Learning Forum

[yew511]

Greetings, I wanted to calculate the Nusselt number of 3D microchannel. To do this, I need to have wall temperature and bulk temperature. For calculating the bulk temperature, I create several isosurface parallel to the inlet until the outlet. For wall temperature, I have no idea what is the correct way to extract the result. I have tried to create a line at the middle of the wall (between solid and fluid), but I notice there is variation of temperature along the x direction. Z direction is the length, Y direction is the height and X- direction is the width. I need to extract the wall temperature at each position (z = 0mm, z = 1mm, z= 2mm, until z=10mm) to enable me to calculate the local Nusselt number. Any recommendation to use which postprocessing techniques?

[Rob]

There's a wall temperature, I think that's what you're after. If you create an iso-surface of constant x (or y - you want to perpendicular plane to the surface) and select the wetted surface when you create the iso surface you'll get a line along the surface.

[yew511]

Is it using the polyline feature as showed in the image above? I wanted to extract the surface temperature of blue shaded region, and I think polyline well suit my needs because i can extract the data at different location which is what I demand. However, when I compare the area weighted average of the entire surface temperature with average temperature of all polyline I created, there is much difference. The area weighted average got 322 K while the average (sum of of all polyline divide by the number of polyline) got 318 K. In this case, it seems that it is not correct to use this method to extract the surface temperature. Any insight how to get proper temperature

[Rob]

The polyline will be a "line" so it depends on how different the surface temperature is relative to the single line. If you want the values at set positions (length ways) create a serious of isosurfaces of mesh but make sure the blue surface is picked in Fluent when you create the isosurface. That will result in a surface that's the intersection of the blue surface and isosurface: actually it's a line, but Fluent thinks it's a surface, don't ask.

[yew511]

The white colour is the isosurface line that I have created.
The area weighted average of all these values are 319,71..
But the average weighted average of the blue region (interface liquid) is 322.82.

Here, the average all lines and average of the interface get different value. Hence, is there any way to extract the results such that it would give a nearer reading. It is because this will affect the Nu number calculation because wall temperature is different when using lines from isosurface and from interface liquid (blue region)

[Rob]

The line is on the surface. The blue face (possibly the inlet) is the whole channel at that position. I would expect them to be different.

[yew511]

Hi, the blue interface i mentioned is the previous image published during 29 April

[Rob]

The image on 29th, the blue surface is the wetted wall. On 30th it's an inlet.
If you're averaging an average you're also missing if the surface temperature change is uniform, if it's not then an average of point/line values may not be representative.

[yew511]

Yes, I would like to get local temperature (which is at different location) from the wetted wall as shown in the image on 29th The image on 30th, I show some portion where how i extract the surface temperature which is using the isosurface with intersection method (which results in the white "U" shape at certain distance from the inlet).

First method
I give some distance for example, each 2.5 mm until 10 mm, which is about 40 isosurface to extract the surface temperature of the wetted wall.

Second method
Then, I compared the surface temperature directly taking the area weighted average of the wetted wall.

Currently, i required to use the first method because i need to get surface temperature at different location. But, when compared to second method, the deviation in surface temperature is about 4 Kelvin which is quite large.Hence, I am finding reason why the surface temperature using first method is much different compared to second method

[Rob]

Plot contours on the surface and see how it looks. Plot with node values off. Both methods are reporting what is correct for that definition of area average, but is the average of the average the same as the overall average? There's probably a very complicated statistics lecture on this somewhere.....

[yew511]

Hi, when i plot the countours, I only can get minimum and maxmimum value of temperature, I cannot get a average value of the selected region is it?

[Rob]

Correct, but we're not just interested in the average, we want to see what is going on. In an experiment you may only get point values because that's where the instruments are. In CFD we get data at every cell centre: as you want to understand what's going on we need more information.