CFD Online Discussion Forums

CFD Online Discussion Forums (https://www.cfd-online.com/Forums/)
-   FloEFD, FloWorks & FloTHERM (https://www.cfd-online.com/Forums/floefd-floworks-flotherm/)
-   -   Enclosure Heat Transfer (https://www.cfd-online.com/Forums/floefd-floworks-flotherm/188548-enclosure-heat-transfer.html)

jjnud June 1, 2017 14:29

Enclosure Heat Transfer
 
1 Attachment(s)
Hello, I'm very new to FloEFD/CFD, and am having trouble finding some information online relevant to what I'm looking for.

I'm designing an aluminum enclosure that houses 2 boards, each with a chip that generates 9W of heat, a heat sink on each chip, a power source that generates 24W of heat, and polystyrene insulation. I've defined the materials, the heat sources, and the goals as max temp of the chip and convective heat transfer rate from the heat sink.

Below is a picture of the design. My question deals with how to handle the boundary conditions. I've defined ambient temp outside the enclosure to be 40 C, and the exterior walls to have a convective coefficient of 10 W/m^2/K. However, I also need the simulation to account for convection inside the box. If I set up boundary conditions inside the box so that all surfaces (chips, heat sinks, pcb's, and insulation) are real walls, and there is a convective coefficient of 5 W/m^2/k, I also need to define the temperature inside the box, which I don't know, and rather wouldn't assume.

Is there any way to leave the fluid temperature inside the box undefined, but still account for convection inside the box? Any other suggestions, or relevant resources online for simulating heat transfer inside an enclosure? By the way, I have it set to external flow, while including internal components. I believe that's the correct configuration.

Thanks in advance!

Boris_M June 29, 2017 04:23

Hi Joseph,

If you run an internal analysis, the external heat transfer coefficient is a good way to model the convection on the outside of the box without actually simulating the outside. But for the inside, you don't need that if you want to consider convection, as soon as you activate the gravity with the correct vector setting, the convection is automatically taking place based on the physics of the natural convection (warm air has lower density and rises etc.).
If you specify real wall with a heat transfer coefficient then you define that as fixed and it is not calculated by the solver based on the actual flow.

The temperature inside the box does not need to be defined if it is a steady state simulation. You can leave it with the default initial contidions the simulation will start with. Over the iterations, the temperature will develop based on the heat dissipated by the chips etc.
It seems to me you have used other more traditional CFD tools where you usually have to define the wall of a body as a wall. For FloEFD this is of course normal. If there is a wall then this is a wall and it is not necessary to define it as a wall. You can apply specific properties to it if you want it to have some special properties such being an ideal wall or having a constant temperature etc. but a wall is a wall and doesn't have to be defined as such. The same goes for the flow. If you specify you have airflow (that means not selecting the "heat conduction in solids only" option in the wizard/general settings) then there will be air in the domain and the initial conditions are by default set to ~20C (20.05C). Changing that setting makes sense if you know that the everage will be at a different level at the end of the simulation which can reduce the time till it is converging or in case of a transient simulation where the initial conditions will be the conditions at t=0 of course.

Not necessarily, you can run an internal analysis for this as well if you can live with the simplification of this and at the most defining heat transfer coefficients for the external walls in the wall condition option in the wizard/general settings. If your model is water tight then you can do an internal analysis but you can of course also do an external analysis but then you don't need to define the outer walls with a heat transfer as that will be calculated by the solver based on the flow that develops.

I hope that helps,
Boris


All times are GMT -4. The time now is 23:01.