Dear STKO team,
I created a model of a steel-concrete composite member constituted of a concrete fiber section with lower and upper rebars embedded, and a steel section coupled to the concrete section through zeroLength elements. For the concrete section, I use concrete04. The steel parts use steel01 (lower and upper rebars) and steel02 for the coupled steel section.
It is a half-span model of a 4-point bending test, which I try to analyze in displacement control (the midspan node is the controlled node), to perform a pushover analysis.
My experience is, that the analysis becomes difficult when the elements get closer to the peak strength of the concrete constitutive law and softening initiates. I can add some robustness, when I use elastic materials for the rebars. Moreover, I see a kink in the lambda-midspan deflection curve when I switch to adaptive time steps, presumably due to concrete softening. It seems, that the analysis terminates for a good reason, but I believe some more increments should be possible with the displacement controlled algorithm.
I also tried to set up a monitor for the fiber strains. Is this possible? They did not appear in the monitor even after adding sections.fiber.strain to the recorder list.
Your suggestions would be very welcome. Thank you.
Kind regards,
Robert
Composite pushover
Composite pushover
- Attachments
-
- T1_MWE.zip
- (147.74 KiB) Downloaded 74 times
-
kesavapraba
- Posts: 468
- Joined: Sat Mar 28, 2020 2:25 pm
Re: Composite pushover
Hi, Below are some of my observation about your model, you can look at them.
1] More than one constraints are applied to the same node and DOFs. For example, please look at the end nodes, where both multi-point and single-point (boundary conditions) constraints are applied. Also, you applied both equalDOF and rigidLink at the same nodes.
2] In your concrete04 material, the residual stress goes to zero. You can consider using more stable concrete material, like concrete02.
3] For fiber section number 11, the torsional constant GJ is too less. Please consider using a closer value for the section. Whereas, for the fiber section number 13, the 1e10 seems to be fine.
4] You might consider using Lobotto integration with 3 or 5 integration points for the beam sections.
5] Monitor doesn't provide option for stress or strain of fiber point. However, for monitoring the stress or strain at a fiber point, you can use custom TCL script. For example,
1] More than one constraints are applied to the same node and DOFs. For example, please look at the end nodes, where both multi-point and single-point (boundary conditions) constraints are applied. Also, you applied both equalDOF and rigidLink at the same nodes.
2] In your concrete04 material, the residual stress goes to zero. You can consider using more stable concrete material, like concrete02.
3] For fiber section number 11, the torsional constant GJ is too less. Please consider using a closer value for the section. Whereas, for the fiber section number 13, the 1e10 seems to be fine.
4] You might consider using Lobotto integration with 3 or 5 integration points for the beam sections.
5] Monitor doesn't provide option for stress or strain of fiber point. However, for monitoring the stress or strain at a fiber point, you can use custom TCL script. For example,
where, y and z are local coordinates of the fiber point in the section. Please also refer to the following page for understanding more on recording response in fiber section. https://portwooddigital.com/2021/07/25/ ... -response/recorder Element -file fiber_response.out -time -ele 1 section 1 fiber y z stressStrain
:: With best wishes ::
Prabakaran Kesavan
Prabakaran Kesavan
Re: Composite pushover
Just use adaptive time step.
Your model is very brittle, so it will reduce the time step near the collapse:
Your model is very brittle, so it will reduce the time step near the collapse:
- adaptibe.png (154.88 KiB) Viewed 398 times