Problems Encountered in Seismic Time-history Analysis Using Zero-length Elements

[michael-GA]

Dear STKO Team：
I encountered the following problem when conducting seismic time-history analysis using zero-length elements in the single-degree-of-freedom shear model：
My zero-length element seemed not to be activated in the time-history analysis. I attempted to derive the stress at the Gaussian integral point corresponding to the element, but found that the stress value was always zero.I checked each step of mine and tried to change some Settings, but still got the same result.
I uploaded my work file. I'm extremely eager to know where the problem lies and how to solve it.
Any advices and helps that you provide are highly appreciated.

[STKO Team]

Errors:

• Your zero length as only 1 material along the 1st (Ux) DOF, so in the top node you have to fix the Uz DOF (otherwise your model is singular)
• Mass cannot be negative! and you need a mass in the Ux DOF (not Uz), I don't really know what your were trying to do here
• Now you will see the stress in the 1 DOF of your model using a gauss plot

[michael-GA]

Thanks for your replys!
Following your guidance, I have resolved the issue mentioned above. However, during the subsequent analysis, I encountered a convergence problem. The system prompted me:


“WARNING: CTestNormDispIncr::test() - failed to converge
after: 2000 iterations current Norm: 4.43469e-22 (max: 1e-23, Norm deltaR: 1.77636e-15)
NewtonRaphson::solveCurrentStep() -the ConvergenceTest object failed in test()
DirectIntegrationAnalysis::analyze() - the Algorithm failed at time 0.119946
OpenSees > analyze failed, returned: -3 error flag
ERROR: the analysis did not converge
while executing
"error "ERROR: the analysis did not converge""
("for" body line 30)
invoked from within
"for {set STKO_VAR_increment 1} {$STKO_VAR_increment <= $initial_num_incr} {incr STKO_VAR_increment} {

”
I have checked my basic settings and steps and found no issues such as boundary conditions. I have uploaded my model file and look forward to your assistance!

[STKO Team]

Your tolerance is too small.
You are trying to reach a displacement error of 1.0e-23.
Take into account that double-precision floating-point numbers have 16 significant digits. So, unless your typical displacement is very small, you will never reach that value.

[michael-GA]

Dear STKO team:
Thank you very much for your guidance!
But another problem emerged in the following test.I am attempting a time-history analysis calculation for a multi-particle system. Between the particles of each layer, I have used bending springs and rigidlink-beam elements to simulate the rods between the springs. However, in the analysis results, I found that the bending spring on the second floor seems to have no effect, and the rotation angle of the upper node of this zero-length element is very small.
I have uploaded the result video and model file in the attachment.
How should I modify the model to enable the bending deformation to be transmitted upward along the height?
Looking forward to your guidance. Thank you!

[STKO Team]

The issue is that the transformation method cannot handle overlapping MP constraints where a node is used as a master in one constraint and as a slave in another.

The quick way to solve this, is to use the penalty constraint handler (try penalty 1.0e15, 1.0e15 as you used this value in the stiff materials of your zeroLength spring).

The clean way is to avoid overlapping constraints. In your model it happens only in the middle. You can replace the 2 rigid link in the middle, with only 1, whose master is the central node

[michael-GA]

Dear STKO team:
the multi-constraint model was successfully implemented with all constraints taking effect following your guidance. However, I have built a simple multi-degree-of-freedom bending spring model connected by twonodelink（I have uploaded the established model to the attachment）, and during the time-history analysis, I encountered three issues:
First, the bending spring seems to only be effective at the bottom layer, and is not activated in the upper structure;
second, I checked the OPENSEES help documentation, which mentioned that the two-node translation and rotation of the twonodelink element are coupled during calculation. Does this mean I need to establish three twonodelink connections between the two nodes to decouple them?
Third, there is an "orient" option in the twonodelink element. After checking it, I'm not sure what effect it has. Does it make the local coordinate of the element coincide with the global coordinate?
My ultimate goal is to establish a multi-degree-of-freedom model with a reasonable deformation mode and connected by curved springs.Each layer's curved spring is activated.
I look forward to your reply on the above issues. Thank you!

[STKO Team]

OPENSEES help documentation, which mentioned that the two-node translation and rotation of the twonodelink element are coupled during calculation. Does this mean I need to establish three twonodelink connections between the two nodes to decouple them?

I don't understand this. If you want the DOFs to be uncoupled, just use zerolength elements

[michael-GA]

Here's the situation: The zero-length element cannot be applied to my model because it doesn't have the translational and rotational coupling at a specific length. I have to use the twonodelink element, but during the usage, the first problem mentioned occurred. Could you please take a look at the model file I uploaded earlier? I want to figure out why the translational degrees of freedom of all layers except the bottom layer are forced to be unified and how I can solve it. Looking forward to your reply!

[michael-GA]

Please ignore this point. I have already resolved this doubt in my subsequent research. Please take a look at my first question. Thank you!