Layered Shell

Telec
Posts: 27
Joined: Sun Oct 11, 2020 8:27 am

Layered Shell

Post by Telec » Wed Oct 28, 2020 10:25 am

Hello,
when using the layered shell, what is the order of insertion of the layers? For example, for a shell arrangement as in the drawing, is the insertion adopted in the image okay? With the rebar layer placed at the bottom, but in the insertion instead it results at the top. Is there a way to see in the post processor if the input is correct? Thank you
e.PNG
e.PNG (4.63 KiB) Viewed 163 times
1.jpeg
1.jpeg (52.84 KiB) Viewed 163 times

STKO Team
Posts: 339
Joined: Tue Oct 29, 2019 8:45 am

Re: Layered Shell

Post by STKO Team » Wed Oct 28, 2020 11:29 am

It is correct, the frist layer in the TABLE is place at the bottom of the shell, while the last one in the TABLE goes to the top.
To visualize the layers in the post processors, you need to use a fiber plot, where you can visualize results such as section.fiber.strain or section.fiber.stress

dudchenkoav
Posts: 22
Joined: Mon Feb 17, 2020 3:27 pm

Re: Layered Shell

Post by dudchenkoav » Thu Nov 26, 2020 9:35 am

Dear Stko Team,

could you please clarify what concrete models are available for modelling of concrete multylayer shells? I ask this question because I got the following message(see the attached screenshot "OutPut") in the terminal once I tryed any other model except for DamageTC3D.
Another question is the tensile strain at a peak tensile stress in Damage TC3D model. According to the uniaxial tests,I have performed, it takes the value 0.0001. At the same time, the value of strain corresponding to peak tensile stress equals to 0.00015 for concrete according to FIB bulletin 70 "Code-type models for concrete behaviour" (see the attached screenshot). May be I am wrong here as I have not figured out Damage TC3D model completely. Could you please clarify these issues for me?

Best Regards,
Aleksandr
Attachments
EE-SS.PNG
EE-SS.PNG (31.46 KiB) Viewed 98 times
OutPut.PNG
OutPut.PNG (84.58 KiB) Viewed 98 times

STKO Team
Posts: 339
Joined: Tue Oct 29, 2019 8:45 am

Re: Layered Shell

Post by STKO Team » Thu Nov 26, 2020 11:05 am

could you please clarify what concrete models are available for modelling of concrete multylayer shells? I ask this question because I got the following message(see the attached screenshot "OutPut")
Actually you can use any nDmaterial. However there are not so many alternatives for concrete. The UserPlaneStressMaterial you are using internally calls the PSUMAT subroutine that is not available in the OpenSees version you are using. Since there is code restriction for that, you can find it only in the pre-compiled sequential version obtained from the OpenSees website.
According to the uniaxial tests,I have performed, it takes the value 0.0001. At the same time, the value of strain corresponding to peak tensile stress equals to 0.00015 for concrete according to FIB bulletin 70 "Code-type models for concrete behaviour" (see the attached screenshot). May be I am wrong here as I have not figured out Damage TC3D model completely. Could you please clarify these issues for me?
It is not clear to me what you are asking. Looking at the image you are trying to model a tensile behavior that has a first intial hardening from 0.9*fctm to fctm. This is not possible with DamageTC3D, since the uniaxial tensile law it uses is an exponential softening. Note however that you are not going to see much difference.

dudchenkoav
Posts: 22
Joined: Mon Feb 17, 2020 3:27 pm

Re: Layered Shell

Post by dudchenkoav » Thu Nov 26, 2020 1:14 pm

Thank you for your response. What material models would you suggest for modeling of concrete using layered shell property?
Actually I was asking about tensile strain corresponding to the peak tensile stress value of concrete or concrete strength in uniaxial tension(\epsilon_ct in the graph). This value is specified in FIB Model Code along with the fracture energy and other parameters for various concrete strength classes. In relation to that I have another question. Can we directly use the parameters (Gt,Gc, epsilon_p ,etc.) for concrete from FIB model code or European norms in Damage TC3D model? May be you have the recommendations on how to choose these parameters.

STKO Team
Posts: 339
Joined: Tue Oct 29, 2019 8:45 am

Re: Layered Shell

Post by STKO Team » Fri Nov 27, 2020 4:04 pm

hank you for your response. What material models would you suggest for modeling of concrete using layered shell property?
Actually there are not so many alternatives. The only other concrete model is the PlasticDamageConcrete3d, which is very similar to the DamageTC3D, however it is much more difficult to calibrate, and it has many convergence problems.
So I would suggest using the DamageTC3D.
Can we directly use the parameters (Gt,Gc, epsilon_p ,etc.) for concrete from FIB model code or European norms in Damage TC3D model? May be you have the recommendations on how to choose these parameters.
Exactly you can use the FIB model code ones. Keep in mind that Gt and Gc are input as real fracture energy (and not as specific fracture energy) so you should use the -autoRegularization option

dudchenkoav
Posts: 22
Joined: Mon Feb 17, 2020 3:27 pm

Re: Layered Shell

Post by dudchenkoav » Wed Dec 02, 2020 11:06 am

Dear STKO Team,

Thank you very much for your responses.
Having tested layered shell property I have found several issues related to it. Could you please clarify them for me? Because I am not sure whether I have made mistakes or these are the property features.
For all the cases I considered two material models: Damage TC3D and Steel02 for concrete and steel respectively. For tension/compression and bending problems material parameters are taken the same, whereas for the problem with shear they are taken to fit experimental results for a beam 2DB100 from the work: https://www.researchgate.net/publicatio ... rification.

1. Tension/Compression modeling
The model is a plate with a plane of symmetry at one short edge and axial displacement acting on the opposite edge along the plate. Both tensile and compressive displacement growth linearly from 0 to the specified values.
The obtained peak reaction of the plate corresponding to its strength is lower than the analytically calculated value for such plate by 20-30% and additional strange boundary effects can be seen in post peak regime(see the attached plots TensionRfhPost.png, TensionDispPost.pnf) and in the pre peak regime for the reactions although the displacement distribution seems to be correct (TensionRfhPre.png, TensionDispPost.pnf). This is quite strange as in 3D solid models we did not have such an effect for boundary elements and the reaction distribution was equal along the plate width.

2. Bending
The same model as in the first problem is considered with change in the boundary conditions. One edge of the plate is fixed while displacements are applied along the axis perpendicular to the plate on the opposite edge. Again,in this case reaction moments are not equally distributed along the boundary (see the attached plot BendingRm.png) and the total sum of reaction moments at nodes has a strange peak (TotalReacttm.png) that exceeds the plate strength calculated analytically. At the same time, post peak horizontal part of the reaction curve (TotalReacttm.png) ideally corresponds to the analytically calculated strength of the plate.

3. Shear
In this case the results fit the verification problem much better in terms of maximum load and crack patterns. The only problem is the deflection at maximum loading which is underestimated.Could you please have a look at the model and check if all the misc parameters are defined correctly.

All the models can be assessed using the link below:
https://drive.google.com/drive/folders/ ... sp=sharing

Best Regards,
Aleksandr
Attachments
TotalReacttM.PNG
TotalReacttM.PNG (9.63 KiB) Viewed 44 times
TensionRfhPre.PNG
TensionRfhPre.PNG (52.91 KiB) Viewed 44 times
TensionRfhPost.PNG
TensionRfhPost.PNG (118.49 KiB) Viewed 44 times
TensionDispPre.PNG
TensionDispPre.PNG (79.81 KiB) Viewed 44 times
TensionDispPost.PNG
TensionDispPost.PNG (163.89 KiB) Viewed 44 times
BendingRm.PNG
BendingRm.PNG (52.64 KiB) Viewed 44 times

STKO Team
Posts: 339
Joined: Tue Oct 29, 2019 8:45 am

Re: Layered Shell

Post by STKO Team » Wed Dec 02, 2020 2:46 pm

The obtained peak reaction of the plate corresponding to its strength is lower than the analytically calculated value for such plate by 20-30% and additional strange boundary effects can be seen in post-peak regime
Those strange boundary effects are due to the drilling formulation used by the ShellDKGQ: The drilling DOF follows the Allman-type corner rotation formulation and is not compatible. It is (almost) fine as far as the problem is elastic, but when it goes into softening regime it gets messy.
We developed ASDShellQ4 to overcome those issues. Try that, and you will see a constant state of deformation without those strange boundary effects.

Regarding the reaction forces at the corners: it makes perfect sense, they should be 1/2 the reaction forces of the inner nodes (assuming all the elements have the same length), because the corner nodes have half the tributary length of the inner ones.

Clarabella
Posts: 34
Joined: Wed Jul 08, 2020 3:50 pm

Re: Layered Shell

Post by Clarabella » Wed Dec 02, 2020 4:55 pm

Dear dudchenkoav,
we are really proud of the new shell implemented by massimo, here you can find some videos and info:
https://github.com/OpenSees/OpenSees/pull/365
and here some tests to see the performance:
https://asdeasoft.net/?opensees_validation
Let us know if you need additional info!

dudchenkoav
Posts: 22
Joined: Mon Feb 17, 2020 3:27 pm

Re: Layered Shell

Post by dudchenkoav » Thu Dec 03, 2020 8:35 am

Dear STKO Team,

thank you for your answers. Yes, implementation of this element resolves the problems I had, except for the bending one. Because I still have the unphysical peak that can be explained by appearance of stresses along y-axes (sigma2) near the boundary that is fixed. This is strange as only one point in the middle of the plate edge is fixed in this direction to exclude the effect of Poisson's ratio and plate width change(https://drive.google.com/drive/folders/ ... sp=sharing).
Thank you for the materials on the element I will read it carefully as I have not payed attention on it before.

Best Regards,
Aleksandr
.
Attachments
S22Bending.PNG
S22Bending.PNG (319.15 KiB) Viewed 22 times
RMy.PNG
RMy.PNG (64.94 KiB) Viewed 22 times

Post Reply