Need help for LED luminary thermal simulation

[alexandwirtz]

Dear community, I am trying to make a thermal simulation for an LED luminary but don't get any useful results. Is there someone with some experience in this and could help me set up the simulation?

Attached is a pic of my results. Applying a radiation load fails the simulation.

I can give access to the design via private message.

Thank you,

Alex

[chrisplyler]

@TrippyLighting are you around today? This question might be right up your alley.

[TrippyLighting]

@chrisplyler I've been thinking whether this would be useful for me but have not had the time to look into it. Thanks for tagging me!

However, @brianrepp might be able to connect us to someone from the Fusion 360 Simulation team the can answer this question with authority as I am also very interested in this particular area of simulation.

[John_Holtz]

Hi @alexandwirtz

You mentioned a radiation load, so the thought that comes to my mind is what happens with other types of loads? In other words, is the problem related to the radiation load, or is the problem related in general to how the model is setup with other loads. For a steady state analysis, you need the following:

1. a source of heat, which in a case like this would probably be a heat generation (Watts per volume) in one of the parts. 
2. a way to remove the heat, which includes conductions to the other parts, convection to the environment, and maybe radiation to environment.

Other factors that need to be checked include the contact between the parts, material properties, and mesh size.

I will contact you separately to get a copy of the model.

[Anonymous]

Did you define contacts or convection? I just wrapped up a course on this and could use some early feedback to see if it's thorough enough. If you have time...heres a free link:

https://www.udemy.com/fusion-360-thermal-simulations/?couponCode=THERMALFREE

You will probably care most about the examples related to the circuit board/electronics. It's a beginner/intermediate course if you already know thermal simulations and obviously f360 doesn't include the CFD element that a higher end solver does.

[John_Holtz]

Hi @alexandwirtz. I think that @Anonymous is correct. The problem is that the parts are not connected together, so the heat applied to the LED cannot conduct to the other parts of the model that include the convection and radiation load. That is what the "Solve > Pre-Check" is indicating when it lists that the "model contains 4 thermally unconstrained groups".

It could be that the parts have a gap between them, or the parts have a large interference (so they are not "touching").

• If there is a gap, you can use an offset bonded contact. Think of this as filling the gap with a material that conducts the heat from one face to the other face.
• If there is an interference, you can use the "Simplify" environment to remove the interference. This brings the part into contact (a 0 gap), and the bonded contact conducts the heat from one face to the next.

The other thing that I noticed is your radiation setup. Radiation is a load from the face of the model to the surrounding environment. Since you have the radiation load applied to faces inside the volume, you are indicating that heat is radiating through the surrounding material to the ambient environment of 30 C. If this is what you wanted to do, then it is okay. Unless the outer part is transparent to radiation, heat transfer will not happen in that way in reality. (If you were hoping to calculate the radiation from one part to another, Fusion does not have this capability.)

Let us know what happens when you make the changes.

[alexandwirtz]

Hi together,

thanks for all your help. So I was able to get the simulation going when I simplified the model a lot and adding an offset-contact. Building it up again, I added one part that had interference (thread) with the other, removed the volume, the pre-check was fine but the following error occurs:

Error: SINGULARITY DETECTED AT GRID 3826 COMPONENT 1
Cause : A singularity has been detected in the stiffness matrix during decomposition.
Action : Investigate the degree of freedom at which the singularity occurred for a lack of
stiffness. Also, check your model for a lack of constraint. If using shell elements,
either set the K6ROT model parameter to a value between 1.0 and 100.0, or set the
SHELLRNODE model parameter to ON. If this does not solve the problem or you are unable
to locate the source of the singularity, consider setting the SOLUTIONERROR model
parameter to ON. Note that solution accuracy may be degraded with this parameter set
to ON. Check the EPSILON value in the Model Results Output or Results Summary File.
Values greater than 0.001 indicate a possible loss of accuracy.
Error: Solver Error
Error: An error occurred while solving the model.

about that I have no clue what it means. Any suggestions on that?

Thanks for your input!

[John_Holtz]

Hi @alexandwirtz

The singularity error is trying to indicate that there is a part that is not connected to the rest of the model, so the loads cannot conduct into or out of the part.

The following is just a guess. Maybe the pre-check is based on the CAD model, so it is able to detect when two parts are in contact even if the contacting area is small. But the analysis is based on the mesh. If the mesh size is large compared to the size of the contact surface, there may not be any nodes in contact between the parts, like this sketch. This would cause the analysis to miss that the parts are in contact.

If this is the problem, increasing the "Maximum Activation Distance" for the contact pair should fix the problem. This will connect the nodes on part 1 within X mm to the faces of the elements on part 2. Since the problem started when you added the last part, the contact pairs involved with that part need to be edited.

The other possibility is that a distorted element is causing the problem. (The solver only knows that it cannot solve the matrix of equations. It does not know why the matrix cannot be solved.). If you do not see any problems with the contact between the parts, try a finer mesh on the model.

[alexandwirtz]

Ok, thank you for the hints, I will give it a try. -Meanwhile I got this phenomenon that the colour of the results do not match the temp. shown. Is there something I can do for that?

Thank you!

[John_Holtz]

Hi @alexandwirtz

It is hard to tell from the static image if the max and min markers are pointing to something on the visible model, or are they pointing to components that have been hidden.

There is an option to control whether the legend shows the range based on the entire model, or the legend shows the range based on the components that are visible. Click the "Legend Options" button (the gear symbol) and change the checkbox for "Visible Only". Does that change your display?

[alexandwirtz]

Hi John,

changing the visibility did not change the displayed colours. But I found that when I sliced the simulation result and zoomed into the LED that here the colour gradient was showing. So there was some bad geometry inside the model which I changed. Also the LED had contacts that where not necessary (gaps with 0,1 mm) and maybe also caused the problem. So this now works fine. 

the luminary I still need to check.

Thanks!

[John_Holtz]

Hi @alexandwirtz,

I wanted to check with you to find out how the simulation is going. Do you have results now? Do you have any follow-up questions? 

Please let us know.

[alexandwirtz]

Hi,

thanks for your replies!

so it is not working too well for me, the offset contact does not conduct any heat, although i played with all the settings:

only when i thicken the inside diameter and fill the gap in simplify mode it works (but now it is an ideal conductor which is not the real case):

Any thoughts on this?

Thanks,

Alex

[John_Holtz]

Hi @alexandwirtz

The result with the offset bonded is unexpected.  I just tried an analysis, and it appears that the offset bonded either does not work, or maybe it works but not as I expected. I will try some more tests to see if I can determine what it is doing.

However, your statement that bonded contact behaves like an ideal conductor is not entirely true. You can define a contact resistance (actually, the inverse value as "Thermal Conductance") which will create a temperature change from one part to the next. This input may give you the effect that you want. Depending on what the temperatures are in the results, radiation between surfaces may be what you really want, but Fusion does not have that capability.

[John_Holtz]

Hi,

I did some more tests and convinced myself that the offset bonded contact is not working in a thermal analysis. I have reported this issue to the developers.

Removing the gap between the parts and using bonded contact is the best solution to this issue that I can think of.

[alexandwirtz]

Hi John,

thanks for your testing and report to the developers!

Alex