Velocity Curve Plot is not constant

[Anonymous]

I have selected adv 5 for this problem. Forced convection inside and natural outside. Max plot of temp is smooth but all velocity and pressure plot has jig saw profile and taking lot of time to solve as it is solving HT And flow both . Temperature is coming down very slowly. Mesh is also fine and given finer for fan.

Can u please help me in this

[Jon.Wilde · Accepted solution]

Looks like a nice setup, cool.

I have a few thoughts and my main one is that the fan curve might need a little modification. Right now it has a region where you can have two different flow rates for the same pressure drop. 
Could I suggest smoothing that out a little?

Lower the velocity and pressure relaxation parameters a really small amount (turn off ISC to do it then back on again).

I would probably also use the mixing length turbulence model.

I have this running smoothly without tweaking the fan curve, just the other changes.

Do you need radiation on too?

See how that goes?

[Anonymous · Accepted solution]

Can you pls share modified model?
Secondly, is k epsilon not a good choice? Which model you used?
Which turbulence model you are suggesting?
Do we need to specify mixing length too?
Can you please specify guide for relaxation parameters too?
Will radiation contribute to lower down the temp, i was thinking it will create problem for convergence and will not contribute much.

[Anonymous]

Can you pls share modified model? Secondly, is k epsilon not a good choice? Which model you used? Which turbulence model you are suggesting? Do we need to specify mixing length too? Can you please specify guide for relaxation parameters too? Will radiation contribute to lower down the temp, i was thinking it will create problem for convergence and will not contribute much.

[Anonymous]

Yes I need radiation on too. I have given 0.9 EMISSIVITY (ref temp 55deg) to PCB surfaces, and for air 0.3 ( created new air material) (As per infor in Autodesk CFD, emissivity value of solid will override air value,hence air value will not have that much effect as it will take attached solid values). For other materials i kept default values.

Modified curve by providing descending value to flow. Velocity & pressure control to 0.49 from 0.5. Under Solver control selected radiation ON.

I started analysis, but still velocity and pressure is tweaking, but its just 190 iterations (hoping it will get converged). Apart from this temperature is almost stable now and decreased also but it seems temperature is decreased significantly. But I doubt about temp distribution.

By the way I refer to max value in plot for tweaking , or we should look at AVG value.

I am really thankful for your help.

[Anonymous]

What should be ref temperature while providing Radiation boundary condition, i kept it ambience temp 55 deg C. Is it OK?

[Jon.Wilde]

Hi,

ke is OK, I used mixing length as you have both natural convection and forced convection here. Feel free to test both.

For the relaxation parameters, lower them a small amount, exactly as you have. I probably lowered them more than was necessary as I went to 0.45. If 0.49 converges smoothly, go with it as it will be faster.

It sounds a lot like you assigned internal boundary conditions to your solid surfaces? These all need to be removed - the emissivity is control from the material settings.

As we said before - we must avoid internal surface boundary conditions 😉

You also need to be really sure that the material properties are correct, otherwise the radiation calculation can throw the thermal results. So check the emissivity values of each part match reality.

If you have air emissivity (within the air material) set to 0.3, that means that 70% of the radiation will bounce back at the walls of the domain where there is no solid. I would use a higher value 😉

I ran this in CFD2018 by the way, just incase it makes a difference - it might actually. Are you on the latest service pack of 2017 at least?

The convergence looks pretty smooth, I would guess that any bumps are down to needing more mesh than anything else. These sort of models should not provide CFD with any sort of difficulty.

To add a comment to something you said on another thread (where @Anonymous is helping and is also totally correct) - you need the film coefficient on the very top surface so that CFD knows what the air temperature is of air that is entering - I guess you have some recirculation here...

Hope that helps,

Jon

[Anonymous]

What shud i keep for air 0.7 or 0.9

[Anonymous]

Thanks John for continuous help.

1. Does it mean that we should not provide Radiation BC to any surface? Though it same as material emissivity. Will it change results.

2. If there is no circulation, still we have to give film coeff on top of ext volume @ ref temp of ambience. But it will be along with 0 guage pressure or should be only film coeff.

3. Bottom surface of external environment will have 0 guage pressure and ambience temp. Other walls should also have BC of temp only . Am I right?

4. What should be the good guess for Air emissivity shall i go with 0.8 for all air volumes including external air?

5. Other materials have default emissivity provided in your database. Is it OK?

[Jon.Wilde · Accepted solution]

Yes, stick to the default, which I think is 0.9

To answer your questions, here goes 🙂

1. Does it mean that we should not provide Radiation BC to any surface? Though it same as material emissivity. Will it change results.

Yes, do not do this, it will cause you problems, just use the material definition.

2. If there is no circulation, still we have to give film coeff on top of ext volume @ ref temp of ambience. But it will be along with 0 guage pressure or should be only film coeff.

Both, P=0 and Film Coeff, just good practice and it means you do not need to worry if there is recirculation

3. Bottom surface of external environment will have 0 guage pressure and ambience temp. Other walls should also have BC of temp only . Am I right?

Up to you, personally I usually do not assign anything to the side walls incase it lowers the temperature artificially. It depends how close they are to the model - my approach is to leave them without

4. What should be the good guess for Air emissivity shall i go with 0.8 for all air volumes including external air?

0.9 as default

5. Other materials have default emissivity provided in your database. Is it OK?

Yes and No. If the emissivity is the same as your real world materials, all is OK. If it is different then you need to change it. CFD has default settings but they cannot account for customised finishes or different paint colours etc. It depends what you have in your model 🙂

With radiation this can have a huge impact so be sure that the values in CFD are the same as your actual model.

[Anonymous]

Many thanks John,

Does it mean that we can create short External volume for natural convection problem, and apply film coeff of 2 to all sides? 😉

[Anonymous]

Ok. I have considered 0.9 for all air volumes including internal and external.

[Jon.Wilde]

No, what you have is fine. Please do not keep making changes that move you backwards 😄

Smaller can mean you artificially accelerate the flow around the unit.

[Anonymous]

I made it 0.45, but after 150 iterations it started tweaking. Its 750 iterations till now and on 3rd day run is still ON.

Temperature is almost stagnant now (with radiation ON), TKE TED is almost constant, but vx vy vz and pressure is still tweaking. I look at the MAX value plot. But Average value plot is not having any tweaking. Do we need to refer AVG or MAX plot for tweaking?

[Anonymous]

Hi Jon,

I  started run with 0.48, but it is taking long time from last 4 days it is running and still ON. Radiation is also ON.

Is it due to radiation... because temperature distribution keep updating but max temp is almost constant.

In between Vel keeps tweaking...observed in detail it has very small tweakings. Any suggestion to expedite the simulation process. I have 36 cores computer and it seems it is utilizing all.

[Jon.Wilde]

Might it just be that there are some small transient features in the natural convection domain?

From what you describe, I would guess so and would just stop it now and use the results.

This should only need a few hours of run time.

[Anonymous]

There are no transient features its simple boundary conditions to external air as you suggested.

[Anonymous]

It seems that 2 internal fan flow may be the problem or radiation on as it solves for radiation every iteration. It is the same model with but with emissivity and radiation on . Air emissivity 0.9 . I will share model tomorrow.
Its same model with BC discussed earlier.

[Anonymous]

Sending you files. Please chk still it is running. its 6th day.

CFD Version 2017

OS : WIN 10

Pls help.

[Anonymous]

It seems velocity & pressure is a problem...shall i change relaxation param to 0.2 for pr & vel?