FSI problem

amir foad · June 9, 2015, 12:42

Hi guys,

I am working on a micro pump by coupling structure and CFX.
I want to do transient CFD analysis, but after structural time step finishes, ANSYS says an error: fatal error, can not read mesh deformation data from interface. This happens when the membrane goes up in diaphragm.

I have not this problem in steady-state analysis.
I would appreciate if you help me.

Thanks in advance

mvoss · June 12, 2015, 10:43

Check the FEM results.... the solver is telling you " i cannot get the displacement..." because ANSYS crashed in the very last timestep. Often you can check the ANSYS.rst for what has happend in the "crashed, unconverged" last load step to get an idea what probably caused the problem.

amir foad · June 15, 2015, 13:33

Dear Mathias,

I checked the .rst file, but still don't understand where the error comes from

mvoss · June 15, 2015, 15:09

Did the last ansys loop reach convergence?

amir foad · June 16, 2015, 07:11

No it doesn't converge at the last step

mvoss · June 16, 2015, 07:40

Did you check where the residual is bad? Activate the NewtonRaphson-option in the Ansys results and re-run the case.
How big was the displacement for this last time step? Did the cfx solver reach converge?
At which point in the simulation did the error occur?
How was the interface convergence?
How did you set the relaxation of the transferred variables?
There are many reasons for the coupling to go wrong.

amir foad · June 16, 2015, 11:09

I set Raphson-Newton to Full.
at the last step the deformation is 9 microns (the geometry is the circular shell with 750 microns radius.
after 5 Iteration in interface, the simulation stops.
CFX solver didn't even start.
Interface residual is almost constant after 3rd iteration.
I set the convergence to 1E-4 RMS for mesh displacement and 1E-2 for interface variables.
The Initial condition is set to zero for all velocities and relative pressure.

And eventually I got this error in structural : Element 1 has excessive thickness change

By the way, thank you for your help man.

amir foad · June 16, 2015, 11:56

I forgot, set under-relaxation to 0.75

Thanks Matthias

mvoss · June 16, 2015, 13:00

What do you mean by "constant after third iteration"? Did the error rise up in the First Time step? Sounds like the initial deformation/stress is to much. How do you change fluid conditions from initial to working conditions? Can you Post a pic of both CFD And FEM Part?

amir foad · June 16, 2015, 13:01

residual values neither do go up nor down
actually I reduced the number of steps and the error removed but I got negative volume element instead

amir foad · June 16, 2015, 13:02

actually I reduced the steps and the error removed but I got negative volume element instead

ghorrocks · June 16, 2015, 19:13

FAQ: http://www.cfd-online.com/Wiki/Ansys..._went_wrong.3F

unihit4392 · August 9, 2015, 22:27

Hi,
I am trying to do FSI simulation of sinusoidal flow in elastic tube. Even i am having same error (excessive change of thickness in element). Did you resolve it?

amir foad · August 10, 2015, 05:47

Hi,
you should find an optimum element size to avoid negative volume.
This link may help you: http://www.cfd-online.com/Wiki/Ansys..._went_wrong.3F

unihit4392 · August 10, 2015, 07:34

ok thanks for the info

June 9, 2015, 12:42	FSI problem	#1
amir foad New Member Amir Foad Farshchi Join Date: Mar 2015 Posts: 11 Rep Power: 11	Hi guys, I am working on a micro pump by coupling structure and CFX. I want to do transient CFD analysis, but after structural time step finishes, ANSYS says an error: fatal error, can not read mesh deformation data from interface. This happens when the membrane goes up in diaphragm. I have not this problem in steady-state analysis. I would appreciate if you help me. Thanks in advance

June 15, 2015, 15:09	FSI problem	#4
mvoss Senior Member Matthias Voß Join Date: Mar 2009 Location: Berlin, Germany Posts: 449 Rep Power: 20	Did the last ansys loop reach convergence?

June 16, 2015, 13:00	FSI problem	#9
mvoss Senior Member Matthias Voß Join Date: Mar 2009 Location: Berlin, Germany Posts: 449 Rep Power: 20	What do you mean by "constant after third iteration"? Did the error rise up in the First Time step? Sounds like the initial deformation/stress is to much. How do you change fluid conditions from initial to working conditions? Can you Post a pic of both CFD And FEM Part?

Similar Threads
Thread	Thread Starter	Forum	Replies	Last Post
Simple FSI (fluid structure interaction) Problem	kilroy	Main CFD Forum	0	January 7, 2015 14:42
Defining contact regions in FSI problem.!!!	sreemvlk	ANSYS	0	June 12, 2014 00:33
prestrain in FSI problem	mina.m	STAR-CCM+	0	October 31, 2013 11:39
divergent of a FSI problem	edooo0	FLUENT	0	August 5, 2013 17:54
FSI MODELING PROBLEM (pinball)!!!	smn	CFX	1	September 16, 2009 00:33

June 12, 2015, 10:43		#2
mvoss Senior Member Matthias Voß Join Date: Mar 2009 Location: Berlin, Germany Posts: 449 Rep Power: 20	Check the FEM results.... the solver is telling you " i cannot get the displacement..." because ANSYS crashed in the very last timestep. Often you can check the ANSYS.rst for what has happend in the "crashed, unconverged" last load step to get an idea what probably caused the problem.

June 15, 2015, 13:33		#3
amir foad New Member Amir Foad Farshchi Join Date: Mar 2015 Posts: 11 Rep Power: 11	Dear Mathias, I checked the .rst file, but still don't understand where the error comes from

June 16, 2015, 07:11		#5
amir foad New Member Amir Foad Farshchi Join Date: Mar 2015 Posts: 11 Rep Power: 11	No it doesn't converge at the last step

June 16, 2015, 07:40		#6
mvoss Senior Member Matthias Voß Join Date: Mar 2009 Location: Berlin, Germany Posts: 449 Rep Power: 20	Did you check where the residual is bad? Activate the NewtonRaphson-option in the Ansys results and re-run the case. How big was the displacement for this last time step? Did the cfx solver reach converge? At which point in the simulation did the error occur? How was the interface convergence? How did you set the relaxation of the transferred variables? There are many reasons for the coupling to go wrong.

June 16, 2015, 11:09		#7
amir foad New Member Amir Foad Farshchi Join Date: Mar 2015 Posts: 11 Rep Power: 11	I set Raphson-Newton to Full. at the last step the deformation is 9 microns (the geometry is the circular shell with 750 microns radius. after 5 Iteration in interface, the simulation stops. CFX solver didn't even start. Interface residual is almost constant after 3rd iteration. I set the convergence to 1E-4 RMS for mesh displacement and 1E-2 for interface variables. The Initial condition is set to zero for all velocities and relative pressure. And eventually I got this error in structural : Element 1 has excessive thickness change By the way, thank you for your help man.

June 16, 2015, 11:56		#8
amir foad New Member Amir Foad Farshchi Join Date: Mar 2015 Posts: 11 Rep Power: 11	I forgot, set under-relaxation to 0.75 Thanks Matthias

June 16, 2015, 13:01		#10
amir foad New Member Amir Foad Farshchi Join Date: Mar 2015 Posts: 11 Rep Power: 11	residual values neither do go up nor down actually I reduced the number of steps and the error removed but I got negative volume element instead

June 16, 2015, 13:02		#11
amir foad New Member Amir Foad Farshchi Join Date: Mar 2015 Posts: 11 Rep Power: 11	actually I reduced the steps and the error removed but I got negative volume element instead

June 16, 2015, 19:13		#12
ghorrocks Super Moderator Glenn Horrocks Join Date: Mar 2009 Location: Sydney, Australia Posts: 17,712 Rep Power: 143	FAQ: http://www.cfd-online.com/Wiki/Ansys..._went_wrong.3F

August 9, 2015, 22:27		#13
unihit4392 New Member hallow rocks Join Date: Apr 2014 Posts: 9 Rep Power: 12	Hi, I am trying to do FSI simulation of sinusoidal flow in elastic tube. Even i am having same error (excessive change of thickness in element). Did you resolve it?

August 10, 2015, 05:47		#14
amir foad New Member Amir Foad Farshchi Join Date: Mar 2015 Posts: 11 Rep Power: 11	Hi, you should find an optimum element size to avoid negative volume. This link may help you: http://www.cfd-online.com/Wiki/Ansys..._went_wrong.3F

August 10, 2015, 07:34		#15
unihit4392 New Member hallow rocks Join Date: Apr 2014 Posts: 9 Rep Power: 12	ok thanks for the info