Hi Ewen ,
My initial thoughts are that you have some sort of convective circulation.
Its a little dfficult to know with the limited information.
Apply some flow trajectories to you heat source and have a look at the result.
Let us know how you get on
Yes, you're right. There are convective circulations being trapped.
I presumed the heat would diffuse out evenly.
What would be the right approach to negate these?
As one of my mentors used to say " .... never presume .... "
Difficult to do completely , heat will find a way.
Block any path between your heat source and the surface in question
Remove the heat some other way (heat pipes ?)
Again without more detailed info its difficult to advise
Sorry for the late reply. I've tried a bunch of things but still receive the same results.
I had wanted to model a common back lit LCD module.
Now this is a very simple model, The yellow is the panel and the brown is the Plastic Chassis and the Blue is the metal chassis.
On the inside I show on the first couple of pictures I posted the different sheets. Polariser, Dbef etc.
After removing any air gaps that I could visibly see, like you said, made the surface plot on the panel a lot better
But the other sheets are more or less similar to the first picture I posted.
I applied the flow trajectories again and see that there are still some convective circulations. I do not particularly want to modify this model due to the simplicity it is already at. I am currently modelling this with just a surface heat source plate, but later would like to add LEDs.
I am looking for the differences in temperature within the sheets
I have assigned the material PET to the sheets. I feel as though that poses a problem as the thermal conductivity and specific heat of the material is defined constant, whereas they would probably change in reality.
I would like to know if there is anything just fundamentally wrong with the way I am modelling this ?
Very simple models are fine for trying out initial ideas however to get accurate results you will need an accurate model with the correct materials / values for all the parts and then , of course , the correct setup of the CFD environment.
There are numerous things that can be done to further refine the model / setup
- Is this setup as an Internal or External analysis ?
- What are your boundary conditions and do they reflect the real life conditions ?
- Setup the correct materials (see below)
- Setup the correct part geometry
- How big is your mesh ? Too few cells or cells in the wrong place.
- Have you performed a mesh sensitivity study yet ? This is crucial if you want to trust your results.
- Can you apply the heat source more realistically ?
- Do you have results from a practical test of the real unit ? Useful for checking your simulation.
As far as the material you are using is concerned both the Specific Heat and Thermal Conductivity values can be defined against temperature in the Engineering Database , you will need to create a User Defined Material .
One thing I do whilst preparing a simulation is to write down two lists , everything I KNOW and everything I ASSUME about the part , environment and materials.
Very often the ASSUME list is longer than the KNOW list, thats when I start working on gathering information to move the ASSUME items into the KNOW column.
Thanks for the quick replies.
I agree that a simple model can be useful to find an initial understanding of the problem at hand.
I'm not particularly trying to find accurate results, but more so results that have the same trend as a real test that I have conducted, that is why I do not think that refining the model to be accurate would be useful for the time being.
Your list of questions are very helpful as a checklist, however I feel as I as I have fulfilled all the criteria.
All it is, is a box with different layers all tightly fitted within it. As you said the convective circulation is a problem, and cannot really be reduced.
What would you suggest (not the problem) a solution is to make all the sheets produce the uniformity I get on panel as shown in the last reply?
I think looking at it from this perspective enables me to see what I did differently to the current model.