how to calculate vibrations in metal cutting

81 Views
Last Post 21 January 2019
Topic Is Solved

ttgf1 posted this 20 January 2019

Hi,

I am trying to model vibrations in metal cutting. For cutting (without vibrations), I use explicit dynamics. It gives me strains and strain rates.

My cutting simulations (no vibrations) are very similar to:

About the vibrations of the cutting tool and the workpiece, which are my main interests: I think I need to couple a modal analysis with Explicit dynamics? I have no idea on how this coupling works, if it is possible and what are the loads that need to be input to the modal analysis from the explicit dynamics?

- thanks

#Permalink
0
0
0

explicit-dynamics modal-analysis

4 Comments

Order By: Standard | Newest | Votes

jj77 posted this 21 January 2019 - Last edited 21 January 2019

Explicit time integration can be used also for dynamics (say wave propagation or high freq. vibrations, or impact vibrations), so you do not need to use a modal analysis (you could not couple the two anyhow, since one is in freq, domain the other time marches in time).

(Normally not used for very long events of low freq. vibrations, say of an seismic event).

You just need to look with a probe say on the Y displacement on the part of interest and plot these. From there you can do an FFT and see for any frequencies, if you like that. That could of course be compared to a separate modal analysis of the cutting tool, in order to identify its freq.

In order to capture, the vibrations of the part, you also need to have a dt (time step) in explicit that is at least 10 times smaller than the highest frequency of interest. Normally this should be OK, since explicit uses a very small time steps (dt~L/csound),

In any way running explicit means that you care about the event of cutting and the dynamic stresses generated there. If you just run a modal analysis that is only to identify how the tool might vibrate, it does not tell you though how it will vibrate during the cutting process and how large the stresses will be.

For that you need to run explicit transient dynamics (or implicit if possible), since the cutting process is a transient event

#Permalink
1
0
0

Liked by

ttgf1 posted this 21 January 2019

Thanks for the response.

I see. However, there are a few BCs that may cause trouble: the bottom side of the part is fixed in x,y,z. So no displacement occurs. all the six faces of the part are fixed in z. The side of the part which is farthest from the tool is also fixed in z and x. All the six sides of the cutting tool are fixed in y and z (it moves along x).

With all the above boundary conditions,which are added for numerical stability and convergence, I am not sure if I get anything other than zero for the Y displacements of the tool. I can try to remove some of the BCs to see how it goes.

#Permalink
0
0
0

Moderator

peteroznewman posted this 21 January 2019 - Last edited 21 January 2019

ttgf1,

I agree with jj77 that you should try to build a model for the implicit Transient Dynamics solver instead of the Explicit Dynamics solver. The reason is there is a tremendous amount of "noise" in the Explicit Dynamics solution because the sound waves are bouncing all around the inside of the part creating a lot of noise if you are trying to measure a stable vibration effect. I think it will get drowned out by the bouncing sound waves. That won't happen in the implicit solver.

Regards, Peter

#Permalink
1
0
0

Liked by

jj77 posted this 21 January 2019 - Last edited 22 January 2019

well if it is all restrained, then there is not any vibration as you say. Ideally one will have the same BC as in real life if possible or as close as possible.

I doubt though that it will give any sensible results since the excitation of the cutting tool, is probably depending on the surface roughness of the cut tool and the cut surface, their unevenness (perhaps some surface undulations), perhaps type of stick slide phenomena, and other complex forms of dynamic excitation which I very much doubt can be captured in an accurate way in an FEA model.