Analysis using ASDAbsorbingBoundary3D

[kolozvari]

Thank you for detailed responses. It makes sense. So, it seems that sfsi example is the way to go for me. I already used ASDAbsorbingBoundary3DAuto in my model and I will just proceed with analysis steps. Since sfsi example has a lot of stuff going on (construction stage analysis, etc.), here is what is my understanding of what analysis steps should be for my case (only apply gravity and then seismic). Can you please check/correct?

1) add recorders
2) add soil domain - is this necessary since in this model I have only one stage (soil)?
3) add soil constraints - not sure if I need this as this might be done automatically by ASDAbsorbingBoundary3DAuto
4) add soil self-weight
5) activate absorbing boundaries - is this automatically going to switch from contraints to free field column?
6) apply damping
7) dynamic analysis

Please advise. It is still not clear to me how the automation for switching between static (supports) and dynamic (free field column) works in the ASDAbsorbingBoundary3D element.

Thank you!

[STKO Team]

Sorry I forgot to answer to this point:

3) add soil constraints - not sure if I need this as this might be done automatically by ASDAbsorbingBoundary3DAuto

Static constraints are internally taken care of automatically by the absorbing elements. In the SFSI model, I needed to add a constraint pattern because I was attaching 2 solids with incompatible meshes using the ASDEmbeddedNodeElement condition. So this is not necessary in your case.

2) add soil domain - is this necessary since in this model I have only one stage (soil)?

This is not necessary if you have only 1 stage. By default, the whole model is loaded (in OpenSees) at the very beginning. Instead, if you want to build your model by stages, you need to use the modelSubset analysis command.

activate absorbing boundaries - is this automatically going to switch from constraints to free field column?

Yes, you just need to use the AnalysisSteps.Misc_commands.ASDAbsorbingBoundaryActivate. Once you do this, the element, which in stage 0 behaves as a set of SP and MP constraints using the Penalty approach, will:

• store the reactions computed so far
• switch its computation to absorbing boundary + free field
• restore the reactions as external forces

Your steps should be:

1. add recorders
2. add soil self-weight
3. run static gravity analysis
4. activate absorbing boundaries
5. apply damping
6. dynamic analysis

[kolozvari]

Thank you for clarification. I have made the analysis steps in my model. I have a quick few follow up questions.

1) When defining "Soil Self Weight" condition in the sfsi example you used a volume force of -18631.4 in Z direction. How did you get this values and what is this values supposed to represent? In my model, I am defining material density rho in Mg, and I am thinking that my volume force should be -9.81, since my units are kN and m. Can you please clarify/confirm?

2) What is the difference in applying gravity load as static vs transient analysis? I see that in some examples you are doing static, and in some transient.

Thank you,

[STKO Team]

1) When defining "Soil Self Weight" condition in the sfsi example you used a volume force of -18631.4 in Z direction. How did you get this values and what is this values supposed to represent? In my model, I am defining material density rho in Mg, and I am thinking that my volume force should be -9.81, since my units are kN and m. Can you please clarify/confirm?

It's exactly a volume force (force per unit volume) which represents the self-weight of the soil, i.e. rho (1900 kg/m^3) * 9.806 (m/s^2).
Your volume force (kN/m^3) will be -9.81 if and only if you have a mass density of 1 Mg

2) What is the difference in applying gravity load as static vs transient analysis? I see that in some examples you are doing static, and in some transient.

No difference, a transient analysis with extremely large time steps does not produce inertia effects so they are the same. It was just a way to show the 2 alternatives

[kolozvari]

Thank you for clarification, it makes sense.

It seems to me that using static analysis for gravity is much less computationally demanding since it can be applied in essentially one step. Also, I am not sure how would I calculate body force for soil condition with multiple layers where each layer has different density.

So, I plan to use static analysis to apply gravity, similarly to what you did in model_3d_basic example (demonstrating the "old" way of implementing free-field columns) from April 6, 2020. In that example, you defined body force in the elements definition as well as soil density in the material definition. In my understanding, one is used for mass (density), the other one is used for gravity load (body force). Can you please confirm?

Thank you,

[STKO Team]

So, I plan to use static analysis to apply gravity, similarly to what you did in model_3d_basic example (demonstrating the "old" way of implementing free-field columns) from April 6, 2020. In that example, you defined body force in the elements definition as well as soil density in the material definition. In my understanding, one is used for mass (density), the other one is used for gravity load (body force). Can you please confirm?

yes. But also in that case (body force in element definition) you need to create as many element properties as the number of layers.
I typically use the VolumeForce approach, because it can be applied incrementally with a linear time series in a pattern, while the body force in the element is applied all at once. It can be a problem with a highly nonlinear material.

[kolozvari]

Hi! A quick holiday question.

I am trying to validate absorbing boundaries by comparing with theoretical solution for a sinusoidal motion of Sarma et al. I have noticed that when I have absorbing boundary at the bottom, the sinusoidal motion that I apply at the base of the absorbing boundary is not the same as at the base of the soil domain, which makes sense. However, this makes it very difficult to validate the model because I cannot apply exactly the input motion that I want. I tried to remove the bottom absorbing boundary and use fixed base, but I have convergence problems (no problems with absorbing boundaries at the base) and I have noticed that accelerations in the soil domain are unusual.

My questions are:
1) is the ASDAbsorbingBoundary3D element intended to be used without bottom layer of ASDAbsorbingBoundary3D? Is there anything in particular that I need to pay attention when doing that?
2) what if I increase stiffness of the ASDAbsorbingBoundary3D at the bottom? Would that give me what I need?

Please let me know your thoughts.

Merry Christmas and Happy New Year!

[STKO Team]

I have noticed that when I have absorbing boundary at the bottom, the sinusoidal motion that I apply at the base of the absorbing boundary is not the same as at the base of the soil domain

I did not fully understand this sentence.

1) is the ASDAbsorbingBoundary3D element intended to be used without a bottom layer of ASDAbsorbingBoundary3D? Is there anything in particular that I need to pay attention when doing that?

You should use it if you are modelling an elastic half-space at the bottom and you want your waves (reflected from the top free-surface) to be absorbed in the bottom layer.
The only reason for neglecting the absorbing boundary at the bottom is that you modeled the whole soil domain down to the bedrock, and you assume that the bedrock is infinitely rigid (not true anyway)

2) what if I increase stiffness of the ASDAbsorbingBoundary3D at the bottom? Would that give me what I need?

No, it will not work properly. The G modulus should be the same as the one of the soil.

Here is a quick tip:
If you have an acceleration history at the free-surface (out-crop motion) you have to apply at the base of the AbsorbingElement a velocity-time history obtained from a deconvolution.
If your soil column is homogeneous, you can avoid the deconvolution process for a simplified approach: apply half of the velocity

[kolozvari]

Thank you for your response. Sorry, maybe I wasnt clear in my previous message.

At this point I am jus trying to validate side boundaries by applying sinusoidal motion and verifying it with theoretical solution. I have noticed that when I have bottom boundaries the motion I apply as input is not the same as it is observed at the bottom of the soil, which makes it difficult to perform validation (see screenshot). I am applying motion of 1.0 m/s2 amplitude. Any suggestions on how to overcome this issue?


When I remove boundaries at the bottom to avoid this issue (only for the purpose of validation, not for real application), then I run into convergence problems. Note that I do not have any structure on top yet, so there shouldn't be any reflection of the waves in vertical direction. Any suggestion on how to improve convergence (soil is modeled as elastic with stdBrick element)?

Finally, in my understanding, accelerations that are extracted using STKO are relative accelerations. Please confirm.

Thank you,

[STKO Team]

I have noticed that when I have bottom boundaries the motion I apply as input is not the same as it is observed at the bottom of the soil, which makes it difficult to perform validation (see screenshot). I am applying motion of 1.0 m/s2 amplitude. Any suggestions on how to overcome this issue?

The motion in terms of accelerations that you see at the bottom of soil (not the bottom of absorbing elements, there it should be 0), should be exactly the motion that you input. But your input (time series assigned to the absorbing element) should be the velocity associated with the acceleration you want. For example:
Ax = acceleration you want at the base
You should apply Vx = integrate(Ax)
If the material properties (G, Poisson, and rho) of your absorbing boundary at the bottom match the ones of the soil model, you should have exactly the input you want.

When I remove boundaries at the bottom to avoid this issue (only for the purpose of validation, not for real application), then I run into convergence problems.

Of course, keep in mind that the bottom absorbing boundary also acts as a vertical constraint...

Note that I do not have any structure on top yet, so there shouldn't be any reflection of the waves in vertical direction.

I mean that the wave that travels from the bottom to the top, hits the free surface and is reflected. it will then travel downward and you should be able to absorb it...

Finally, in my understanding, accelerations that are extracted using STKO are relative accelerations. Please confirm.

It is exactly what opensees gives, we do not post process it... If you have a uniform excitation (fixed base) they are relative. If you have a multi-support-excitation it is absolute.
In this case, since you are applying the input as a boundary traction (your input velocity transformed to boundary traction by the ASDAbsorbingElement) the accelerations will be absolute, that's why you should see your acceleration at the base and not zero.

If your share your model here I will check what's wrong