[Horton]

I am looking for some help with Flotherm XT, please.
I used a simple cylinder ( heat block) in free air to represent a resistor dissipating 3W, but the resulting surface temperature is much higher than measured.
The aluminum cylinder is 14mm diameter, and 28mm in length. Ambient is the default of air at 25C. Essentially the same conditions as the beginner's example provided in the manuals.

For the cyliner in question, radiation is on, and emissivity is set to 1.00.
The surface temperature produced by XT is 120C, but in practice, is 65C.
Can somebody please explain?

[Gabriel_C]

hello,

I tried the same setup in fth 10.1.

I made a aluminum cylinder with the above dimensions and attached aluminum (201W/m*K) and surface with emisivity = 1.
The results are (@25°C air temperature) 162°C for the vertical position and 142°C for horizontal positioning of the cylinder. This is much more than what you measured. Also a ratio of 4:1 between convection and radiation has been observed (approx 2.4W are lost through convection and 0.6W by radiation).

How did you performed the measurement: thermocouples or infrared measurement?

I guess the measurement shows less mainly because (in order of importance):
- the power is not really 3W inside the cylinder but much smaller
- the wires that connect to the cyclinder cool by conduction the cylinder or it is in contact with a "heatsink" (table, clamps...etc). So the cylinder does not stay in free air without any conduction cooling path.
- Thermocouples are not in good contact with the cylinder so you do not measure the actual temperature.
- you made the measurement too fast and didn't reached "steady state" temperatures.

Although I have a fth XT license, I don't have time to test it also in XT but I trust that you simulated correctly


Hope this helps, Gabriel

[Horton]

Thank you very much for your prompt reply, Gabriel.

The resistor is '25Watt', aluminium-clad, and horizontal. I measured the temperature using a thermocouple (probe), and ensured steady state was reached. The wires are very thin, and the resistor is supported away from any surfaces by polystyrene foam. I measured the supplied current, and the voltage at the resistor terminals. I am sure it's 3W.

The resistor too hot to touch for an extended period, but certainly not 120C, let alone 140C. The hottest point is the upper surface at 72C. There is some variation according to probe angle, but a few degrees either way.

I am baffled by the simulated result, and assumed it was my ignorance of XT. That led me to doubt the way the resistor was simulated (as a heat block) so I tried a Joule heated carbon rod, but, still, the temperature was much the same as the aluminum version. Naturally, the actual emissivity is less than 1.00, so the difference is all the greater. Thanks again.

[Gabriel_C]

I guess for this more detailed modeling is necessary.
This type of resistors have wires inside (that define resistance) winded on a ceramic cylinder and then all placed in mica and on top aluminum. also the cooling fins change the overall area of the device and mounting plate ads more mass to the assembly.

One on top of the other is possible that the temperature inside the resistor is much bigger and due to bad heat transfer the aluminum case stays at 70°C.

Try finding how such a resistor is constructed and improve the model a little.

Best regards, Gabriel

[Horton]

I tried materials of different thermal conductivity and density, as well as a rectangular shape. The difference is still large.
If there were a large difference between the core and surface temperatures of the resistor, then I would have expected the highly conductive aluminum to have a lower average temperature, so lower surface temperature.
However, it's good to know that you obtained similar results for the cylinder as it is. I will contact the resistor manufacturer to see if they have some free air data - which is not supplied on the datasheet.
Thanks for your time, Gabriel.

[CFDfan]

Originally Posted by Horton

I tried materials of different thermal conductivity and density, as well as a rectangular shape. The difference is still large.
If there were a large difference between the core and surface temperatures of the resistor, then I would have expected the highly conductive aluminum to have a lower average temperature, so lower surface temperature.
However, it's good to know that you obtained similar results for the cylinder as it is. I will contact the resistor manufacturer to see if they have some free air data - which is not supplied on the datasheet.
Thanks for your time, Gabriel.

This is an interesting discussion and I'm waiting for Gabriel's inquiry from the manufacturer. I'be been using SWFS (essentially FloEfd) for along time and noticed that in natural convection cooling models the program produces higher temperatures than the reality. The differences are not so dramatic however - about 5-10%.
Since flothermXT uses the solver and the mesher of FloEfd, one could expect higher temperature as well. I'm not familiar with XT, but a colleague of mine who uses it suggested to select the finest meshing option of XT and rerun the model.
Also there is a simple rule of thumb. The film coefficient including radiation in natural convection applications rarely exceeds 10-12W/(K.m^2). You just calculate the surface area, including the fins, put the power dissipation and you'll get an idea how much temperature rise to be expected.