Separation contact - Loose surface contact after few steps

[tony_berset]

Hello,

I always encounter the same problem with simulations where a first part is moving and a second part is sliding or interacting with the first one. In every cases I use separation contacts and the analysis involve large displacements.

More often, the simulation reach the end of the calculation without any problem but the surfaces loose contact after few steps.

A typical example is the short video attached.

Could you help me achieve these kind of analysis with contacts kept during the full simulation please ?

Thank you in advance for your precious help.

[John_Holtz]

Hi @tony_berset . Nice analysis so far.

I can think of a number of possibilities for why the contact "ends".

1. The Maximum Activation distance needs to be large enough to created contact elements between the nodes at time 0 that will eventually come into contact later. For example, the last set of teeth may be D mm apart at time 0 (and obviously 0 mm apart when they engage toward the end of the analysis). The maximum activation distance needs to be larger than D mm in order to create the elements. 
2. The stiffness is too low, so the parts are penetrating. Change the "Contact > Advanced Options > Max Penetration Distance" to something on the order of 10% of the mesh size.

If those items do not solve the contact issue, feel free to zip the assembly and part files and attach the zip file to the forum.

[tony_berset]

Hi John,

Thank you for the answer.

Unfortunately your hints does not work for me. Do you have other items to explore that could help me resolve this problem ?

[John_Holtz]

The Explicit analysis (either explicit dynamics or explicit quasi-static) would be the next thing to try. Start with the explicit analysis to see how long it takes to run. Keep in mind that the duration needs to be very short (like 0.001 or 0.01 seconds depending on the model) because the time step size for the calculation will be on the order of 1E-8. 

If the Explicit dynamics behaves properly regarding the contact, then you can use the explicit quasi-static to let the solver "stretch" out the duration. This gets the answer closer to a static solution even though the math is still based on dynamics; hence, the name "quasi-static". (It is not truly static, just close enough to minimize the dynamic effects such as the kinetic energy.)

[tony_berset]

I already tried Explicit quasi-static, but for a relatively acceptable amount of time it gives me 8 days of estimated calculation time.

For a shorter time of simulation, the contact is creating like an impact between parts at the beginning of the simulation instead of slinding gently.