penalty values for zeroLengthContact3D element

[bozturk]

Dear STKO team,

I am currently working on simulating the interaction between double rubber pads located between the pier top and the deck in my OpenSees model. I came across an element (zeroLengthContact3D) on the OpenSees wiki that seems appropriate for this purpose, and I have incorporated it into my model. However, I'm facing some challenges with setting appropriate penalty values, as I chose them arbitrarily based on parameters used in a similar thesis I came across.

For the element definition, I've used the following parameters:

Normal Penalty (kN): 1E12
Tangential Penalty (kT): 1E6
Friction Coefficient: 0.5 (between the organic glass and the plastic rubber)
Cohesion: 0
I've placed the element in the +Z direction, with the top pier node as the master and the deck node as the slave.

In my research, I've been scaling the PGA of the input motion to different intensity levels. However, I've noticed a significant and unrealistic spike in the calculated acceleration of the deck element (directly above this contact element) when I scale the PGA to 0.60g. This has led me to suspect that during stronger shaking, the contact element may not be effectively maintaining the integrity of the components, causing them to separate.

• I'm curious if there are any general guidelines or rules of thumb for determining suitable penalty values for such cases. I tried to manually adjust the penalty values by orders of magnitude (increasing and decreasing by +-1E1), but the results have not shown much improvement.

• Also, I'm not entirely sure if the current element I'm using is the best choice for simulating rubber pads. This was the only simple element I could find that allowed me to define friction between the nodes. I'd appreciate it if you could suggest any alternative elements that might be more appropriate for modeling this situation.

To provide a clearer understanding of the connection, I've included a figure below.

Screenshot 2023-08-15 at 10.02.17 PM.png (222.11 KiB) Viewed 5086 times

I would greatly appreciate any insights, suggestions, or help you could provide to help me solve this issue.

Regards,
Burak O.

[STKO Team]

Normal Penalty (kN): 1E12
Tangential Penalty (kT): 1E6

In a contact element, those values are penalty values. It means that in theory the should be INFINITY, but in practice, you can't, so you chose a "high enough" value. And it really depends on your model. For example, if the typical stiffness Ktyp = EA/L (where L is the size of the mesh) of the pier and deck elements has an order of magnitude = 10.0^X, make Kn some order of magnitude higher ... Kn = 10^(X+3).

The above-mentioned approach is fine for hard contact. If instead, your contact element is actually simulating an object with a finite stiffness, use that one. But as I said it really depends on your problem.

In my research, I've been scaling the PGA of the input motion to different intensity levels. However, I've noticed a significant and unrealistic spike in the calculated acceleration of the deck element

It can be due to many issues. The contact stiffness is probably too high. Anyway, try to use a transient integrator with numerical damping to damp out spurious high-frequency modes. Try the TRBDF2 or the HHT

may not be effectively maintaining the integrity of the components, causing them to separate

This is normal in a contact element, they avoid compenetration but not separation

[bozturk]

Normal Penalty (kN): 1E12
Tangential Penalty (kT): 1E6

In a contact element, those values are penalty values. It means that in theory the should be INFINITY, but in practice, you can't, so you chose a "high enough" value. And it really depends on your model. For example, if the typical stiffness Ktyp = EA/L (where L is the size of the mesh) of the pier and deck elements has an order of magnitude = 10.0^X, make Kn some order of magnitude higher ... Kn = 10^(X+3).

The above-mentioned approach is fine for hard contact. If instead, your contact element is actually simulating an object with a finite stiffness, use that one. But as I said it really depends on your problem.

In my research, I've been scaling the PGA of the input motion to different intensity levels. However, I've noticed a significant and unrealistic spike in the calculated acceleration of the deck element

It can be due to many issues. The contact stiffness is probably too high. Anyway, try to use a transient integrator with numerical damping to damp out spurious high-frequency modes. Try the TRBDF2 or the HHT

may not be effectively maintaining the integrity of the components, causing them to separate

This is normal in a contact element, they avoid compenetration but not separation

Thank you very much for your explanation. I really appreciate it.

[STKO Team]

You're welcome