nodes position through time

smagmon · June 18, 2009, 21:58

Hi there,

i did a simulation in order to reproduce a experiment that involves wave propagation in water and also fluid structure interaction. This one consists on a confined fluid, in which pressure waves propagate through the fluid until finding a compliant wall. Those waves give moviment to the compliant wall that are measured.
I simulated this system and now I want to validate my model comparing the results with the experimental results. In my simulation I can observe the motion of wall, what I want to compare with my experimental data. The problem is that I dont know how. I mean, how does CFX solves the mesh motion? Is possible using CEL to get the coordinates of the nodes of a specific 2d region for each instant? If not, how can I get the node coordinates depending on time that I can observe using CFX-Post?
Any idea or tip on that direction will be very well received! Thank you in advance.

ghorrocks · June 19, 2009, 00:12

Hi,

You can get CFD-Post to display the X/Y/Z coordinate of the wall patch. If you want to extract the absolute positions for external analysis of you can get that out of CFD-Post using the export command.

Was the simulation done as an FSI simulation? If so then the mesh motion comes from the deformations predicted by the FEA solver.

Glenn Horrocks

smagmon · June 19, 2009, 06:28

Hi Mr. Horrocks,

thanks for your fast answer. I am using MFX and I set in my FEA (ANSYS Mechanical) only two boundary conditions.
!============================
!...
!BC1 - fix the boards of my "plate"
sf,boards,all,all,0
!BC2 - set FSI
sf,plate,fsin,1
!...
!============================
then I set the CFX simulation especifying by CEL the mesh displacement that generate the pressure waves. And also set the FSI region.
Therefore, I am looking for the mesh displacement in both sides. It is also to check if the exchange between CFX and ANSYS is ok for me. Due to convergence and stability I changed some settings, relaxation for example, and now I need to evaluate those changes. (some advises on that field has also high value to me!)

Back to my doubts, I could get from some points but it was a hard work getting those data and exporting. Is it possible to get x,y,z from each node in a 2D region through time using that approach you suggested? Does that approach permit exporting the data x(t),y(t),z(t) from each node in the surface?

Thanks

ghorrocks · June 19, 2009, 08:14

Hi,

You can get the x/y/z location of every point on a patch at every time step by one of two ways, neither of them are nice:

1) Make a monitor point on each point. Then you can output the x/y/z location of it through the solver manager
2) output a results file every time step. Then export the x/y/z points from CFD-Post.

This will generate a huge file and I wonder why you would bother. CFD-Post is post-processing software designed so you don't need to handle large datasets elsewhere. Why do you want to export it anyway?

Glenn Horrocks

smagmon · June 19, 2009, 08:50

Some parts of my surfaces present opposite displacements of other parts, what can give me a null volume displaced. But it is not what really happens. To avoid that, i want to use a RMS value from my displaced volume. As we are applying to the experimental data. Maybe it is possible using CEL, is it?
In my case, the second way is better, I think! You mean, trn file for each step? I already ran the simulation outputing for every step, (And you are right... it generates a huge file, but it is already there). And as I understood, what I have in that files, among other information, is the node position in each step. Then, can I export those data all together or do I need to export point by point? I mean, is possible to have a "surface" coordinates for each timestep or do I need to export each node position (P,i= x(t), y(t), z(t); where i is node index)? What I want to do is:
1 - define some very small areas using the nodes position (from three nodes position I can define the position of a element of area);
2 - take the component y of that element position;
3 - make the product of that component by the respective element of area;
4 - make the summation of that product for each element of area.
It gives me a volume, that can become zero because some position phase differences among the elements. And it is what I want to avoid applying the rms value of that product.

If you have any idea to make my work more effective, i will be really grate,

Smagmon

ghorrocks · June 20, 2009, 08:00

Why do you want to export the data? You can do this calculation in CFD-Post (or as a monitor point during a solver run) then you don't need to export anything. MUCH easier and more efficient. Have a look in the reference manual under the CEL Expression language for the types of functions you can do in CEL.

Glenn Horrocks

smagmon · June 23, 2009, 07:22

Hi,
Thanks for the tip. I will try to do that, I am learning with the reference material.

Sorry for not replying until now. I was out for a couple days.

June 18, 2009, 21:58	nodes position through time	#1
smagmon New Member Join Date: Mar 2009 Posts: 13 Rep Power: 17	Hi there, i did a simulation in order to reproduce a experiment that involves wave propagation in water and also fluid structure interaction. This one consists on a confined fluid, in which pressure waves propagate through the fluid until finding a compliant wall. Those waves give moviment to the compliant wall that are measured. I simulated this system and now I want to validate my model comparing the results with the experimental results. In my simulation I can observe the motion of wall, what I want to compare with my experimental data. The problem is that I dont know how. I mean, how does CFX solves the mesh motion? Is possible using CEL to get the coordinates of the nodes of a specific 2d region for each instant? If not, how can I get the node coordinates depending on time that I can observe using CFX-Post? Any idea or tip on that direction will be very well received! Thank you in advance.

June 19, 2009, 08:50		#5
smagmon New Member Join Date: Mar 2009 Posts: 13 Rep Power: 17	Some parts of my surfaces present opposite displacements of other parts, what can give me a null volume displaced. But it is not what really happens. To avoid that, i want to use a RMS value from my displaced volume. As we are applying to the experimental data. Maybe it is possible using CEL, is it? In my case, the second way is better, I think! You mean, trn file for each step? I already ran the simulation outputing for every step, (And you are right... it generates a huge file, but it is already there). And as I understood, what I have in that files, among other information, is the node position in each step. Then, can I export those data all together or do I need to export point by point? I mean, is possible to have a "surface" coordinates for each timestep or do I need to export each node position (P,i= x(t), y(t), z(t); where i is node index)? What I want to do is: 1 - define some very small areas using the nodes position (from three nodes position I can define the position of a element of area); 2 - take the component y of that element position; 3 - make the product of that component by the respective element of area; 4 - make the summation of that product for each element of area. It gives me a volume, that can become zero because some position phase differences among the elements. And it is what I want to avoid applying the rms value of that product. If you have any idea to make my work more effective, i will be really grate, Smagmon Last edited by smagmon; June 19, 2009 at 09:19.

Similar Threads
Thread	Thread Starter	Forum	Replies	Last Post
Physical Reason for stability of Implicit Schemes?	radhakrishnan	Main CFD Forum	26	October 3, 2023 23:05
SimpleFoam k and epsilon bounded	nedved	OpenFOAM Running, Solving & CFD	16	March 4, 2017 09:30
Is there a way to write the time step size, time a	may	FLUENT	6	November 22, 2009 12:52
DPM UDF particle position using the macro P_POS(p)[i]	dm2747	FLUENT	0	April 17, 2009 02:29
Unknown error	sivakumar	OpenFOAM Pre-Processing	9	September 9, 2008 13:53

June 19, 2009, 00:12		#2
ghorrocks Super Moderator Glenn Horrocks Join Date: Mar 2009 Location: Sydney, Australia Posts: 17,871 Rep Power: 144	Hi, You can get CFD-Post to display the X/Y/Z coordinate of the wall patch. If you want to extract the absolute positions for external analysis of you can get that out of CFD-Post using the export command. Was the simulation done as an FSI simulation? If so then the mesh motion comes from the deformations predicted by the FEA solver. Glenn Horrocks

June 19, 2009, 06:28		#3
smagmon New Member Join Date: Mar 2009 Posts: 13 Rep Power: 17	Hi Mr. Horrocks, thanks for your fast answer. I am using MFX and I set in my FEA (ANSYS Mechanical) only two boundary conditions. !============================ !... !BC1 - fix the boards of my "plate" sf,boards,all,all,0 !BC2 - set FSI sf,plate,fsin,1 !... !============================ then I set the CFX simulation especifying by CEL the mesh displacement that generate the pressure waves. And also set the FSI region. Therefore, I am looking for the mesh displacement in both sides. It is also to check if the exchange between CFX and ANSYS is ok for me. Due to convergence and stability I changed some settings, relaxation for example, and now I need to evaluate those changes. (some advises on that field has also high value to me!) Back to my doubts, I could get from some points but it was a hard work getting those data and exporting. Is it possible to get x,y,z from each node in a 2D region through time using that approach you suggested? Does that approach permit exporting the data x(t),y(t),z(t) from each node in the surface? Thanks

June 19, 2009, 08:14		#4
ghorrocks Super Moderator Glenn Horrocks Join Date: Mar 2009 Location: Sydney, Australia Posts: 17,871 Rep Power: 144	Hi, You can get the x/y/z location of every point on a patch at every time step by one of two ways, neither of them are nice: 1) Make a monitor point on each point. Then you can output the x/y/z location of it through the solver manager 2) output a results file every time step. Then export the x/y/z points from CFD-Post. This will generate a huge file and I wonder why you would bother. CFD-Post is post-processing software designed so you don't need to handle large datasets elsewhere. Why do you want to export it anyway? Glenn Horrocks

June 20, 2009, 08:00		#6
ghorrocks Super Moderator Glenn Horrocks Join Date: Mar 2009 Location: Sydney, Australia Posts: 17,871 Rep Power: 144	Why do you want to export the data? You can do this calculation in CFD-Post (or as a monitor point during a solver run) then you don't need to export anything. MUCH easier and more efficient. Have a look in the reference manual under the CEL Expression language for the types of functions you can do in CEL. Glenn Horrocks

June 23, 2009, 07:22		#7
smagmon New Member Join Date: Mar 2009 Posts: 13 Rep Power: 17	Hi, Thanks for the tip. I will try to do that, I am learning with the reference material. Sorry for not replying until now. I was out for a couple days.