XANSYS Message: 93029 [Go back to message list]
[bookmark on del.icio.us]
No rating yet
Subject: Re: Ansys FSI nightmare
Author: Peter Joseph Attar
Date: 2008-04-25 22:46:50Tan,
As somebody mentioned the key question here is what is the physics that you
are trying to understand? FSI problems are extremely difficult without
buckling. In my dealings in aeroelasticity, we have much simpler structural
models in that we don't expect buckling and the constitutive models I usually
use are linear elastic. With all of this, we still have problems. One thing
that I do that helps with numerical issues is to try and get a good flow
solution which has been integrated through any transients first and
then "release" the flexibility of the structure. So for example if I am
interested in a wing at 25 degrees angle of attack, I first simulate just the
flow physics around a rigid delta wing at 25 degrees and then once I've
solved for a while and all of the fluid transients have died out, I then
start with the fully coupled solution. So, unless you are really interested
in the dynamics of going from an unbuckled to a buckled state, you might want
to solve the structural problem only first to get the buckled deformation
field then perhaps solve ONLY the fluid problem through this "rigid"
geometry. Then after a while you can "turn on" the flexibility of the
structure. This will perhaps alleviate some of the problems you are having
with mesh movement when the system undergoes the dramatic change from buckled
to unbuckled state.
If you are indeed interested in the dynamics of this, then my guess would be
that you need to refine the heck out of the mesh for BOTH the fluid and
structure in the region where you are having issues. Also I've had better
luck with shells than solsh190's when it came to large deformation, so if you
can, you might want to switch to something like shell181s for the structure.
I have never actually used the FSI in ANSYS so I am not sure of the
limitations on the typs of elements which can be used for FSI analysis.
Also my above idea about the different solutions may not be possible without
the correct restart capabilities.
Anyway good luck. It sounds like a fun problem.
Peter
On Friday 25 April 2008 17:08, Y Y Tan wrote:
> Hi guys,
>
> Thanks for the quick reply. Thanks for caring. :)
>
> First of all, i'm an undergrad, and I have to hand up my full dissertation
> next week. I havent even started writing it yet. :(
>
> My supervisor,does not know how to do FSI simulations and so far I dont
> think there's anyone in my department who does this.
>
> Christopher, I'm not too sure bout the manual part. I dun think i'll have
> anytime for that now. I'm in desperate need of a result to put into my
> dissertation.
>
> Mason, this is currently the simplest it can be.
>
> Anup, the aim of my exercise is to develop a model for blood flow inside a
> vein and the physics of the vein is exactly similar to flow inside a
> flexible tube. Thanks for letting me know that sequential is better. I have
> tried using Flotran before this but i don't really know how to set up the
> ALE and there's not much guide on it. The MFX seems a little more friendly
> since I can do it with Workbench.
>
> Let me tell you guys what I did and what happend to my model:
>
> 1) Create Solid model in Simulation, which is just a quarter of a pipe
> since the pipe would buckle into a 2 lobe shape. Fixed the ends and applied
> symmetry on the symmetry planes. Applied an external pressure as well. I
> modelled my geometry to be slightly off-circle so that I could get the
> model to buckle. I'm using line-search, autotimestepping and non-linear
> stabilization. The element used is Solsh190, element formulation was
> programme controlled. Material properties nearly incompressible with
> poison's ratio 0.49.
>
> 2) Create CFD model in CFX, applied Inlet velocity, outlet pressure,
> symmetry, and the interface. Set coupling time duration 1s and coupling
> time step 1s and simulation type steady-state. Did not apply any
> initialization. Under solver control, maximum iteration:50, fluid time
> scale control set to conservative auto timescale. For the coupling control,
> max iteration 20, tried playing with the under-relaxation factors but
> didn't really help much because I dont really understand what
> under-relaxation does.
>
> Problems that occured:
> 1) Negative volume in CFD, manage to tackle this with a finer mesh at first
> but then the problem still came, so I change the mesh displacement
> stiffness to 1/boundary^10, problem still occured so I did try to change
> the pseudo CFX time and also lower the under-relaxation factors which have
> helped. So as far as negative volume is concerned, its not happening yet as
> the solver crashes before this could happen.
>
> 2) Mesh highly distorted for the solid, element turning inside out, element
> warping. I've had all this in the solid part, which doesnt happen when I
> run the solid part alone. I've already got quite a good mesh. I used to
> think that this should not happen because the net pressure which is applied
> to the solid is much smaller during the FSI than in the single Solid
> solution but I have recently notice one big problem in my model. I use
> Ansys and keep looking at the results file as the simulation goes on, one
> thing is that in one stagger iteration, the pipe buckles with the vertical
> part collapsing inwards and the horizontal part moving outwards. It is
> suppose to happen this way since I model the vertical direction to have a
> slightly smaller radius than the horizontal part. Anyway, this goes on like
> that and it seems like it would be converging but then suddenly, after
> maybe the 5th stagger iteration, the horizontal part starts collapsing
> inwards while the vertical part moves outwards and the solution starts
> diverging. So yea, this is the MAIN problem.
>
> >From your experience, do you guys think that it is a good idea if I write
> > my
>
> dissertation focusing on the problems which occured and the steps which I
> undertook and have no results? or this would definately not put me in a
> good position?
>
> Regards,
> Tan Yi Yong
> The University of Sheffield
>
> ^--------------------------------------------------------
>
> | XANSYS - www.xansys.org |
> | The Discussion List for users of ANSYS, Inc. Software |
> | Hosted by PADT - www.padtinc.com |
>
> ^--------------------------------------------------------
--
Peter Attar
Assistant Professor
The University of Oklahoma
Dept. of Aerospace and Mechanical Engineering
865 Asp Avenue, Felgar Hall Room 212
Norman,OK 73019-1052
phone: 405-325-1749
fax: 405-325-1088
^--------------------------------------------------------
| XANSYS - www.xansys.org |
| The Discussion List for users of ANSYS, Inc. Software |
| Hosted by PADT - www.padtinc.com |
^--------------------------------------------------------