CFX solver interpolation

Fahd · July 1, 2018, 10:47

Hi guys,

I used user defined remeshing through junction box routines. CFX solver interrupt the calculation at each time step, then replace the mesh file with new one and interpolate the result of old mesh into the new mesh.

which type of interpolation CFX uses.
do I have access to change it or not and how?
Does interpolation scheme affect results?

regards,
-Fahd

ghorrocks · July 1, 2018, 18:50

It is almost certainly linear interpolation. I have not found a reference to justify that but I would imagine higher order interpolation would cause more trouble than it is worth due to unphysical overshoots.

I do not think you have access to change the interpolation done by CFX. If you want to change the way this is done you can write your own initial condition in user fortran and use what ever interpolation scheme you like.

Yes, interpolation scheme will affect results. Getting an interpolation scheme which is conservative is challenging. Also getting one which accurately keeps rapid changes in variables is also challenging. Getting a scheme which does both is just about impossible.

Lance · August 3, 2018, 01:53

The expert parameter

Code:

    interp second order data = t

will change the interpolation to second order. But beware, the .res files will become bigger.

ghorrocks · August 3, 2018, 06:25

I did not know that, thanks!

But be aware that higher order functions are prone to gross overshoots. Caveat emptor!

July 1, 2018, 10:47	CFX solver interpolation	#1
Fahd New Member Fahd Mohamed Elabd Join Date: Sep 2016 Location: Qatar Posts: 8 Rep Power: 9	Hi guys, I used user defined remeshing through junction box routines. CFX solver interrupt the calculation at each time step, then replace the mesh file with new one and interpolate the result of old mesh into the new mesh. which type of interpolation CFX uses. do I have access to change it or not and how? Does interpolation scheme affect results? regards, -Fahd

July 1, 2018, 18:50		#2
ghorrocks Super Moderator Glenn Horrocks Join Date: Mar 2009 Location: Sydney, Australia Posts: 17,703 Rep Power: 143	It is almost certainly linear interpolation. I have not found a reference to justify that but I would imagine higher order interpolation would cause more trouble than it is worth due to unphysical overshoots. I do not think you have access to change the interpolation done by CFX. If you want to change the way this is done you can write your own initial condition in user fortran and use what ever interpolation scheme you like. Yes, interpolation scheme will affect results. Getting an interpolation scheme which is conservative is challenging. Also getting one which accurately keeps rapid changes in variables is also challenging. Getting a scheme which does both is just about impossible. __________________ Note: I do not answer CFD questions by PM. CFD questions should be posted on the forum.

August 3, 2018, 01:53		#3
Lance Senior Member Lance Join Date: Mar 2009 Posts: 669 Rep Power: 22	The expert parameter Code: interp second order data = t will change the interpolation to second order. But beware, the .res files will become bigger. brunoc likes this.

August 3, 2018, 06:25		#4
ghorrocks Super Moderator Glenn Horrocks Join Date: Mar 2009 Location: Sydney, Australia Posts: 17,703 Rep Power: 143	I did not know that, thanks! But be aware that higher order functions are prone to gross overshoots. Caveat emptor! __________________ Note: I do not answer CFD questions by PM. CFD questions should be posted on the forum.

Similar Threads
Thread	Thread Starter	Forum	Replies	Last Post
fluent divergence for no reason	sufjanst	FLUENT	2	March 23, 2016 16:08
thobois class engineTopoChangerMesh error	Peter_600	OpenFOAM	4	August 2, 2014 09:52
CFX solver doesn't start through Workbench	oj.bulmer	CFX	3	June 30, 2012 08:44
viewing cfx post while working on cfx solver manager	HMR	CFX	5	March 9, 2011 22:33
[ICEM] trouble with mesh quality from ICEM in CFX Solver	escher25	ANSYS Meshing & Geometry	0	February 28, 2011 07:38