Learning Forum

[mehdiyahia1]

Hello, I'm trying to do a simulation of conjugated heat transfer, the source is a 10*10 m2 cube located in a building of 50*50*20 m2, the building is in the atmosphere, my study is stable,realizable k-epsilon, when I checked the results I see that I do not have a semmytey around my cube, if there is an idea to improve the result, I would be very grateful.

[mehdiyahia1]

[mehdiyahia1]

[mehdiyahia1]

[Rob]

Why do you expect to see a symmetrical result? 

[mehdiyahia1]

Thank you for the reply Rwoolhou

I have unconvincing results when I use the symmetrical. 

[Rob]

Have a look at von Karman vortex street on Google: you may not have a symmetric flow. 

[mehdiyahia1]

I think  Von Karman vortex street, is only for the transient state regime

[Rob]

Yes and no. How good is your convergence? 

[mehdiyahia1]

For the convergence i have 0.1 for all except  epsilon

[mehdiyahia1]

Yes and no.  what does that mean?

[Rob]

It means we can have a transient effect in the results when using the steady solver: it's usually identified by the way the residuals behave.  Just curious, are you using the University licence? 

[mehdiyahia1]

Yes i used the university licence . Are you have  some advice explains this phenomenon ?

[Rob]

Can you post the residuals plot as it's easier to explain if I can see that. We're very limited on here due to export laws, so the more images the less trouble I can get into! 

[mehdiyahia1]

[Rob]

Perfect, thanks. 

I'd be more concerned with the temperature limits if you're not expecting 5000K in the domain, 1K may also be a concern. Open up Solution Controls > Limits and increase the turbulent viscosity ratio by about 4 orders of magnitude.  

You've also not run anything like enough iterations to converge the solution: run another 500-1000 and let's see how it looks then. 

[mehdiyahia1]

I already have a turbulent viscosity 1.

[Rob]

No, the viscosity ratio limit, not the boundary condition. 

[mehdiyahia1]

[Rob]

Yes, that's showing a transient result: or more precisely that the solver is seeing many correct steady state results. Now compare the result from this solution with the plot you added earlier on. 

I think there's a good solution on our (customer side) system and add the first 10-ish lines of the licence file here so I can find the local ANSYS contact. 

 

[mehdiyahia1]

It' s impossible to obtain the university license.

[mehdiyahia1]

I don't know if there's any other way you can help me.

[Rob]

Very simply, you have a solution that's not steady state. You can either use monitors to check the values of interest are fairly stable or switch to transient and use monitors to see how the values of interest change with time. 

I can't advise on which approach is suitable as I don't know exactly what you're doing and what you need from the results. 

[mehdiyahia1]

I want to know the temperature distribution in my domain.

[mehdiyahia1]

And the height that reaches the plume.

[Rob]

And if you monitor those values do they fluctuate?

[mehdiyahia1]

Excuse me, but what you mean by monitor?

[Rob]

In Fluent we can monitor (look through the list of commands in the tree - left side of the interface) data from within the solution. This is often done to check convergence (steady state calculations) or trends in transient. Since you're using 2019Rx click on "Help" and it'll take you to the documentation, from there you'll find the Fluent tutorials. Do the first few as one of those will have monitors as part of the set-up. 

[mehdiyahia1]

I have tried all your advice but I am still far from good results.
Here's my job. https://yadi.sk/d/OkivnVFPTUVtIw

Can you help me?

[Rob]

I'm not permitted to open/download attachments. Please post images, and explain why they're not what you're expecting. 

[mehdiyahia1]

Now, I use the method of symmetry.
In the first image, I have a discontinuity in the temperature values.
In the second image this discontinuity, nor that the temperature is 290, is physically inexplicable.

if I'm right, how can I explain it? 

[mehdiyahia1]

[mehdiyahia1]

[mehdiyahia1]

350 K

 

[mehdiyahia1]

290 K

[mehdiyahia1]

The temperature in the cube is 2000 k

[Rob]

The surface looks like you're picking up the mesh facets, have you checked you're fully resolving the flow? 

[mehdiyahia1]

How can i do it?

[mehdiyahia1]

"The surface looks like you're picking up the mesh facets" I don't understand what you mean?

[Rob]

If you look at the isosurface there are hard edges/features in the shape. This tends to imply the surface is following the mesh, or that the mesh isn't well refined. 

To check the flow is resolved you either need to know what to look for or do a mesh dependency study.  Note, using symmetry will prevent/alter eddy shedding so you may find you don't get as good agreement with the experiment. 

[mehdiyahia1]

This is my mesh.

[mehdiyahia1]

As I understand it, work with the symmetry it doesn't recommend in my case.

[Rob]

That'll explain some/most of the problems. None of the zones are connected (you will have contacts which is OK, just not ideal given how simple the geometry is), and more critically the jump is cell size is awful. Have a look in the documentation tutorials and remesh the model. I suspect the results will be a lot better once you've fixed the mesh. 

Note, having all good mesh metrics usually means the cells are OK. It does not mean the mesh (ie all the cells) is any good. 

[mehdiyahia1]

You're advising me to use ICEM for the mesh (my knowledge is very weak in the mesh).

[Rob]

No, I'd stick (unite) the volumes together and use Workbench or Fluent Meshing with the watertight geometry workflow. 

https://www.youtube.com/watch?v=zFsa1Ezot8Y

ICEM CFD is very good for hex meshes, but not so straightforward to use. 

 

[mehdiyahia1]

Thank you very much sir for this helpful video,
I used Fluent Watertight geometry for the mesh, but I have a problem, I don't have my source in cell zone conditions, in return iin Workbench after DM and Mesh , this problem doesn't exist.

[mehdiyahia1]

With Fluent Watertight geometry

 

[mehdiyahia1]

With Workbench

 

[Rob]

Check the zone definitions in Fluent Meshing: they may need defining during the meshing process.