Time Step with mesh deformation

rbarrett · July 1, 2011, 14:20

Hi,

I was just wondering if anyone has any advice on the size of a time step to use with a deforming model.

I am attempting to simulate the compression of a piston within a cylinder, which has an initial acceleration phase, a CV phase and a deceleration phase. The total compression time is 16.6 ms, the acceleration is 2079 m/s^2, deceleration is 23493 m/s^2 and the CV is 12 m/s. What I am seeing is that the simulation fails when I reach the short deceleration phase. If I reduce the value of this deceleration by an order of magnitude, the simulation runs to completion perfectly, with a timestep of 16.6ms/32.

Is there any guidelines available on what size of a time step I should be using with this type of simulation and is there a specific way to deal with this high of a deceleration term?

Thanks in advance

ghorrocks · July 2, 2011, 07:55

The fluid time scales are usually far smaller than the motion timescales, so whatever time step you need to resolve the motion, you are bound to need something far smaller for the fluids to be modelled accurately.

But obviously with such massive decelerations you need an appropriate time step to capture it, and the fluid mechanics associated with it are likely to require an even smaller timestep.

rbarrett · July 3, 2011, 08:43

Thanks a million for your reply. What I am seeing with a very small timestep is that the mesh folds on itself; does this mean that my mesh must be much more dense to accommodate this much reduced timestep?

Also, in terms of a numerical value for the timestep, is there any guideline as to what size of a timestep I should be using, or is it a matter of reducing it until it works for this particular simulation? Could I use a function that has a larger value of timestep for the acceleration, CV and post compression stages and refine it for the deceleration phase in a bid to reduce computational time?

Thanks in advance for any help.

ghorrocks · July 3, 2011, 08:56

The folding should have nothing to do with the timestep size. The fact you say it runs at a coarser time step is weird. But I would fix the root cause of the problem rather than trying different time step sizes. There are a few things you can do, varying mesh stiffness, remeshing, different mesh types etc.

Can you post an image of what your geometry looks like near where you have the problem?

rbarrett · July 4, 2011, 05:18

Please find attached below an image of the mesh just prior to the mesh folding. As the mesh is quite dense, I have only included a portion of the mesh. As you can see from the image, the mesh seems to be ok at this stage.
63.jpg

stumpy · July 4, 2011, 10:17

Double precision?

rbarrett · July 6, 2011, 11:08

Thanks a million for all your help. I had already double precision selected. I think that I have solved the problem as it was related to a poor mesh and is no longer occurring.

jfmorissette · July 7, 2011, 09:30

Stumpy, do you think that with mesh deformation and FSI, double precision is a must?

stumpy · July 7, 2011, 10:44

I would use DP for all mesh deformation and FSI cases.

July 1, 2011, 14:20	Time Step with mesh deformation	#1
rbarrett New Member Richard Barrett Join Date: Jun 2011 Posts: 14 Rep Power: 15	Hi, I was just wondering if anyone has any advice on the size of a time step to use with a deforming model. I am attempting to simulate the compression of a piston within a cylinder, which has an initial acceleration phase, a CV phase and a deceleration phase. The total compression time is 16.6 ms, the acceleration is 2079 m/s^2, deceleration is 23493 m/s^2 and the CV is 12 m/s. What I am seeing is that the simulation fails when I reach the short deceleration phase. If I reduce the value of this deceleration by an order of magnitude, the simulation runs to completion perfectly, with a timestep of 16.6ms/32. Is there any guidelines available on what size of a time step I should be using with this type of simulation and is there a specific way to deal with this high of a deceleration term? Thanks in advance

Similar Threads
Thread	Thread Starter	Forum	Replies	Last Post
How to write k and epsilon before the abnormal end	xiuying	OpenFOAM Running, Solving & CFD	8	August 27, 2013 16:33
[snappyHexMesh] snappyHexMesh won't work - zeros everywhere!	sc298	OpenFOAM Meshing & Mesh Conversion	2	March 27, 2011 22:11
Full pipe 3D using icoFoam	cyberbrain	OpenFOAM	4	March 16, 2011 10:20
DPM UDF particle position using the macro P_POS(p)[i]	dm2747	FLUENT	0	April 17, 2009 02:29
Computational time	sunnysun	OpenFOAM Running, Solving & CFD	5	March 16, 2009 04:32

July 2, 2011, 07:55		#2
ghorrocks Super Moderator Glenn Horrocks Join Date: Mar 2009 Location: Sydney, Australia Posts: 17,872 Rep Power: 144	The fluid time scales are usually far smaller than the motion timescales, so whatever time step you need to resolve the motion, you are bound to need something far smaller for the fluids to be modelled accurately. But obviously with such massive decelerations you need an appropriate time step to capture it, and the fluid mechanics associated with it are likely to require an even smaller timestep.

July 3, 2011, 08:43		#3
rbarrett New Member Richard Barrett Join Date: Jun 2011 Posts: 14 Rep Power: 15	Thanks a million for your reply. What I am seeing with a very small timestep is that the mesh folds on itself; does this mean that my mesh must be much more dense to accommodate this much reduced timestep? Also, in terms of a numerical value for the timestep, is there any guideline as to what size of a timestep I should be using, or is it a matter of reducing it until it works for this particular simulation? Could I use a function that has a larger value of timestep for the acceleration, CV and post compression stages and refine it for the deceleration phase in a bid to reduce computational time? Thanks in advance for any help.

July 3, 2011, 08:56		#4
ghorrocks Super Moderator Glenn Horrocks Join Date: Mar 2009 Location: Sydney, Australia Posts: 17,872 Rep Power: 144	The folding should have nothing to do with the timestep size. The fact you say it runs at a coarser time step is weird. But I would fix the root cause of the problem rather than trying different time step sizes. There are a few things you can do, varying mesh stiffness, remeshing, different mesh types etc. Can you post an image of what your geometry looks like near where you have the problem?

July 4, 2011, 05:18		#5
rbarrett New Member Richard Barrett Join Date: Jun 2011 Posts: 14 Rep Power: 15	Please find attached below an image of the mesh just prior to the mesh folding. As the mesh is quite dense, I have only included a portion of the mesh. As you can see from the image, the mesh seems to be ok at this stage. 63.jpg

July 4, 2011, 10:17		#6
stumpy Senior Member Join Date: Apr 2009 Posts: 531 Rep Power: 21	Double precision?

July 6, 2011, 11:08		#7
rbarrett New Member Richard Barrett Join Date: Jun 2011 Posts: 14 Rep Power: 15	Thanks a million for all your help. I had already double precision selected. I think that I have solved the problem as it was related to a poor mesh and is no longer occurring.

July 7, 2011, 09:30		#8
jfmorissette New Member Jean-François Morissette Join Date: Dec 2010 Location: Montreal Posts: 4 Rep Power: 16	Stumpy, do you think that with mesh deformation and FSI, double precision is a must?

July 7, 2011, 10:44		#9
stumpy Senior Member Join Date: Apr 2009 Posts: 531 Rep Power: 21	I would use DP for all mesh deformation and FSI cases.