CFD Online Discussion Forums

CFD Online Discussion Forums (http://www.cfd-online.com/Forums/)
-   FLUENT (http://www.cfd-online.com/Forums/fluent/)
-   -   interpolate error of boundary profile (http://www.cfd-online.com/Forums/fluent/44915-interpolate-error-boundary-profile.html)

justin June 4, 2007 07:42

interpolate error of boundary profile
 
Hello eveyone,

I am using experiment data of atmospheric boundary layer to write inlet boudary profile by point style. the results show that velocity profile along the inlet is not same, but the tke profile is very well. The maximum error of velocity profile is about 10%. is its due to interpolate? How can I reduce this error as the experiment data is not enough? Any suggestion?

Best regards

Justin

Henrik June 4, 2007 09:25

Re: interpolate error of boundary profile
 
Hi,

I don't know if this is the actual reason for the deviations you see, but depending on how you set your velocity profile you might be changing node or cell values. You might want to try plotting using the other kind of data to see if the interpolation has anything to do with this.

Hope this helps,

Henrik

justin June 4, 2007 10:47

Re: interpolate error of boundary profile
 
Henrik, thanks. I have tried to densify the grids, the error still eixsts, but it drop down to 5%. I can't afford to densify the grid again. If it cannot disappear, is it acceptable?

Regards

Justin

Henrik June 4, 2007 10:54

Re: interpolate error of boundary profile
 
I guess what's acceptable is up to you to decide? :)

If you're setting values in your boundary condition and you don't get what you set in the program, then I would not be satisfied, however...

What are you setting, node or cell values? Are you giving pointwise information or are you using a formula and looping over faces? Are you sure that what you examine is the boundary and not an interpolation to the interior of the domain or something else? Are you initializing from this boundary or have you started calculating and are looking at an unconverged solution maybe? (should not affect the boundary anyway, however)

Good luck!

/Henrik

justin June 4, 2007 16:29

Re: interpolate error of boundary profile
 
Hi, Henrik,

Thanks for you much more information.

The point profile used cell value, but what I set was the plan of inlet.

I didn't set pointwise and keep it as as default one (x,1,y,1,z,1), as I used velocity magnitude.

I plot xy of two vertical line on the inlet after the fisrt iteration step and compared them.

Do you think my setting is all right?

Regards Justin


Henrik June 5, 2007 04:01

Re: interpolate error of boundary profile
 
Hi again

As far as I understand, you set three velocities on the inlet (one in each coordinate direction)? And then you make an XY-plot after one iteration and display cell values?

If you display cell values, then you cannot expect to see exactly the same value as the one you set as the boundary condition (since this is normally valid on the face only). After having made one iteration, the solution inside the domain has most probably changed, which means the cell value will be an interpolation between your boundary condition and the values of the nodes in the cell corners inside the domain. It is then very probable that you will see some differences, as indeed you have.

If you want to check your boundary conditions once and for all, I suggest you set them and then initialize the solution from that boundary. Then (without iterating) make the same XY-plot of the lines on the inlet and display the node values. From what I know, these are the values that should be exactly the same as what you set in the boundary conditions panel.

Good luck!

/Henrik


All times are GMT -4. The time now is 13:10.