QUALITATIVE Torque calculation 3

[Alfninux]

Hi everyone. I've a question related to the QUALITATIVE computation of the required torque to make a rotation.

The image shows a cross section. Two rigid bodies (hollow cylinder and elliptic base cylinder) and a deformable element are shown.
Supposing to know all the mechanical properties required, how do I compute (qualitatively) the torque required to make a 90deg rotation of the inner elliptic body in order to deforme the deformable element?

The deformable body and the outer cylinder are linked along their contact surface, and cannot slide

 

[handleman]

Do you know what the word "qualitatively" means? What the heck would a qualitative answer be? "Half a whole bunch-ish"? "More than I want"?

 

[CWB1]

As alluded, I believe you want to quantify the torque necessary which involves knowing the material properties of each part so you can take into account a combination of deflection and sliding friction. Depending on the accuracy necessary you're also going to want to take into account the manufacturing tolerance of each part.

 

[rb1957]

this would be a really nasty problem to solve. For example, will the insert deform (as obviously it should) or will the insert slip and rotate ?

another day in paradise, or is paradise one day closer ?

 

[btrueblood]

And will it deform smoothly and elastically, or will it deform plastically, or will it just rupture into little pieces?

 

[GregLocock]

That is very similar to how many rubber bushings in a car work, except you have a free floating inner rather than it being bonded to the rubber. Even in that simpler case it is very hard to calculate this (or model it) to better than about a factor 1.2 or so.

In your case you are also going to have friction.



Cheers

Greg Locock


New here? Try reading these, they might help FAQ731-376

http://eng-tips.com/market.cfm?

 

[Jboggs]

From Google:

"What is difference between qualitative and quantitative?

What is the difference between quantitative and qualitative research? In a nutshell, quantitative research generates numerical data or information that can be converted into numbers. Qualitative Research on the other hand generates non-numerical data."

So, you want non-numerical data? How does that work?

 

[IRstuff]

I'm guessing the OP meant they wanted an "equation" as opposed to a "number," and being from Italy and all that...

TTFN (ta ta for now)
I can do absolutely anything. I'm an expert!

https://www.youtube.com/watch?v=BKorP55Aqvg

faq731-376 forum1529 Entire Forum list

http://www.eng-tips.com/forumlist.cfm

 

[BUGGAR]

Similar to what Greg said, this could be a rubber bushing for an inner control arm joint with some progressive action? Can you make it solid and vary the density of the elastomer in the shape you have shown to eliminate the friction part?

 

[BrianE22]

I'd set up the free body diagram for different angles on the cam. I'd then break down the forces between the cam and the rubber into components into the rubber and tangent to the rubber. That's the easy part. Knowing how the rubber reacts those components would probably require some high level elastomeric experience. Maybe the people in the "Rubber Engineering" forum could help at that point.

 

[hydtools]

Somewhat like torsion axles.

https://www.alltrailer.ca/2018/01/31/torsion-axles-vs-spring-axles-which-is-better/

Ted

 

[Alfninux]

Sorry. I wasn't so clear.
For "qualitative" I meant that I don't need a precise number but a good procedure to understand more or less the amount of required torque. The deformable part is an elastic element (rubber-like) and it deforms just elastically. I'd like to know (in the case I have to choose a servo motor) the required torque under the assumptions:
-elastic deformation of the deformable element;
-no deformation of the rigid bodies;
-no sliding of the deformable element.
-friction between the elliptic inner cylinder and the deformable element.

I hope I've been clearer

 

[BrianPetersen]

The geometry that you have created and the materials that you are suggesting (I wouldn't call it "chosen", yet) are going to behave highly non-linearly. Build a model and test it.

 

[FEA way]

Because of the complexity of this model (unusual shape, elastic deformation of some parts and rigifity of another ones, friction and so on) I think that the best way to solve it is to use Finite Element Analysis. It may also be the only way apart from physical experiment because I doubt you can find any useful formula for that. All this assumptions you've mentioned can be included in FEA (rigid bodies, frictional contact between inner rigid part and deformable part, no sliding contact between outer rigid part and deformable part) but you can also eliminate some and solve even more complex problem. For example there's no need to make these parts rigid. Required torque can be calculated using some parametric study - apply different torque in each analysis and solve it to find out the displacement (rotation). If it doesn't match required value, repeat the analysis with different torque. The only problem can be the time required to perform all these analyses. But with some simplifications it may not take too much time to solve. Also keep in mind that rubber-like material probably should be modeled using hyperelastic formulation but we can assume it's linear elastic.

 

[rb1957]

I'm with Brian. I think this is way too complicated to get real numbers from FEA and a physical test is what you need. Once you've done a bunch of tests you may have learnt enough to create a general (algebraic) solution or to make a model.

another day in paradise, or is paradise one day closer ?

 

[FEA way]

Well, it’s always experiment > FEA > hand calculations. Of course apart from situations when exact analytical solution exists (like for beams).

But I would try with FEA anyway. Good nonlinear solver (like Abaqus) should provide results that will be close enough to reality.

 

[BrianPetersen]

I never said the model had to be a physical one A finite-elements model is another way.

 

[rb1957]

No, you didn't say he had too, but you did "recommend" it. personally, I can't even imagine the properties required to run this in FEA …

so I strongly for testing, Brian's either way (?), and FEA_way is "all the way with FEA" …

many ways to skin cats ...

another day in paradise, or is paradise one day closer ?

 

[Alfninux]

In case I want to perform a FEA before building a model, which software would you advise to be used? (I'm looking for some simple software, because in a first approximation I don't need accurate results).

 

[BrianPetersen]

FEA and "simple software" do not exist in the same space and time.