for your reference
Thank you in advance,
Shubham
Attachments
case.sif
(4.73 KiB) Downloaded 296 times
raback
Site Admin
Posts: 4812
Joined: 22 Aug 2009, 11:57
Antispam: Yes
Location: Espoo, Finland
Contact:

Re: Flow and deformation in an elastic tube

Post by raback »

Hi

You should first perform the "test load" phase as in:
https://github.com/ElmerCSC/elmerfem/bl ... ox2/ac.sif

Unfortunately ElmerGUI offers only a fraction of the features of ElmerSolver. In this case you can make the 1st draft with the GUI but you probably need to abandon it quite early to dig deeper into the field.

-Peter
Post Reply