Atom topic feed | site map | contact | login | Powered by FluxBB | réalisation artaban

You are not logged in.

- Topics: Active | Unanswered

Pages: **1**

**psmith****Member**- From: California
- Registered: 2018-06-27
- Posts: 6

All,

I am looking for an option to implement semi-infinite boundary to my problem involving soil mass. Looking through the documentation, one possible option is to use CALC_MISS (but the detailed documentation is restricted and my understanding is quite limited). In the wake of such, could anyone point me towards the right direction in finding an appropriate boundary condition that is already available for general use in Code_Aster to mimic semi-infinite boundary conditions?

Thanks.

-----

Paul

Offline

Hi Paul

Take a look at 3D_ABSO, document U3.14.09 in the documentation section. This element provides a 'rubber band mesh' on the boundary that mimics anechoic conditions. It is not perfectly anechoic though.

Of interest may also be the possibility to excite unbounded regions using impinging wave sources, ONDE as well as the possibility to excite using Fourier Modes.

As far as I understand, MISS3D is an analytic layered soil package that coupled to Code Aster (CA). You should be able to get a perfect anechoic BC with this sort of coupling though at a considerably larger work/modelling effort. One advantage is the ability to excite/pick up response at a distance outside of the model.

For what it is worth, I was involved in a project where we did something similar with a 5M element structure model in Nastran, a code similar to MISS3D plus a supercomputer. Solution time was in the ballpark 1 Hr/Hz.

Then, there always is the Hilla Billy approach. You can make a large model with physically modeled outer layers that have matching soil properties where you tweak up the loss factor. These outer layers can be sampled wrt to the wavelength and the model changed with frequency. A bit clumsy, but it does the job.

I'd try 3D_ABSO first.

Sincerely

Claes

Offline

**psmith****Member**- From: California
- Registered: 2018-06-27
- Posts: 6

Hi Claes,

Thanks for your valuable inputs on this. 3D_ABSO might be acceptable solution for my problem. I will revert back once I have implemented this in my model.

Work carried out by you on 5M element structure model in Nastran sounds quite interesting. Any chance that this is available in the form of report/paper somewhere?

Thanks again.

Paul

P. S. I hope core documentations and validation of CALC_MISS will be available for general public in the future.

Offline

Hi Paul

I cannot post links on the forum, but try and google the terms maxlab vibration Mechanical Stability and you should find some of the work.

I believe there were some papers written by the dept of Building Mechanics at Lund University on the work. We did use an old Nastran trick to help speed things up.

You know, the analytic model is first use to compute the green field response. You take out response points where the nodes of your FE mesh should go. Next, you must dig a hole by inverting this matrix and putting zero pressure where the hole in the soil is. This step is quite laborious. We avoided it by instead inserting two meshes. The first mesh is the structure that goes into the hole, the second mesh is a regular mesh of the layered soil - but, with negative material data. The latter mesh digs a hole while avoiding matrix inversion.

/Claes

*Last edited by CLF (2019-03-16 23:45:57)*

Offline

**psmith****Member**- From: California
- Registered: 2018-06-27
- Posts: 6

Hi Claes,

I tried using 3D_ABSO (as also detailed in validation problems SDLV120A and SDLV121A). My simple model (1m x 1m x 1m) runs successfully for transient analysis, but when the model is switched to harmonic analysis, the model fails as run is carried out using two models in one (3D and 3D_ABSO) for harmonic analysis. Any suggestions on the workaround? Thanks.

I tried looking into U3.14.09 for acceptable analysis using 3D_ABSO, but could not find one (though the manual states that only elastic behaviour is accepted).

Paul

Offline

Hi Paul

Sorry, I cannot help you there other than stating that the rubber band mesh will never be perfectly absorbing.

I suggest that you first try using clear cut cases, e.g. like a pipe where you constrain wave types to just one type - just to make sure there are no bugs. You should then be able to compare with the data you get for a situation with an impedance boundary condition.

The forum and developers may know more on this issue as I have worked only vibroacostic problems so far, i.e. not dealt with multiple wave types in soils and such.

/C

Offline

**psmith****Member**- From: California
- Registered: 2018-06-27
- Posts: 6

Hi Claes,

I appreciate your help in pointing towards the right direction in the first place. I look forward to getting some response on this from other members of the community and/or developers.

I will be posting, if I manage to solve this issue.

Thanks again.

Paul

Offline

Pages: **1**