[bemomb]

Hello, everybody.

I would like to simulate a simplified swashplate pump. It essentially consists of a rotating disc on which a piston presses axially. In the appendix you can see a sketch. The rotation causes friction at the point of contact, which leads to heat development. My question now is, how do I map the heat development with Ansys CFX? My first thought was to calculate the friction work using the force of the piston and to define it as the source at the interface. Does that make sense? Or is there another possibility?

thanks in advance

[evcelica]

That sounds like the right way to do it.

I can't remember how the input works exactly though? Whether you define it on the interface itself? Or on both sides of the interface separately, where I would then specify half on each side of the interface. There may be no difference if there is no thermal resistance at the interface, but I would see if it makes any difference.

[bemomb]

Hi evcelica,


thanks for your answer, I will try that out btw if I rotated the plate, would it be necessary to use a frozen rotor? Or can I just rotate the domain?


thx in advance

[urosgrivc]

It sounds like the right way to do it, yes, but when you try it just doesn't work.
please update me if I am wrong and you find it to work.

This is way more complex than just setting 50% on each side of the interface.
The part which is rotating is able to take away far more energy than the piston as fresh material roles into contact constantly than the heat has time to conduct inwards till it comes in to contact again.
this is a similar example to (brake-disc brake-pad) if I understand correctly.
This gets more complex if you have different materials in contact (with different thermal properties)

I had huge problems setting an interface like this at the beginning.

Then I actually faked the interface as a separate interface volume with volumetric heat production and orthotropic thermal conductivity.
This led to simulation which automatically determined heat distribution between the two bodies.
It gets even more interesting as you are able to observe that heat input is not constant

Something like this... https://drive.google.com/open?id=1LC...QFPdPQbLDP9i32
But this took a lot more than basic settings

If possible, if geometry lets you, do not rotate the mesh as this will complicate
the contact further.

[bemomb]

Thank you so much for your answer.
I have already tried and compared the evcelica approach. The results were very similar.

Yes, the contact between brake disc and brake piston is a very similar example.

Could you explain your approach on how to set the interface as interface volume with volumetric heat production and orthotropic thermal conductivity?

So you suggest to simulate everything standing still without rotation?

[bemomb]

have just seen your video, looks super difficult.


I guess I have to find a simplified solution

[urosgrivc]

well thanks

yes without rotation.
In the settings it is possible to "rotate the heat numerically" but the mesh stays in the same place.
this simplifies things (timesteps, mesh) a lot when possible but geometry must be a simple cylinder

This was also included in my model but my model has geometry which must rotate because of the cooling fins ass seen from the video (not a simple cylinder). This is why I have an additional domain
Domains are stacked from the centre (symmetry) of the brake disc out like this:
(rotating domain cooling ribs),
(stationary brake disc with the heat going around and around-simple cylinder),
(Volume interface-special properties to fake volume to act like 2D element),
(stationary brake pad)

simplifying things:
You could compute the heat distribution coefficient by hand then set this as mentioned before this factor will not be 50/50% but might be 90/10% if you are ok with making a fair enough assumption.
There are equations made for this
I vent all in on simulation as now I am able to optimize the shape of brake pads

[bemomb]

Okay, where do I set that only the heat should rotate and not the geometry?

and how do I create this fake interface?

[urosgrivc]

read this..
https://studylib.net/doc/10035486/mo...s-cfx-tutorial
hint; page 28,29..

but read all of it its usefull

[evcelica]

I still think the heat could still be applied to each side at a 50%/50% ratio.

For every action, there is an equal and opposite reaction. The piston is not doing more work to the rotor, than the rotor is doing to the piston. They exert the same forces on each other.

Yes, The rotor will end up taking away more of the heat since it will be the cooler part with more cooling area. 90%/10% sounds more than reasonable. It depends on the cooling to each part. But that 90% of the energy to the rotor is just it's 50%, plus 40% from the piston, which transfered through the interface of the hotter piston to the cooler rotor.

If you have no thermal resistance at the interface, it will not make much difference either way. But with a high thermal resistance it would. I stand by a 50/50 application of heat being correct.

[bemomb]

Hi evcelica


There should be no resistance, as both components touch each other. It is difficult to imagine it physically correct, but I also think the heat is split and then there is an additional heat conduction from piston to disc.

[bemomb]

I found a thread of tacitosantos with a similar issue.


What I dont understand is why his results show a heating that follows the rotation of the plate while mine is only taking place at the interface (pls see attached video)

[urosgrivc]

Thank you evcelica for the explanation, I will definitely test it out.
As I was doing this quite some time ago I don't exactly know what the problem was but something weird was going on, and I think that there is also some distribution factor you can set which as I remember is not mentioned in the papers, I think I was confused by that back then.

[bemomb]

little update from my side


changed the motion of the plate from domain motion to solid motion and now there finally is a heat input which follows the rotation. What i dont understand is the temperature distribution on the interface

[urosgrivc]

Try to plot the temperature of the interface on its own temperature scale (temperature difference between max and min temp might be very small).
Maybe check the mesh as temperature gradients are present its not bad to add some thinner layers near the contact region to really capture the temperature gradients.