Learning Forum

[NatKoba]

I am new to ANSYS CFX and currently trying to demonstarate bubble assisted natural convection method of liquid immersion system. 

My current model consists of CPU and two cold plates immersed in a cooling tank filled with fluorinert FC3283.

The tank has a nozzel inlet at the bottom face for bubbles and top face set as outlet. 

CPU is set to generate 150W and Cold Plates are to absorb the whole 150W. 

I was able to demonstrate natural convection and monitor flucuation of CPU temperature and fluid velocity at different air flow rate and air diameter. 

However I believe the model needs something with fixed temparature as a reference value. 

So I created walls around cooling tank (fluid domain) and set those calls as solid domain and isothermal at 25C.

This changed the heat transfer setting of both CPU and Cold Plates; conservative interface flux is no longer an option. 

Could you please tell me how I can maintain the interface flux state of CPU and Cold Plates while adding an isothermal solid domain. 

Thank you very much for your time and help in advance.

Nat

 

[DrAmine]

Can you post a screenshot of the model/geometry from CFX and highlight the boundaries used.

[NatKoba]

Dear Amine,

Thank you very much for getting back to me. 

I am not quite sure how to provide the information requested in one image. 

So please allow me to prosent the geometry image of my model and also a list of boundary settings as following;

Cooling tank (the cubic box): Fulid domain (FC3283+Air),  Thermal energy

Walls (attached to 4 faces of the tank) : Solid domain, Isothermal @ 25C

CPU (the one in the middle): Solid domain, Thermal , Thermal energy 150W, Heat Transfer (Heat flux for now*)

Cold Plates (two on each side of CPU): Solid domain, Thermal energy -150W, Heat Transfer (Heat flux for now*)

Inlet (the bottom face): inlet, Bulk mass flow rate, Static temp @25 C

Outlet (the top face): outlet 

*when being set at interface flux, natural convection was observed. 

 

 

I hope the above gives you enough info. 

Kind regards,
Nat

 

[DrAmine]

Something to with the mesh you are using: on solid-fluid interface conservative flux should be used. To simplify remove the outer box and back to the working case: how are the outer faces modeled?

[NatKoba]

Good morning Amine, 

Okay so back to the cooling tank without any attached solid walls. 

That model has a cubic box (cooling tank) what is fluid domain with thermal energy. 

The defult setting is as following;  Wall boundary with no-slip and smooth wall that is adiabatic and uses volume fraction

And Interface is set as interface boundary with no-slip and smooth wall and use of volume fraction too. 

Its heat transfer as conservative inferface flux. 

Outer faces has an inlet at the bottom which is set as inlet with subsonic flow regime. 

Bulk mass flow rate ( air flow rate), normal flow direction to boundary condition, medium tubulence and static temp @20.  Volume fraction  air :1.0 and fluid : 0    

The outlet at the top of the outer faces is set as outlet boundary with subsonic flow regime. 

Average static pressurem relative ressure @ 0Pa and pres. profile blend, and average over whole outlet pressure

 

And again, I am trying to set something solid and isothermal for a referece value. 

Thank you for your help in advance. 

Kind regards,

Nat

 

 

 


 
 


 
 


 
 


 
 

 

[NatKoba]

Dear Amine,

Although I posted all the setting info of my model earlier, i might have been off the point. 

By your message, were you saying that the problem is due to the mesh I'm using?

Here is the mesh details;

Quadric

Default elemert size

Adpative Sizing 

Resolution 6

Slow transition 

Fine Angle Center 

Initial Size Seed: Assembly 

Check mesh Qualit Medium Sommthing 

No mesh metric

No use of Automatic inflation

 

I hope this time I answered your question. 

 

best regards,

Nat

[DrAmine]

Please Share pcitures of the domain abd BC's hard to read all settings.

[DrAmine]

And Why are you using multiphase flow?

[DrAmine]

You wrote volume fraction . 

[NatKoba]

Hi Amine, 

Earlier, I was confused multiphase flow with singe-phase cooling method. 

Yes, I am using the multiphase flow as the cooling tank consists of fluid and bubbles.

We are trying to see the effect of bubbles on natural convection.   

I will send you the image of settings later today.

Kind regards, 

Nat

[DrAmine]

in s nutshell stick to the working case. I do not think that having resolved solid outer walls will bring more evidence

[NatKoba]

Dear Amine, 

Following your advice, I am now working on the previous model which has no outer walls. 

I am trying to create something that is isothermal in the model to make it more realistic. 

Ideally, those Cold Plates (two long solid objects) should have fluid inside that is isothermal.

And then I'd like to monitor the Cold Plates' surface temperature as they extract the heat generated by CPU (the little object in the center).   

Is it possible to make those Cold Plates fluid domain with water inside that is set as isothermal while having thermal walls with interface flux boundary? 

If not, what are the alternatives?

Looking forward to hearing from you. 

Kind regards,

Nat

[DrAmine]

So the green bodies to be filled with water right? If yes then I would stick to conservative flux at the interface and do not assume any isothermal fluid. If you really want to do that remove the green bodies and just use isothermal wall

[NatKoba]

Hi Amine, 

Thanks for your quick reply!

Yes, the green bodies in real system have water running through and are connected to a heat exchanger (simplied in my model). 

Let me double check that with isothermal wall, conservative flux is still possible? 

Conservative flux is no longer an option as soon as I changed the outer walls into isothermal. 

So if I want to keep conservative flux, I can not have any isothermal object in the model, right?

Hugely appreciating your patience and help.

BR,

Nat

[DrAmine]

Still not clear: which outer walks are you talking about?

[NatKoba]

Amine, 

I meant outer box wall (Cooling Tank). 

I am trying to figure out a way to have something isothermal in the tank. 

Otherwise, the simulation is simply based on the heat transfer between CPU and Cold Plates. 

Kind regards,

Nat

 

[DrAmine]

The outer wall of the Big Tank can be set to whatever (convective boundary is the most realistic one) but what does that do with the other issue?

[NatKoba]

Amine, 

Convective boundary of outer wall !!! That's it! 

That still keeps the conservative flux for both CPU and Cold Plates right?

How do I set that?

Boundary details of the big tank doesn't give such option.

 

BR, Nat

[DrAmine]

Under Heat Transfer.

[NatKoba]

Amine, 

Under Heat Transfer, there are ;

adiabatic, temperature, wall heat flux, wall heat transfercoefficent, system coupling and fluid dependent. 

I should choose wall heat flux?

Kind regards,

Nat

[DrAmine]

What about reading the manual?

[DrAmine]

Wall Heat Transfer Coefficient should be used.