[raishigeta]

I need to write a UDF that will read a point coordinate from a table and provide me with Velocity(u and v) at that point. That point might not be at a node, so fluent would have to interpolate.

Is there a UDF line command to get the Velocity at a specific X,Y location?

[AlexanderZ]

my vision:
you may use loop over cells and get coordinates of centers of these cells,
compare them with data from table, extract velocity and interpolate it yourself

[raishigeta]

Originally Posted by AlexanderZ

my vision:
you may use loop over cells and get coordinates of centers of these cells,
compare them with data from table, extract velocity and interpolate it yourself

Thank you for taking your time to consider my problem. I really appreciate it.
I was working on that way that you suggested.
However, I couldn't find a good way to interpolate it for myself.
I'm afraid to ask that Do you know any good way to interpolate it from the data of centers or nodes, kind of using any free-software or some?
Or Could I ask your vision again?
I'm not majoring this kind of things and I need to graduate soon...
Is it easier to do it myself without using UDF or TUI things?

[vinerm]

Hi Rai

Do you wish to extract value from the cell in which the point lies or do you wish to interpolate the values from the adjacent cells or nodes to your point and report that? For the later, you have to use some interpolation algorithm. If mesh is fine, I would suggest using 0 order interpolation, i.e., the nearest node value or picking up the value from the center of the cell in which the point exists. There are a couple of macros available within Fluent to find the id of the cell in which a given point lies.

You may use DPM_Locate_Point(x,&a,0.0,0)

where x is the vector containing point coordinates and a is the variable to cell ID. There certainly could be better solutions available depending upon the version of Fluent. Once you have the cell ID, you may loop through all nodes or faces and retrieve interpolated value.

Vinerm

[raishigeta]

Originally Posted by vinerm

Hi Rai

Do you wish to extract value from the cell in which the point lies or do you wish to interpolate the values from the adjacent cells or nodes to your point and report that? For the later, you have to use some interpolation algorithm. If mesh is fine, I would suggest using 0 order interpolation, i.e., the nearest node value or picking up the value from the center of the cell in which the point exists. There are a couple of macros available within Fluent to find the id of the cell in which a given point lies.

You may use DPM_Locate_Point(x,&a,0.0,0)

where x is the vector containing point coordinates and a is the variable to cell ID. There certainly could be better solutions available depending upon the version of Fluent. Once you have the cell ID, you may loop through all nodes or faces and retrieve interpolated value.

Vinerm

Thank you for your quick reply. I really appreciate it. I feel like I understood better than before now.
But it seems like that what I need is 1 order interpolation (linear interpolation?).
The detail of my circumstances is here: I'm doing 2D air flow analysis in square region(100mm×100mm).
I need the value of the points that are located at (x,y)=(1,1),(1,2)...(99,99).
Now I've got the values of 0 order interpolation which is like the same values as the closest nodes.

But thanks anyway, I really appreciate that you took your time to consider my problem.

[vinerm]

Hi

Linear interpolation would be very expensive since you have to calculate the distance of each node from the point within the cell and then calculate the value. On the other hand, averaging over nodes would not be that expensive. However, it is difficult to justify to go into all this. 0 order interpolation is quite close to 1st order if mesh is fine enough.

[sbaffini]

Once you get the ID of the cell containing your point, it is as simple as retrieving both the cell value and the gradient of the variable you want to interpolate, then just:

phi_point = phi_cell + SUM(grad_cell(i)*(x_point(i)-x_cell(i)),i=1,ND)

Depending from what you want to do, it might make sense to properly distinguish between pure gradients and reconstruction gradients, both of which are available.

You might need to use the TUI to tell Fluent to avoid freeding the memory for the gradients.

[han.velthuis@tno.nl]

Hi, I was looking for the same, thought it would be a straightforward question, however did not find a working example for Fluent 18. I started UDF programming but ran into the problem of undocumented functions, type conflicts nd so on. As a quick workaround I made a script (any language will do) that makes a Fluent journal file "MakeJou.jou". The scripts reads a file ("points.txt") with points (x, y, z), generates the TUI commands to make the point-surfaces, to export the flow data at the point-surfaces to file "DataAtPoints.txt", and to delete the point-surfaces again, as not to leave traces in Fluent.

Code for generation of journal file MakeJou.jou (this case Mathcad 15)
[IMG]C:\tmp\tmp.bmp[/IMG]

Example:
----------

File with points (x,y,z, coordinates) "points.txt":
0.55 -0.04 0.0045
0.3 0.05 0.0001
-0.55 0.05 0.0001

Autogenerated Journal file "MakeJou.jou":

q
/surface
point-surface
Point-1
0.55
-0.04
0.0045
point-surface
Point-2
0.3
0.05
0.0001
point-surface
Point-3
-0.55
0.05
0.0001
/file/export
ascii
DataAtPoints.txt
Point-1
Point-2
Point-3

yes
xvelocity
yvelocity
zvelocity
velocity
pressure
temperature
density
viscosity-lam
q
no
yes
0
q
/surface/delete-surface
Point-1
/surface/delete-surface
Point-2
/surface/delete-surface
Point-3

File with output for each data point "DataAtPoints.txt":
nodenumber, x-coordinate, y-coordinate, z-coordinate, viscosity-lam, density, temperature, pressure,velocity-magnitude, z-velocity, y-velocity, x-velocity
1, 5.500000000E-01,-4.000000000E-02, 4.499999762E-03, 8.906351952E-06, 8.156205730E-06, 2.969481164E+02, 9.988928086E+00, 1.775980586E+01,-8.117933759E-01, 1.176474198E+00, 1.769677456E+01
2, 3.000000104E-01, 5.000000000E-02, 9.999999404E-05, 8.837132723E-06, 1.846997675E-05, 2.935214512E+02, 2.235924084E+01, 9.385820385E+00, 1.460540274E-04, 0.000000000E+00, 9.385820342E+00
3,-5.500000164E-01, 5.000000149E-02, 9.999999404E-05, 8.840741512E-06, 4.097106808E-05, 2.937000280E+02, 4.962854152E+01, 1.299232211E-01,-8.431707945E-05, 0.000000000E+00, 1.299228378E-01

[han.velthuis@tno.nl]

https://paste.pics/ad0e64c49e39978173f01302e6f09d08