Free surface flow deformed shape error

messbalint · October 27, 2010, 19:50

Hello

We have a problem with the deformed shape of three dimensional free surface flow driven by surface tension. The intersection at the inlet is like a racetrack, the problem is at the curved section of it. It seems like a mesh problem, but we have tried many types of meshes (with and without O-Grid) and they didn’t work. We’ve tried these options: curvature underrelaxation factor, Laplacian volume fraction smoothing, volume fraction coupling, harmonic body force averaging and decreased timescale at volume fraction convergence control.
Time is running and we can't get closer to the solution, so please try to help us!

Thank you in advance for your help!

ghorrocks · October 28, 2010, 19:08

Surface tension flows are highly sensitive to mesh quality. You really need hex elements at as close to 1:1 aspect ratio as possible for any hope.

Is this a steady state or transient model? What are you modelling anyway?

messbalint · October 29, 2010, 07:51

Thank you for your reply Glenn!

We will simulate a liquid sheet injected from a nozzle into the atmosphere and the model you have seen in the pictures would be a simple case to get to know the behave of these kind of models. The models are steady state and we use homogeneous multiphase model. Both phases (air and water) are continuous. We are interested in the deformation of the sheet but not in its disintegration.
A model is running which results seems to be great, and from that I think the keys will be the aspect ratio (as you have said) and that the elements side must try to be parallel every place with the gas-liquid boarder-line. Do you think it could be right?

ghorrocks · October 29, 2010, 08:15

No, the important issue is to have all elements to be cubes or as close to it as possible. No need for alignment with the free surface.

Free surface models in steady state are very hard. I use CFX and Fluent for free surface models with surface tension and neither can handle steady state stuff. You really have to run everything transient, and just march it out to steady state. Time consuming, but it is the only way.

messbalint · October 29, 2010, 08:20

I will try! Thank you so much!

Similar Threads
Thread	Thread Starter	Forum	Replies	Last Post
Building OpenFOAM1.7.0 from source	ata	OpenFOAM Installation	46	March 6, 2022 14:21
CGNS Compiling	Diego	Main CFD Forum	17	December 21, 2014 02:40
OF 1.6 \| Ubuntu 9.10 (64bit) \| GLIBCXX_3.4.11 not found	piprus	OpenFOAM Installation	22	February 25, 2010 14:43
user subroutine error	CFDUSER	CFX	2	December 9, 2006 07:31
user defined function	cfduser	CFX	0	April 29, 2006 11:58

October 28, 2010, 19:08		#2
ghorrocks Super Moderator Glenn Horrocks Join Date: Mar 2009 Location: Sydney, Australia Posts: 17,872 Rep Power: 144	Surface tension flows are highly sensitive to mesh quality. You really need hex elements at as close to 1:1 aspect ratio as possible for any hope. Is this a steady state or transient model? What are you modelling anyway?

October 29, 2010, 07:51		#3
messbalint New Member Mess Balint Join Date: Oct 2010 Posts: 6 Rep Power: 16	Thank you for your reply Glenn! We will simulate a liquid sheet injected from a nozzle into the atmosphere and the model you have seen in the pictures would be a simple case to get to know the behave of these kind of models. The models are steady state and we use homogeneous multiphase model. Both phases (air and water) are continuous. We are interested in the deformation of the sheet but not in its disintegration. A model is running which results seems to be great, and from that I think the keys will be the aspect ratio (as you have said) and that the elements side must try to be parallel every place with the gas-liquid boarder-line. Do you think it could be right?

October 29, 2010, 08:15		#4
ghorrocks Super Moderator Glenn Horrocks Join Date: Mar 2009 Location: Sydney, Australia Posts: 17,872 Rep Power: 144	No, the important issue is to have all elements to be cubes or as close to it as possible. No need for alignment with the free surface. Free surface models in steady state are very hard. I use CFX and Fluent for free surface models with surface tension and neither can handle steady state stuff. You really have to run everything transient, and just march it out to steady state. Time consuming, but it is the only way.

October 29, 2010, 08:20		#5
messbalint New Member Mess Balint Join Date: Oct 2010 Posts: 6 Rep Power: 16	I will try! Thank you so much!