Natural convection in oil filled enclosure
 

Hello everyone,

I am simulating a problem involving natural convection inside an enclosure with internal heat sources. The enclosure is filled with oil. The heat is dissipated through the enclosure walls (outer wall with heat transfer coefficient and ambient temp was defined on the external enclosure surfaces).
The main problem is that it takes a very long time for the oil temp. to reach SS (the time constant is app. 13000 sec). I have tried flow freezing but gets inconsistent results.

Is it possible to use flow freezing for natural convection (I used the periodic approach to allow the temp. field to effect the velocity field and vice versa)? What is the recommended approach for solving natural convection problems with very different time constants? Any help/advice will be appreciated.

Hi Yoav,
What do you mean by "time constants" for an steady state simulation?
You are talking about steady state simulation, aren't you?
Flow freezing is speeding up the solver a little and yes, you should use the periodic option in order to let the flow adopt again after some time.
Natural convection can take a while but 13000 seconds sounds more like a transient simulation which will usually take longer than steady state simulations. What do you want to get from your simulation?

Regards,
Boris

Further details
 

Hi Boris,

I have a heat source inside a oil filled enclosure. the oil circulate inside by natural convection only. heat is dissipated through walls. I want to get the heat-source and oil temperatures (and corresponding heat transfer coefficients).

by "time constants" I mean that if I multiply the oil thermal capacity by the thermal resistance (sum of oil to wall resistance+wall resistance+external wall to ambient resistance), tao=RC, I get a app. 13000 sec. Which means that it takes a lot of time for the oil temp to reach SS.

When simulating in SS, the heat-source temp rise reached SS after only 20-40 iterations and the oil temp. almost did not change. then the oil temp starts to rise very slowly and it can take several thousands iteration till SS. I tried to specify a varying time step (app. linearly increasing with time) but the solution does not converge.

what is the best way to determine the freeze and no-freeze periods? should I specify a manual time steps (for both freeze and no freeze)?

I don't really need the time history of the solution, only the final results.

thank you in advance.

Hi Yoav,

I understand now what you mean by time constant. When you say "varying time step" I assume you mean the freezing option steps which are not time but iterations or travels in a steady state simulation. That might be confusing.

How big is your heat source?
Yes, convergence might take some time but 40 iterations tell me that the mesh isn't that big and the geometry isn't that complicated. You should refine the mesh on the heat source and the sourounding fluid to better resolve the convection flow around the source. Then also apply min, av and max goals on the important parameters and components you are interested in. The temperature change might be very slow in some applications so you should watch the convergence graph if it is still rising and if it is still to slow you should turn off the goal convergence and maximum travel criteria for the solver to stop. You can do that also in the calculation control options. That way the solver will not stop if the temperature change is too low to be recognized by the automatic convegence criteria definition or stop due to the maximum of 4 travels is reached. you can also manually set the convergence criteria if it is too high. But in such cases it is better to watch the curves and maybe also display the delta value in the table of the graphs. To display that you can right click on the table header line and add that parameter. This shows the detla which has to be smaller than the convergence criteria for the solver to consider the parameter as converged.
Sometimes also radiation cases can run much longer to really converge.

I hope this helps,
Boris

Thanks
 

Hi Boris,

Thanks for your reply, it is greatly appreciated.
Any guidelines for setting the flow freezing parameters?

thanks

That is hard to tell. Depending on the solving time or Iteration per travel you should wait till the flow has basically established and then you can start freezing and depending on the same parameters the periodic freezing has to be adjusted.
I cannot tell the settings exactly without knowing the model and these parameters myself.

Regards,
Boris