[YuguiYo]

Hi everyone,

I'm trying to simulate a MEMS flowrate with Ansys steady state. For those who know a little bit about MEMS, i'm using the thermal time of flight studies. Anyway it's not really important for my issue. I got a problem with the maximum temperature value (which is in milion degres)...

I give you the set up of my system (3D) :

The MEMS is a platinium thermistor which dimensions are 106x10x0,2 µm (0,2 is the thickness). It is "amalgamate" in a SiO2 plinth (106x150x500 µm).

I put the 'topology shared' between the 2 bodies for the thermal conduction.

Then I do the mashing. Due to the thin thickness of the thermistor I put 'mashing sizing at the edges' of the plinth (SiO2) and the 'face mashing', resulting in a pretty good mashing.

Boundaries conditions :

• Internal heat is applicated to the platinium thermistor. A 10 mW power from Joule effect is outgoing from, so the volume thermal flux is 4,17e14 W/m3.

• Convection is applicated to all the exteriors faces (platinium and SiO2) for a natural convection we take a 10 W/m².°C factor

• Adiabatic condition is applicated to the bottom face

When i do the simulation i found overwhelming temperature... But the thing is that all the values taken are coming from laboratory experimentations ! So i don't know from where is coming the mistake

If someone can help me it would be perfect, i'm stuck with this problem for 1 week

Hugo

[evcelica]

Sounds like you don't have the thermal conduction at the interface?
In CFX-Pre, make sure you set "Heat Transfer" at the interface between the two bodies.

This is something you should be able to estimate by hand very easily.
See This:

[YuguiYo]

Hi Evcelica,

Thanks for your answer !

i'm using thermal steady state instead of CFX-Pre. I thought that when the shared topology is applicated to all the bodies, the thermal conduction should be done between them.

I'm more or less ok with your calculations (except that for me the internal heat is divised between conduction and convection flux). Nevertheless, i still found huge temperature (63000-68800°C)...

Do you think that i should use CFD instead of thermal steady state ?

Hugo

[ghorrocks]

Both the FEA approach and the CFD approach should be easily able to model this. But your crazy temperatures are likely to be caused by either the interface between some regions not being defined properly or you not including a temperature condition to constrain the temperature (but your convection BC should have this taken care of).

[evcelica]

As Glenn says, either code will work. I used steady state thermal as well. If you want the two to have perfect thermal contact with each other, select the bodies in geometry (design modeler), and "create new part". Then they will share nodes at the interfaces.
Otherwise, you have to define a contact in mechanical, as well as a contact conductance and pinball region.

[YuguiYo]

Hi,

Thanks for helping me it's really nice !

I've done the calculation again (calculations are in 1D that means that we don't take care about the 3D heat transfer charasteristics; that's why I wanted to know the difference with ANSYS). I found a conduction temperature variation of 3144 °C and a convection temperature variation of 3703°C (so my maximal temperature is 6800°C) and i know that it's way too much comparated to the experimental datas.

By the way Evcelica, I would be interested to understand how you find 104 K for the conduction temperature variation, and from where is coming your 0.5 factor in the formula.

I will try to figure out what can be the difference between the experimentation and the simulation. Maybe i have to set up new expressions or modificated the data i've put..

[ghorrocks]

Are you taking into account radiative heat transfer? At these temperatures radiative heat transfer is likely to be larger than conductive or convective heat transfer.