Two way FSI time stepping for different time scales

QBeast · August 6, 2016, 11:35

Dear forum users,

I'm running a 2-way FSI problem, transient structural(Anays mechanical)-fluidflow(fluent)

Problem description:

In a 2-way FSI problem, I have been struggling with the possibility of resolving the two systems fluid & structural on two different timscales, I'm doing a vertically standing rubber plate subjected to pressure build up on its face due to incoming fluid (air with single longitudinal velocity component at 4 m/s), the model is all set and start iterating, but the fluid flow evolves on relatively smaller time scales and needs say ~ a 5e-4 sec timestep to converge properly, while I managed to solve the transient structural analysis (alone) on a much larger timescale, like 0.05 sec and then 0.1s timestep. Now the problem comes mainly from the analysis settings of the coupling service which overrides the time stepping for Ansys Mechanical and fluent.

Question:

I am wondering if there is a way to avoid solving the structural problem for every fluid flow timestep which is too small for the rubber plate problem, and requiring a lot of CPU time to inverse the matrix by the Transient Structural solver?
.
My physical view of this coupled problem is that the flow field features developed around the obstacle (rubber plate) would include flow structures evolving on a finer timescale compared to the change of the force acting on the rubber plate, so in my opinion one need not necessarily to solve transient structural on the very same tiny timescale as the fluid system.
Ideally, I would’ve used (if possible) a fine time step for CFD say 5e-4 sec to ensure convergence of the turbulent flow and a suitable one for structural 0.05 sec, which is 100 time ratio and avoids inverting the matrix for the structural (100 times), since the load (aerodynamic force acting on the plate) doesn’t change that fast, at least as an integral value over the face.
Thanks in advance, I would appreciate comments or suggestion on this issue.
Best regards

mortazavi · August 10, 2016, 16:09

Hi, I think this is not possible in 2way fsi. the simulation iterates in each time step for both solid and fluid domains.

QBeast · August 11, 2016, 06:58

Thanks mortazavi, that's what I was afraid of, unfortunately this makes the coupled problem too computationally expensive on the structural side.
I'm wodering if there is a scripting I can hook in the mechanical model to workaround the this limitation ? otherwise Ansys should be thinking of this separate timescales issue, since it appears as a serious limitation.

Regards.

August 6, 2016, 11:35	Two way FSI time stepping for different time scales	#1
QBeast New Member Saâd Debbahi Join Date: Oct 2010 Posts: 26 Rep Power: 15	Dear forum users, I'm running a 2-way FSI problem, transient structural(Anays mechanical)-fluidflow(fluent) Problem description: In a 2-way FSI problem, I have been struggling with the possibility of resolving the two systems fluid & structural on two different timscales, I'm doing a vertically standing rubber plate subjected to pressure build up on its face due to incoming fluid (air with single longitudinal velocity component at 4 m/s), the model is all set and start iterating, but the fluid flow evolves on relatively smaller time scales and needs say ~ a 5e-4 sec timestep to converge properly, while I managed to solve the transient structural analysis (alone) on a much larger timescale, like 0.05 sec and then 0.1s timestep. Now the problem comes mainly from the analysis settings of the coupling service which overrides the time stepping for Ansys Mechanical and fluent. Question: I am wondering if there is a way to avoid solving the structural problem for every fluid flow timestep which is too small for the rubber plate problem, and requiring a lot of CPU time to inverse the matrix by the Transient Structural solver? . My physical view of this coupled problem is that the flow field features developed around the obstacle (rubber plate) would include flow structures evolving on a finer timescale compared to the change of the force acting on the rubber plate, so in my opinion one need not necessarily to solve transient structural on the very same tiny timescale as the fluid system. Ideally, I would’ve used (if possible) a fine time step for CFD say 5e-4 sec to ensure convergence of the turbulent flow and a suitable one for structural 0.05 sec, which is 100 time ratio and avoids inverting the matrix for the structural (100 times), since the load (aerodynamic force acting on the plate) doesn’t change that fast, at least as an integral value over the face. Thanks in advance, I would appreciate comments or suggestion on this issue. Best regards mortazavi likes this. Last edited by QBeast; August 11, 2016 at 08:58.

Similar Threads
Thread	Thread Starter	Forum	Replies	Last Post
simpleFoam error - "Floating point exception"	mbcx4jc2	OpenFOAM Running, Solving & CFD	12	August 4, 2015 02:20
Moving mesh	Niklas Wikstrom (Wikstrom)	OpenFOAM Running, Solving & CFD	122	June 15, 2014 06:20
AMI interDyMFoam for mixer nu problem	danny123	OpenFOAM Programming & Development	8	September 6, 2013 02:34
CFL in dual time stepping with implicit scheme	dokeun	Main CFD Forum	0	May 8, 2013 23:01
pisoFoam with k-epsilon turb blows up - Some questions	Heroic	OpenFOAM Running, Solving & CFD	26	December 17, 2012 03:34

August 10, 2016, 16:09		#2
mortazavi Member SMN Join Date: Jun 2009 Location: CANADA Posts: 71 Rep Power: 16	Hi, I think this is not possible in 2way fsi. the simulation iterates in each time step for both solid and fluid domains.

August 11, 2016, 06:58		#3
QBeast New Member Saâd Debbahi Join Date: Oct 2010 Posts: 26 Rep Power: 15	Thanks mortazavi, that's what I was afraid of, unfortunately this makes the coupled problem too computationally expensive on the structural side. I'm wodering if there is a scripting I can hook in the mechanical model to workaround the this limitation ? otherwise Ansys should be thinking of this separate timescales issue, since it appears as a serious limitation. Regards.