SimScale CAE Forum

Attached solids with interfaces for CHT



Newby to Simscale and have freeCAD and Salome to create stp. Anyone have a sucessful step by step method to create solids attached by interface with these softwares.

thanks, Joe


Hi @jbielecki!

Did you already upload a file or at least can show a sketch to give us an idea of what you want to achieve? We will then try to help you to create your geometry as well as running your first simulation.





I have uploaded many times with little success in created just the right solids and shells. You can see one of my failed attempts in LEDbrdbox project I made public. Looking for any suggestions. My goal is PCB powered CHT in air.



Here an image too. Just two solids to start. One fluid ‘box’ and the other the PCB solid.


Currently running a mesh and see what can be done here @jbielecki - will keep you up-to-date.




Hi @jbielecki!

With which software did you create the part? Could you thicken the outer part of your LED as well as allowing some overlap to see what influence this will have on the mesh itself. Want to give that one a try.

We will definitely get this one done :+1:





Thanks for the effort. I have tried Salome many times. It seems to have more tools than freeCAD, but maybe I’m wrong. Maybe you know of another ‘cheap’ preprocessor software that can work better. Yes if you are referring to the board thickness, that could be doubled or tripled in thickness.



Hi @jbielecki!

Salome is definitely a power tool but you can also use Onshape for this purpose. The advantage is that you are way more efficient when you make adaptions and use the import function inside the workbench. Let’s thicken the outer part a bit and try that approach.

All the best!



Do the solids need to be uploaded as one file or can each be imported separately? I can do this, but every time a try to partition a solid in Salome and export, importing to Simscale I lose watertight and solids.


It would be better to have a partitioned solid. Do you mind trying Onshape?

What you can also do is to directly upload the mesh (as UNV, OpenFOAM or MED) if that’s possible in your case.





I successfully imported stp with three solids watertight and ran my first CHT solution of a small PCB with and LED of 1 W heat load. The solution is a little high. I was looking to include radiation heat transfer. Does this solution allow this? Thanks


Hi @jbielecki!

You could use an additional boundary condition (custom) defining a heat flux if you like. In addition to that you can try running a structural heat transfer analysis working with table uploads defining your convection over time (thus transient analysis) - the effect is not identical though as the convection is uniformly distributed over your component(s).

Can you also tell me what you mean with “little high”? Have you done a hand calculation or an experiment to know what you would expect in the simulation?




Yes correct.

I have modeled and tested a slightly different geometry (same surface area though) and substrate is little more complex (multi-layered) but overall should be similar conduction and convection. I would like to have the ability to include radiation (i.e. diffuse or even specular). Other than those ‘custom’, options no optical properties for faces per say?


Hi @jbielecki, I don’t think that kind of option is available. I have in the past used the software in which radiation was calculated as you mentioned by defining reflectivity and emissivity to faces and doing a pre-calculation of exchange between the faces. Unfortunately, I can imagine a viable workaround, however, you are welcome to add this requirement to our vote for features section detailing your use case and how you imagine it working.

Of course, if anyone else has an alternative please correct me.

Sorry, I cant be of more use,


Ok I’m back with an uploaded LED board on a finned heat sink in air box. I have run it several times looking for results that give a realistic delta T distribution on the assembly and have not been successful. It’s been 30 days and my initial trial account is over and now my project does not run. It gives a IO error.

I have tried fixed valued BC but I think zero gradient is probably more realistic for the problem and cannot seem to run the solve long enough to reach convergence. I have run it as long as 5000 sec and still my control surface temperature is still rising.

What is wrong with my project? Was it altered because my account status changed?



Hi @jbielecki, it is very important to define initial conditions, even if they are ballpark figures. If there is no delta T it takes a while for the heat to flow to the rest of the system.

Also, you shouldn’t need to define the interfaces, that should happen automatically (possibly the reason for your latest error).

Let me know how it goes,