Hello,
I am having trouble with my model. I want to find the final falling speed of a liquid (blood) in
a bifurcating piece of pipe. I spend three days making different meshes and experimenting
with the element types. Still I´m getting no convergence. Does anybody have an idea what
else I could try?
Cheers
Tobias
Help for convergence difficulties
Help for convergence difficulties
- Attachments
-
- Y-Verzweigung_5_8.zip
- Test case.
- (432.61 KiB) Downloaded 220 times
-
- Site Admin
- Posts: 4832
- Joined: 22 Aug 2009, 11:57
- Antispam: Yes
- Location: Espoo, Finland
- Contact:
Re: Help for convergence difficulties
Hi
Nice little case you've there. I think that the problem you have comes partically from the fact that you don't fix the pressure level by any means. Normally any Neumann BC (normal velocity not defined) determines the pressure level but you don't have any. You have introduced an artificial compressibility which introduces a term c*(p-p0) where p0 by default refers to the previous pressure. So you can see that this lets the pressure travel if is not fixed by any other means. One way to fix it is to set p0=0 by adding in Material section
Of course this introduces a small consistancy error so you have to be alert on what value of AC you're using.
I'm just running your case with this setting and convergence seems to go nicely... Additionally I set the relaxation to 0.5 and applied my current favorite linear system solver settings, as given below, but I don't know whether they have any real impact.
-Peter
Nice little case you've there. I think that the problem you have comes partically from the fact that you don't fix the pressure level by any means. Normally any Neumann BC (normal velocity not defined) determines the pressure level but you don't have any. You have introduced an artificial compressibility which introduces a term c*(p-p0) where p0 by default refers to the previous pressure. So you can see that this lets the pressure travel if is not fixed by any other means. One way to fix it is to set p0=0 by adding in Material section
Code: Select all
Artificial Pressure = Real 0.0
I'm just running your case with this setting and convergence seems to go nicely... Additionally I set the relaxation to 0.5 and applied my current favorite linear system solver settings, as given below, but I don't know whether they have any real impact.
Code: Select all
Linear System Iterative Method = BiCGStabL
Bicgstabl Polynomial Degree = 4