I try to solve a model coupling thermal and flow
it s a very classical model
This is a box with fins inside,a pipe in and a pipe out, and air flow.
Power is applied outside the box at the bottom part.
boitier_ailette5.FCStd
Flow simulation only works (pressure in, pressure out) (option flow => equation convection set to none) not heat equation
Strange thing, i need to reverse pressure orientation in the dialog box, but it works. There is some velocity over fins inside.
Capture d’écran 2023-11-22 111326.jpg
Capture d’écran 2023-11-24 094321.jpg
Thermal simulation only works too (i applied a thermal convection coefficient on fins)
But when i tried to coupling heat and flow, no solutions no convergence or sometime it crashed.
in this case i want to calculate the convection coefficient on fins.
i tried difrent solver (BICGSTAB, ldrs) diffrent conditions, without success.
i tried to copy parameter in similar case
fem-elmer-tube-thermalflow-velocity-3d.FCStd
I tried to solve navier-stokes in first.
At this time it s very complicated to undertstand how parameters should be adjusted.
In a classical model we just need the thermal convection heat transfer coefficient to coupling heat and flow
Why it s too hard to calculate It here?
Is it possible to limit its range (ex 5 to 200W/m2K) ?
or is possible to use the flow result (without heat) as a input ?
If someone could help me it would be great.
Thanks
I would try refining the mesh, it’s rather coarse for this kind of problem. But CFD with FEM is risky, it might be better to give up on this approach and use OpenFOAM (CfdOF module plus some manual edits if needed) instead.
I’ve already worked with CFDOF to simulate the flow on this casing. The results were good (turbulent flow, not laminar like here). But what I’m looking for here is thermal and flow coupling. I saw in some examples that it was able to calculate the heat exchange coefficient, which interests me.
When I refine the mesh size (0.5 mm) FEM crashes because of the memory. But I can’t increase the mesh size beyond 1 mm, which is the size of the gap between the fins.
I’m having great difficulty understanding which parameters need to be changed for the simulation to converge.
OpenFOAM should handle that too. The ChtMultiRegionFoam solver can be used for this but it’s not supported in CfdOF so you would have to edit the solver files manually after doing the initial setup.
Another option would be to use OpenFOAM + CalculiX coupled with preCICE since it supports CHT too but it’s better to avoid such couplings between independent solvers when not necessary.
GwB : What kind of simulation do you want to perform? You want to simulate …
the heat transfer from the surface into a medium (air, water, …)
the heat transfer inside a solid body and the transfer of heat from the boundaries of this body into a medium (air, water, …)
The first simulation is fully implemented in FreeCad CfdOf. You must de-select “Isothermal” simulation in the physics model:
deselect isothermal.jpg
Now, you can enter a thermal boundary condition. These thermal boundary conditions are combined with “regular” boundary conditions, for example wall:
Thermal BC.JPG
For a thermal simulation, you have 2 ways to model the transition from the solid body into the medium: You can make a very, very dense mesh near the surface. This is not the usual way because in most cases you do not have a computer that can handle this mesh. For this reason, a heat transfer coefficient can be defined via the boundary conditions:
heat transfer coefficient.JPG
Currently for thermal simulation with FreeCad CfdOf, there is no good tutorial available, but I will record a tutorial in the next few month.
If you want to simulate via the 2nd option (heat transfer inside the solid body, the so-called “conjugate heat transfer”), you must edit the files generated from FreeCad CfdOf for OpenFoam. This is very difficult and really nothing for beginners. I am currently looking for ways to set up this simulation and then record a tutorial.
thank for your answer Raedchen and NewJoker
I will take a look on your informations, very interresting.
The aim of this simulation is to look at the efficiency of the fins, in order to modify the sizing if necessary, or change the flow. To do this, the exchange coefficient must be calculated and not imposed. It is the heat transfer in the solid coupled with the heat transfer in the fluid. This exchange coefficient depends on the fluid and its speed, and varies locally.
If you look at the example fem-elmer-tube-thermalflow-velocity-3d.FCStd attached to my first message, you’ll see that the fin exchange coefficient is calculated.
I don’t need a very fine mesh on the fins, but I do need to be fine on the air between the fins because of the small spacing (1mm).
I’m going to look at the solutions you’ve suggested, but I’m also going to try working with a simpler model to understand where the blockages are.
Hi
I’ve just done some tests with a fairly simple model, to try and make the flow model converge (Elmer and Flow), the result is simple when in the flow module we set equation to none and flow model to no convection, the model converges in a few seconds.
Capture d’écran 2023-11-30 113955.jpg
Now if I set the model to full and equation to computed, I never find any result (only inconsistent) even if I change the motor and set a uniform temperature condition.
What’s more, the result obtained is laminar instead of turbulent. The possibility of coupling thermics and flows is feasible on small models, I succeeded with a very simple model but as soon as it becomes more complex I think it’s a wasted effort.
What would have been nice would have been to do a simple flow calculation (even from openfoam-CFDOF), then use the result to estimate heat exchange on fins, and then run the thermal module and repeat on the flow module, or else start again with a complete model but with an input solution, but that’s another debate.
I hope that in a while we’ll have an open foam thermal model integrated into freeCAD-CFDOF, and that it’ll be more relevant than Elmer.
thanks
When the model is more complex, you will not know the physis of your flow. For turbulent flow, you need to use the right turbulent model to suit your flow.
