Learning Forum

[Mohamed Abu-bakr]

I am simulating flow in a pipe and constant uniform heat flux, I choose to module liquid (Fluid) without pipe(Hollow tube), When I calculate the temperature at the liquid surface using chart the values are different from what I have noticed on the contour. Also, I have used the probe at the values are similar to chart and different from the contour however the values on the contour is the correct one. 

The values on the contours are around 306.762 K toward the outlet as shown the following  figure 

and from the chart, it is around 295.5 as shown in the following figure 

The full project file is available on  the following link: https://drive.google.com/open?id=1qnVJjSBAsOhjpc5VYB5EOYGq0Tdc5kqP

[Rob]

ANSYS staff can't download files or attachments, so we'll need a few more pictures. 

Please can you plot the line on the contour plot too? Also plot the mesh so we can see the resolution. 

[Mohamed Abu-bakr]

This is the line relative to the contour (Line in black)
 

The line is generated by polyline in cfd post as the intersection between zx at y=0 plane and the liquid surface.
This is the mesh 

[Rob]

What are your wall boundary settings in CFX? 

Please create a new line that runs up the centre of the domain and plot that: I suspect the results will be very similar. 

[Mohamed Abu-bakr]

My wall BC are as shown 

The chart and the contour actually show the same result at the center.
The temperature in the pipe core is constant and equal to the inlet condition 293 K as the thermal boundary layer is smaller compared to the 60 mm diameter.
This is the line relative to the plot (again in black)

And this the chart

[Rob]

Can you refine the near wall mesh and continue the run: looking at the image of mesh & temperature contour it looks like only the near wall cell is warm. 

[Mohamed Abu-bakr]

Sorry but the solution seems to be temporary when I made another mesh for different turbulence model the problem occurred again.  

[Mohamed Abu-bakr]

I am trying to use k-e standard with scalable wall function, Targeting y plus 30

Free stream velocity 4.5

Density 1000 kg/m^3

Viscosity 0.00099 Pa.sec

Diameter 0.06 m

and these my inflation layer setup

 

 

 

 

[Mohamed Abu-bakr]

Same case same mesh same turbulent model but in fluent as the result is matching.

 

 

[Rob]

How's the convergence? 

[Mohamed Abu-bakr]

All residual are below 10e-7 and Energy is below 10e-7. Also I am monitoring the average wall temperature and it doesn't change much by the end of the run however it is wrong.

[Rob]

And if you plot everything in Fluent with node values deselected? 

[Mohamed Abu-bakr]

Fluent results are correct even in the monitor, when I said that monitor is wrong I meant in CFX, it is wrong but the solution has converged and monitor values became constant but not correct, that what I meant for the cfx not fluent. 

[Rob]

OK, so Fluent is consistent - that's the main thing to clarify. 

Fluent is a cell based solver and CFX is node based. CFD Post is what was CFX Post so when data is transferred across from Fluent there is some interpolation of the data: the different is usually negligible. I tend to use Fluent for all numerical reporting and CFD Post for images: the latter is becoming rarer as Fluent's graphics are improved. 

[Mohamed Abu-bakr]

CFD post with fluent is giving good results.
But CFD post of CFX is what seems to fail.

I have noticed that for CFX results in CFD post
When I change variables from conservative to hybrid it gives better values 

 

However based my understanding of this: https://www.sharcnet.ca/Software/Ansys/16.2.3/en-us/help/cfx_ref/i1298886.html 
The opposite should be the case.

[Rob]

I'll ask a colleague: I don't use CFX enough to be able to answer reliably.