Contact Parameter

Hi Seng,

 I think as a general rule you want to achieve the following:

 - Find a stiffness and exponent value that results in an acceptable amount of penetration for the typical contact event in your system. The 'acceptable' amount should be determined by considering material properties and likely the expected deflection of the structure around the contact point.

 - Next the value can be set so that it provides some numerical stability. Both too little and too much damping can result in spiky contact forces and slow simulation times. Typically a bit of damping (0.1% perhaps, or 1% of the stiffness as a guess) will help solution speed and produce the least-spiky contact forces.

 - Related to the damping is the maximum penetration depth for maximum damping: if you set this to be small then the damping turns on quickly and your contact event can be dominated by damping forces. You likely just want enough damping to provide stability, again, so don't set the penetration depth to be too shallow.

Regards,

Rishi

Download HertzWin. Once started, identify the closest approximation of the contact geometry (sphere-sphere, sphere-plane, cylinder-plane...), add your materials.

Add a low force (1 N), copy the penetration and force to Excel. Change to 2N, repeat, then 5, 10, 20 ....50000 (or whatever is suitable for your model). Plot force vs, displacement. Add a trendline and change the type to power and show the equation. Probably be worth it to increase the number of significant digits on the equation. There you can read stiffness and exponent. This will give very accurate contact parameters, but not necessarily the ones that gives the fastest simulations. So you might want to decrease the stiffness a factor of 100 s=to start with (easier to debug the model if necessary with softer contacts).  You would typically see an exponent in the range of [1.0 , 1.1]

Regarding damping and penetration: Magic! No, not really, but a lot of trial and error might be required. Starting with a damping ~0.1% of stiffness (if you work in N, mm, seconds) is usually a good start. Then you should be able to use a dmax=0.01 mm,

But it could be easier to get the model to run smoothly using c=0.01*k and dmax=1 mm (or about twice the maximum penetration). This aligns better with the 15 different suggested damping methods that I have found suggested for Hertzian contact. With this high dmax, you assure that the damping always increases with penetration, as opposed to using a very small dmax that would have a damping independent of penetration depth. This makes sense, as if you go back behind the Hertz theory and looking at the material equations, stiffness and damping are dependent on penetrated volume and change of volume. I did the exercise last year, starting with the material constitutive equations based on penetrated volume, and all the Hertzian equation fell out automatically for the different geometry cases.

Thank you so much for the advice and support. I'm going to try it and see how it goes.

Thank you so much for the advice and support. I'm going to try it and see how it goes.