Problems in analysis results when the user defines different regions

[aziz_barzuk]

Hello,
I was modeling a simple column composed by 2 soil layers: a deformable bedrock (modeled with the ASDAdsorbingBoundary3D element) and a 40 m soil layer, all is linear elastic.

There is some problem I met by using of STKO in the modeling procedure that I need to fix:

- I can't perform a transient analysis with the Newmark method and the ASD element that seems to work only with the
TRBDF2 method. The analysis converges with the Newmark method, but the results are not good.

-I can't specify different rayleigh coefficients using the region command otherwise the result is uncontrolled.
The result is uncontrolled even if I speficify the global rayleight command.
So seems that in this case (annexed) I can't use it.

I think I did some mistake in the joining of the 2 soil layers with the merge command (I did the intersect of the 2 solids and not the glue option). So what is the best way to model different soil layers and how to keep them together ?

Why can't I use the region command ?
If I use it to specify different rayleigh properties the results go crazy.

This behaviour happens also using the LK surface element at the bottom of the model.

I annex my simple try model

Thanks in advance.

[aziz_barzuk]

I annex the try.scd file with the lk element, if you run the analysis you can see that only the first element in the bottom, the cube is moving the top is not moving.

[STKO Team]

- I can't perform a transient analysis with the Newmark method and the ASD element that seems to work only with the TRBDF2 method. The analysis converges with the Newmark method, but the results are not good.

Pay attention to the Newmark's coefficients! 0.5 and 0.25 for unconditional stability... you set 0.2

-I can't specify different rayleigh coefficients using the region command otherwise the result is uncontrolled.
The result is uncontrolled even if I speficify the global rayleight command.
So seems that in this case (annexed) I can't use it.

I don't understand what problem you had with damping

I think I did some mistake in the joining of the 2 soil layers with the merge command (I did the intersect of the 2 solids and not the glue option). So what is the best way to model different soil layers and how to keep them together ?

They are connected correctly

Why can't I use the region command ?
If I use it to specify different rayleigh properties the results go crazy.

You have to explain the error otherwise it's impossible for us to answer

[aziz_barzuk]

In this case seems that the result doesn't change a lot using 0.2 or 0.25 considering that the result is totally unespected due to some mistake in the model i did.
The problem is that by using the region command the upper part doesn't move (forum2.scd case) or the column seems to have no G stiffness (forum.scd case).

So I can't understand how to model with STKO a column composed of 2 different layers: a bedrock with a lk condition at bottom and a soil layer. Then I want to assign 2 different rayleigh parameters to the 2 layers.

The only way I know to assign the rayleigh parameters is to define 2 regions.
The problem is that when I define the region seems that:
- the 2 layers behaviour is like a very low G column with inf. deformations.
- the bottom part moves itself and the upper part is totally quiet. So the wave is not spreading in the upper part.

Did you try to run and see the results ?

If it is still not clear I will post some images of what I can't understand in the model results.

I think that the problem is in the region command, in particular in the -elerange option maybe.
In a simpler example I used the region command by writing all the element and it worked well.
But I need to learn how to do this thing also in STKO.

[aziz_barzuk]

I try to summarize better the new model in this post:

In the model I posted is represented a simple soil column based on a rock deformable layer of 1 meter thick with a Vs = 800 m/s. In this bottom layer, at the base, there is a LK boundary.
The upper part has a Vs = 250 m/s and it is 40 m thick.
So, I also defined 2 regions: TOP and DOWN and I had assigned 2 different Rayleigh damping for each one.
All the model is elastic.
The normal-x sides are linked with an equDof relation that will be activated in the next “dynamic phase”.
The column can move only in the x direction.
This example is very common, and it is equivalent to the ones that can be analyzed with simple excel sheet like EERA.
The signal is assigned at the base of the bottom part, and it is a ricker wavelet of 0.5 Hz
The calculation is set in this way:
The initialization phase is in K0 conditions, and it is calculated as a fake transient phase, in the next phase it is activated the shaking load at the base of the column and the x fixes are removed.
Then, still in this phase the equalDof at the x sides are activated.
So, the column can move only in x direction and represent a soil bank infinitively extended.
I have a benchmark to compare with and the result obtained is this:




(STKO result in dashed line)

The element used is the 20_8_brickUP one and all the pore pressures are set to 0.
The result is about the output signal at the base of the column.
The problem is that it seems that the Rayleigh command is not considered at all because the result
obtained and reported is the one that could be obtained with no Rayleigh damping.
But I souspect that could be related to the lk element. How is it modeled?
Is the C parameter required in the LK condition referred to all the base ? or is it related to the size of the single element ? or the single dashpot ?
If I use the brickUP element the things go worse in the sense that the analysis doesn’t converge at all.
From the plot of the results, I can’t understand why the Rayleigh is not recognized.
I did another model (by hand) and even if I write something different respect the stko output command the only things that is different is:
the set of the region command that I used writing the list of every single element.
The set of the lk at the base
But in that case, I match the benchmark.

Thanks in advance
G. Cavallo

[STKO Team]

I think you did not add the volumes of the geometry in the selection set, but only the boundary faces (which are not translated to OpenSees elements).

Redefine your selection sets, and when you are adding items in the selection set, make sure to select only sub-solids of your geometry. You can use the selection filter:

subsolid.png (10.52 KiB) Viewed 963 times

[aziz_barzuk]

Thanks for the answer,
Unfortunately it doesn't seems to be the problem:
- you can see that in the generated file, and in the region command the element are listed under the -elerange option, so I think that the element was considered in the right way
- the result doesn't change.

I think that there could be 2 main reason:
- the region command doesn't select correctly all the element
- the LK element does something strange. Infact in the slope tutorial I didn't get why you multiply the load for a (0.4 g) factor.

Thanks
GC

[STKO Team]

Unfortunately it doesn't seems to be the problem:
- you can see that in the generated file, and in the region command the element are listed under the -elerange option, so I think that the element was considered in the right way
- the result doesn't change.

Please, do it carefully. I did it on my side and it works. I also compared the region command assigning the same coefficient of a global rayleigh damping and it works correctly.

- you can see that in the generated file, and in the region command the element are listed under the -elerange option, so I think that the element was considered in the right way

Yes, but as I said, STKO's pre processor mesh has genrally more element than those passed to OpenSees: For example in a CAD, a volume also has boundary surfaces (that generate face elements) which in turn have boundary edges (that generate line elements).
Only sub-geometries with an element property assigned will be converted to OpenSees elements. In your case you selected boundary faces, which are not in OpenSees (in fact you can see from the tags passed to the region command, that are not the same as the tags of your brick elements).

Please, double check your model, I did it and it works

[aziz_barzuk]

Thanks a lot for your efforts. I appreciate a lot.

I understand the problem about the element created from cad and the generation of the elements on OpenSees, and I followed what you suggested by deleting the regions created and re-creating them by using the "select volume" option ON only.

I only ask you kindly if it could be possible for you to annex the model you checked and corrected and that works so that I can see the differences with mine and learn something new.



Thanks
GC

[STKO Team]

check this:

forum_next.zip

(271.29 KiB) Downloaded 76 times

To check it I used a very high damping ratio