How can i get velocity profile like this ?

Falcon35 · April 13, 2015, 12:32

Quote:

Originally Posted by CeesH

I just looked at the Cornell results, and they do see the same overshoot in velocity at the edge of the boundary layer there. With that, I assume there is a small y-velocity component as well. Because in your case, it is there in the solution. If you color by the y-velocity instead of x-velocity, you'll see.

how can i fix it ?

CeesH · April 13, 2015, 12:37

Well, if they find it as well, with the current fix there may not be a fix. Perhaps with a more refined mesh, and stricter residuals you do get closer to the Blasius solution. But if you are doing the exact Cornell tutorial, this is probably what it is.

Falcon35 · April 13, 2015, 12:41

Quote:

Originally Posted by CeesH

Well, if they find it as well, with the current fix there may not be a fix. Perhaps with a more refined mesh, and stricter residuals you do get closer to the Blasius solution. But if you are doing the exact Cornell tutorial, this is probably what it is.

so is it true or false ? and how can i get location of 0,99*Uinlet ?

CeesH · April 13, 2015, 12:52

Physically, it should not be there. Numerically, it is, and it also is in the tutorial. If your goal is to replicate the tutorial, I wouldn't worry about the result. Otherwise, see if increasing mesh density helps.

Falcon35 · April 13, 2015, 12:57

Quote:

Originally Posted by CeesH

Physically, it should not be there. Numerically, it is, and it also is in the tutorial. If your goal is to replicate the tutorial, I wouldn't worry about the result. Otherwise, see if increasing mesh density helps.

man, i won't replicate , i have to do my homework and i need location of 0,99Uinlet to calculate boundary layer thickness numerically please help me ?

CeesH · April 13, 2015, 13:01

increase the mesh density, see if the results change. To get the 0.99-level; with umean is 1, plot the iso-surface where u is 0.99m/s

Falcon35 · April 13, 2015, 13:05

Quote:

Originally Posted by CeesH

increase the mesh density, see if the results change. To get the 0.99-level; with umean is 1, plot the iso-surface where u is 0.99m/s

okay good, how will i do it for 0,99Uinlet ? Can you describe it step by step like this plot>area-weighted average vs vs (for example) again ?

CeesH · April 13, 2015, 13:15

In FLUENT, you could create a custom field function: define > custom field function. Set this equal to Ux/<enter value of inlet velocity>.

Then go to surfaces > iso-surface, and create an iso surface of constant field function, value 0.99.
Next, go to graphics and animations, contours and plot the surface.

This gives you a qualitative view however. For a quantitative view, I would recommend exporting the data to MATLAB or CFD post. Maybe it's possible in FLUENT too, but I don't know how to do that for every x-location. Perhaps you should also take a look at the post-processing part of the cornell tutorial because they explain the steps for plotting in CFD post very clearly, also normalization, and with that you should be able to do it.

Falcon35 · April 13, 2015, 14:13

Quote:

Originally Posted by CeesH

In FLUENT, you could create a custom field function: define > custom field function. Set this equal to Ux/<enter value of inlet velocity>.

Then go to surfaces > iso-surface, and create an iso surface of constant field function, value 0.99.
Next, go to graphics and animations, contours and plot the surface.

This gives you a qualitative view however. For a quantitative view, I would recommend exporting the data to MATLAB or CFD post. Maybe it's possible in FLUENT too, but I don't know how to do that for every x-location. Perhaps you should also take a look at the post-processing part of the cornell tutorial because they explain the steps for plotting in CFD post very clearly, also normalization, and with that you should be able to do it.

May i do it by "write to file" velocity profile for every x ?

Falcon35 · April 13, 2015, 17:55

Quote:

Originally Posted by CeesH

In FLUENT, you could create a custom field function: define > custom field function. Set this equal to Ux/<enter value of inlet velocity>.

Then go to surfaces > iso-surface, and create an iso surface of constant field function, value 0.99.
Next, go to graphics and animations, contours and plot the surface.

This gives you a qualitative view however. For a quantitative view, I would recommend exporting the data to MATLAB or CFD post. Maybe it's possible in FLUENT too, but I don't know how to do that for every x-location. Perhaps you should also take a look at the post-processing part of the cornell tutorial because they explain the steps for plotting in CFD post very clearly, also normalization, and with that you should be able to do it.

brother ????

CeesH · April 14, 2015, 03:53

You can use write to file for every desired x indeed; then import the data into excel, matlab or some other software, and do the quantitative analysis there.

Falcon35 · April 14, 2015, 04:48

Quote:

Originally Posted by CeesH

You can use write to file for every desired x indeed; then import the data into excel, matlab or some other software, and do the quantitative analysis there.

May i see boundary layer thickness in there by 0.99U ?

CeesH · April 14, 2015, 04:59

yes. when you have the numbers, for each value of x you can look up the first number where Ulayer > 0.99U. Write down the position of this value for each x-position, and you get a good view of the boundary layer development

Falcon35 · April 14, 2015, 05:47

Quote:

Originally Posted by CeesH

yes. when you have the numbers, for each value of x you can look up the first number where Ulayer > 0.99U. Write down the position of this value for each x-position, and you get a good view of the boundary layer development

ty, i got 0,025 m boundary layer thickness for x= 0,5 m but i got it 0,034 by analitical method why is it ?

CeesH · April 14, 2015, 05:54

Neither. Both the analytic and numerical approach make assumptions and thereby are approximate. Compare both with experimental data, and see which gets closer.

Falcon35 · April 14, 2015, 06:02

Quote:

Originally Posted by CeesH

Neither. Both the analytic and numerical approach make assumptions and thereby are approximate. Compare both with experimental data, and see which gets closer.

how can i compare both with experimental data

CeesH · April 14, 2015, 06:37

You need to find a source reporting experimental data regarding boundary layer development over a flat plate. It must exist, because solutions like the Blasius solution are developed based on and compared to such data. Look in literature, find the data, do the comparison.

Falcon35 · April 14, 2015, 06:50

Quote:

Originally Posted by CeesH

You need to find a source reporting experimental data regarding boundary layer development over a flat plate. It must exist, because solutions like the Blasius solution are developed based on and compared to such data. Look in literature, find the data, do the comparison.

But teacher wanted us to calculate it by formula and ansys fluent ? i asked you why are they too different ; fluent's value 0,025 , formula's value 0,034 ?

CeesH · April 14, 2015, 07:03

As I said, both the numerical approximation and the analytic solution make certain assumptions which cause their solution to differ. study the assumptions you make in both approaches, compare what their effect on the found profile might be. But remember, neither approach gives a perfect answer.

Falcon35 · April 14, 2015, 07:18

Quote:

Originally Posted by CeesH

As I said, both the numerical approximation and the analytic solution make certain assumptions which cause their solution to differ. study the assumptions you make in both approaches, compare what their effect on the found profile might be. But remember, neither approach gives a perfect answer.

then different results between 0,025 and 0,034 NORMAL ?

April 13, 2015, 13:01		#26
CeesH Senior Member Cees Haringa Join Date: May 2013 Location: Delft Posts: 607 Rep Power: 0	increase the mesh density, see if the results change. To get the 0.99-level; with umean is 1, plot the iso-surface where u is 0.99m/s Falcon35 likes this.

April 14, 2015, 03:53		#31
CeesH Senior Member Cees Haringa Join Date: May 2013 Location: Delft Posts: 607 Rep Power: 0	You can use write to file for every desired x indeed; then import the data into excel, matlab or some other software, and do the quantitative analysis there. Falcon35 likes this.

Similar Threads
Thread	Thread Starter	Forum	Replies	Last Post
[UDF] Relative coordinates in logarithmic velocity profile	cfdworker	Fluent UDF and Scheme Programming	25	July 13, 2017 03:44
[swak4Foam] Inlet velocity profile for turbulent pipe flow using swak4Foam	zordiack	OpenFOAM Community Contributions	5	February 23, 2016 01:52
Logarithmic velocity profile	cfdworker	Fluent UDF and Scheme Programming	0	April 23, 2009 19:09
[boundary condition] logarithmic velocity profile	cfdworker	FLUENT	2	April 17, 2009 23:36
Prescribed inflow velocity profile - how to?	Alan	Main CFD Forum	10	October 28, 2005 12:14

April 13, 2015, 12:37		#22
CeesH Senior Member Cees Haringa Join Date: May 2013 Location: Delft Posts: 607 Rep Power: 0	Well, if they find it as well, with the current fix there may not be a fix. Perhaps with a more refined mesh, and stricter residuals you do get closer to the Blasius solution. But if you are doing the exact Cornell tutorial, this is probably what it is.

April 13, 2015, 12:52		#24
CeesH Senior Member Cees Haringa Join Date: May 2013 Location: Delft Posts: 607 Rep Power: 0	Physically, it should not be there. Numerically, it is, and it also is in the tutorial. If your goal is to replicate the tutorial, I wouldn't worry about the result. Otherwise, see if increasing mesh density helps.

April 13, 2015, 13:15		#28
CeesH Senior Member Cees Haringa Join Date: May 2013 Location: Delft Posts: 607 Rep Power: 0	In FLUENT, you could create a custom field function: define > custom field function. Set this equal to Ux/<enter value of inlet velocity>. Then go to surfaces > iso-surface, and create an iso surface of constant field function, value 0.99. Next, go to graphics and animations, contours and plot the surface. This gives you a qualitative view however. For a quantitative view, I would recommend exporting the data to MATLAB or CFD post. Maybe it's possible in FLUENT too, but I don't know how to do that for every x-location. Perhaps you should also take a look at the post-processing part of the cornell tutorial because they explain the steps for plotting in CFD post very clearly, also normalization, and with that you should be able to do it.

April 14, 2015, 04:59		#33
CeesH Senior Member Cees Haringa Join Date: May 2013 Location: Delft Posts: 607 Rep Power: 0	yes. when you have the numbers, for each value of x you can look up the first number where Ulayer > 0.99U. Write down the position of this value for each x-position, and you get a good view of the boundary layer development

April 14, 2015, 05:54		#35
CeesH Senior Member Cees Haringa Join Date: May 2013 Location: Delft Posts: 607 Rep Power: 0	Neither. Both the analytic and numerical approach make assumptions and thereby are approximate. Compare both with experimental data, and see which gets closer.

April 14, 2015, 06:37		#37
CeesH Senior Member Cees Haringa Join Date: May 2013 Location: Delft Posts: 607 Rep Power: 0	You need to find a source reporting experimental data regarding boundary layer development over a flat plate. It must exist, because solutions like the Blasius solution are developed based on and compared to such data. Look in literature, find the data, do the comparison.

April 14, 2015, 07:03		#39
CeesH Senior Member Cees Haringa Join Date: May 2013 Location: Delft Posts: 607 Rep Power: 0	As I said, both the numerical approximation and the analytic solution make certain assumptions which cause their solution to differ. study the assumptions you make in both approaches, compare what their effect on the found profile might be. But remember, neither approach gives a perfect answer.