Learning Forum

[Alisafian]

Hello everyone,

 

I want to simulate a cylindrical roller with a diameter of 20 mm and a height of 30 mm. It is located on a thick plate, and a vertical force of 4000N is being applied on a line on the top of the cylinder. According to Hertzian contact theory, a contact pressure of 682 MPa will be created. 

I would like to ask, if I delete this cylinder from my 3D model, and instead, I use vertical load on the plate directly or using nodal force on the plate directly, how can I make my simulation result closer to the real situation where I apply the force on the deformable cylinder. My results are too different. 

 

I know that applying force on the cylinder will create a very narrow surface to surface contact and it is not equal to nodal or line force anymore. The thing I am looking for is an equivalent force which mimics the behavior of cylinder (surface to surface contact)

 

Is there any method to solve this problem?

 

Thank you.

[jj77]

That is not working (nodal forces - since you do not know the distribution which depends on the contact).

 

Use contact between parts when doing this type of analysis.

 

See the verification manual to see how this is set up and understand it:

VM63 or VM191 or VM272 it is called:

And here:

https://confluence.cornell.edu/display/SIMULATION/ANSYS+-+Hertz+Contact+Mechanics 

 

[Alisafian]

Dear @jj77,

 

I appreciate your response. I have already watched this video, and that is why this insight showed up in my mind to use equivalent nodal force instead of the sphere/cylinder. Because for an accurate result, we need small mesh size and inflation around the edges. This feature highly increases the computational time if the model is explicit dynamic and the roller is passing over this surface. 

In this video, the surface is rigid and is considered as one element. Suppose we need to consider deformable surface as well. If the surface also needs small mesh size and inflation, the computational time will be increased. Am I right?

 

Thank you.

[jj77]

A Hertzian contact problem is a a nonlinear contact model so we need to model the contact in detail in order to capture the pressure distribution . So this is not simple and can ofcourse not be represented by nodal forces.

 

Now if you have a truck driving over a bridge so you want to see the global response (deflection due to moving load) and do not care about the local pressure distribution on the deck, then yes we can just use an equivalent force for the truck that moves over the structure.

 

So if we care about the local stresses under the contact, then that needs to be modelled in detail with contacts. If we do not care and we care about a global response like for the bridge than a point force is fine.

 

[Alisafian]

Thank you very much for your response.