Ask for an example for simulating damage

Hello,

> some research on fracture damage

could you please be more specific?
Typical scenarios, materials, scales, times, ..........

There are a few distinct/independent approaches in Yade to simulate damage and fracture.
If you provide more information, we can give you more specific (then just very general) answers.

Cheers
Jan

Hi, Jan

Thank you very much for your kind reply.

My topic is to simulate the interaction between sea ice, a brittle material, and ocean structure

and study the fracture process of sea ice in engineering scale.

Thanks for your kind help.

Shenyu

Hello,

thanks for some info (although it is still very general).

> ocean structure

What kind of structure?
Some massive, possible to simulate as a rigid body?
Or some not so stiff/strong and also the structure would need actual simulation (deformation, collapse)?

> the fracture process of sea ice in engineering scale

What is "engineering scale"?

Why do you want to use DEM and not e.g. FEM or some of its derivatives (XFEM)?

I **personally** see the advantage of DEM for the scenarios, where you have massive fragmentation / crushing and the products of crushing influences significantly the further process.
For the case of ice, I would expect cracking and splitting of the ice into a few pieces, which do not significantly influence each other.
Do you have some experimental results or references?

What is / do you expect the typical "mode" of fracture?
E.g. local ice crushing due to local pressures of the ocean structure, splitting an iceberg into a few smaller icebrgs, ...

How would you simulate water?

Of course, ice breaking can be somehow modeled by DEM or DEM-like method(s), just some brainstorming..

Cheers
Jan

Hi, Jan
Thank you very much for your kind reply.

> ocean structure

ship, simulate as a rigid body.

> the fracture process of sea ice in engineering scale

About "engineering scale", what I want to express is that I study the macroscopic breaking process of sea ice.

There are many types of sea ice, I am mainly concerned with packing ice, and the conditions I've studied are similar to [1].

 Hundreds of packing ice float on the water, we simulate the motion of the ship under this condition. Packing ice breaking needs to be considered in the simulation. That's why I want to use DEM.

>What is / do you expect the typical "mode" of fracture?

About that, I'm not very clear about this either. At present, it is mainly bending and breaking

>How would you simulate water?

Temporarily consider using theoretical or empirical formulas to approximate the effect of water on packing ice.

Thanks for your kind help.
Shenyu

[1] https://doi.org/10.1016/j.oceaneng.2020.108264

Hello,

thanks for clarification, now the information is sufficient enough for solid discussion.

Basically you have myriads of options :-D
What choices are good or not is very difficult to tell..

What time span do you have for the project? (one month short project, master thesis, PhD thesis, some large project, ...)
In what environment is the project (academia, industry, ...)?
What computer power do you have (one laptop, one good desktop, a few computers, a HPC cluster, ...)?

just some brainstorming:

individual icebergs:
- modeled as rigid body:
    - straightforward
    - difficult to model breaking (some indirect method for cracking would be needed, e.g. based on external forces)
    - no inertia inside iceberg
    - computationally "cheap"
- modeled as elastic (rigid particles connected with deformable bonds):
    - can capture actual inertia effects inside the iceberg
    - bonds can be breakable, thus it would be possible to model somehow the breaking
        - very challenging thought
    - more computationally "expensive"
- shapes:
    - spheres / circles
    - clump / aggregate of spheres / circles
    - polygons / polyhedrons

Water can probably be neglected in the beginning stage, its incorporation would be further step in complexity.

Something, you can try directly and get some feeling:
- CPM (model for concrete)
    - examples/concrete/uniax.py [1], damage under tension
    - examples/concrete/brazilian.py [2], damage under
- Polyhedrons
    - examples/polyhedra/splitter.py [3], splitting polyhedrons based on external forces
- jointedCohesiveFrictionalPM
    - surely option to consider for the iceberg cracking, unfortunately I have no experience with it and can give no advice myself :-(
- some your new solution :-)

Cheers
Jan

[1] https://gitlab.com/yade-dev/trunk/-/blob/master/examples/concrete/uniax.py
[2] https://gitlab.com/yade-dev/trunk/-/blob/master/examples/concrete/brazilian.py
[3] https://gitlab.com/yade-dev/trunk/-/blob/master/examples/polyhedra/splitter.py

Hi, Jan
Thank you very much for your kind reply.

It's about my PhD thesis, and the research group has an HPC cluster.

So, I am now going to simulate ice breaking with sphere particles and bond models.

And I determined the geometry of the ice in the computational domain and the Sphere particle coordinates used to fill the geometry[1].

But I still have some questions, how is the bond between particles established before the start of the simulation?

In discrete element commercial software, such as EDEM, these bonds are created with the particles and broken during simulation

Shenyu

[1]https://www.dropbox.com/s/7b3eiqyyupp54xw/particle.jpg?dl=0

> So, I am now going to simulate ice breaking with sphere particles and bond models.

Personally I am not sure if this is the best approach, but you will see :-)
- if the packing is very regular (like in the provided picture), the cracks are too much "predefined"
- most domain most of the time does not need to be actually simulated (so the simulations will be unnecessarily "long"). Actually this is my main concern, but maybe it is not relevant..
- I am not sure if the "clear breaking" of the ice is doable by DEM models (but I have never tried). E.g. the proposed CPM model creates "larger damaged zones" rather than real cracks.
There is also CohFrictMat, which have cohesive bonds that may break, but with this too I have almost no experience (you can ask separate new question on the possibility to use CohFrictMat for iceberg breaking).
You can also use completely new Yade model dedicated for ice breaking.
For sure you can make it work somehow this way :-)

I like the idea that the icebergs are rigid body (either clump of spheres or some polygon), the "overall" motion of these large icebergs is solved by DEM and the breaking is solved for individual icebergs on demand (based on external forces and momentum) "externally" - either by DEM or other method.

But it is just some brainstorming, ideas which seems good to me, but might not work at all when it is actually done :-)
After all, it is your PhD, you will have to do all the work yourself, it should be coordinated with your supervisor, you yourself will have to present and defend it etc. etc.
So really take it just as a "non-binding" advises and make the final solution yourself :-)

> how is the bond between particles established before the start of the simulation?

There are several approaches.
You can define the bonds "manually" using utils.createInteraction [4].
Or you can let the bonds be created "automatically" based on their radius, interactionDetectionFactor and aabbEnlargeFactor as in uniax.py [1].

I would suggest to consider also the JointedCohesiveFrictionalPM.
But again, I have no personal experience, so you can ask a separate new question on this topic (if the model could be used to model ice breaking).

Cheers
Jan

[4] https://yade-dem.org/doc/user.html#individual-interactions-on-demand

Hi, Jan
Thanks for your brainstorming about my question, which is important for me.

There is a Chinese proverb called "crossing the river by feeling the stones". It means that in the face of new problems, we can only keep trying.

Good luck for us!

Thanks again for your kind help.

Shenyu