Learning Forum

[Caleb Wood]

Hello, I have a model which contains water (yellow), air (brown), and water with a damping coefficient (red). The water is modelled using MAT_NULL and the air is modelled using MAT_VACCUUM. For some reason there is a blank area where the two water types meet. My process for filling the volume is as follows:

1. Fill entire geometry with air
2. Fill water portion using BOXCOR 
3. Fill the damped water portion using BOXCOR

The boxes are defined so that they meet at the exact same point. I have tried altering the value of NSAMPLE but it has not changed anything. I have seen S-ALE examples online and none of them appear to have this issue which I am having. Does anyone know why this is occuring?

[ido]

Hello Caleb,

LSPREPOST represents the AMMGs via their volume fractions. At the corner, it will reconstruct the interface, in this case, with roughly a 45 deg slanted plane. It is just an artifact. It is not wrong. The finer the mesh the smaller this gap is. If you are curious, look "Young's modified Interface reconstruction methods". It is very tricky and complex. But it does the job.

Regards,

Ian Do

[Caleb Wood]

Thank you, Ian. Maybe you can also answer this other question I have. I am trying to embed a pile into the water surface so that it is partially submerged at t = 0 sec. To do this, I believe I need to remove the water from the footprint of the pile and replace it with MAT_VACCUUM. The issue is when I try to fill the footprint of the pile with MAT_VACCUUM, much of the water remains behind (see photos 1 & 2). This creates instabilities when I run the model, and disturbs the water surface (see photo 3). What would be the correct way of modelling a pile embedded in water ALE elements? (The water in the below photos is no longer yellow, it is brown)

Photo 1

\

Photo 2

Photo 3

[ido]

Hi Caleb,

Mesh resolution issue - right? You have about only 2 ALE elms spanning the width of the pile! The mesh is too coarse so it simply cannot resolve the ALE matrial interface details you need. There is no magic. Just finer mesh and it will look better. But then finer mesh produces too big a model! There engineering judgement has to come into play. That I am not sure how to get around. 

On the other hand you can try a test run and check the force|P on the pile, then see how good it is. Then make adjustments based on that. A mesh resolution study is always a good thing to do to understand your model characteristics.

Regards,

Ian 

[Caleb Wood]

That was my fear. I was hoping there was a trick I was missing which would help fill the volume as I am not sure how fine I will be able to make my mesh since I plan on applying this to a large-scale problem with elastic material models. Maybe I can try to use a fine resolution around the piles and other areas of interest and a coarser resolution in the areas where the wave is traveling across the tank as it seems to be less influenced by the size of the mesh. Thank you!

[Peter Yip]

For your first question, I had posed a similar one a few months ago about why there is a void between two neighboring materials. Turns out that there is quite a body of research in how to handle the contact between differing materials in an ALE framework (especially at the eulerian step/phase of the numerical procedure) and their interface reconstructions. Here is the first link that started my journey in mixture theory and its intricacies. It easily can spiral into many reading hours of interesting papers.

https://escholarship.org/content/qt1fr314x9/qt1fr314x9.pdf

For your second question, I am a bit confused on your goal. It could be that I don't know what you exactly mean by "embed a pile into the water surface"? Does a "pile" of something (rock or whatever) not have material strength? Regardless, one key thing is that the order of the Multi-Matierial Groups DO matter. I would just keep that in mind because your results will differ if the order in which you have the material groups is not what you intend. When I have many materials it is easy for me to get lost as to what order I am trying to have so I thought it would be good to bring up here. Here is a link to the reasoning from Hao Chen. 

https://ftp.lstc.com/anonymous/outgoing/hao/sale/tutorials/Structured%20ALE%20Tutorial%202.pdf

Peter