Separation of elements in deal.ii

[jessie]

I would like to ask if deal.ii is capable of simulating failure due to separation of the elements such as collapse and breakage since the structure is brittle in nature(masonry)? It will be coupled with blastFOAM using preCICE.

Thankyou.

[jessie]

I am expecting this kind of output (https://761vqp-my.sharepoint.com/:v:/g/personal/jessiecorralesb_761vqp_onmicrosoft_com/EfIevVBuMENGipXNVubLMnMB580ZpVOboOoIbpskuopeGA?e=1h5Uwc). Can I do this in deal.ii?

[Wolfgang Bangerth]

> I would like to ask if deal.ii is capable of simulating failure due to
> separation of the elements such as collapse and breakage since the structure
> is brittle in nature(masonry)? It will be coupled with blastFOAM using preCICE.

"Capable" is a word that can have many meanings. If you have several years of
work, you can probably do that. If all you care about is computer game
graphics like several of the snippets in the video, it shouldn't even be very
difficult to just do this with simpler methods. But if you want it to be
accurate and realistic, then this is a major undertaking on the scale of a PhD
project or maybe multiple -- which you can do in deal.II, but you shouldn't
expect it to be easy (nor will it be in any other finite element library,
simply because the *physics* of what you want to do is difficult).

Best
WB

--
------------------------------------------------------------------------
Wolfgang Bangerth email: bang...@colostate.edu
www: http://www.math.colostate.edu/~bangerth/

[jessie]

The idea that is on my mind is basically two parts. The first part is that that an explosive charge (say TNT) inside a structure will be detonated that will cause shockwave and gas pressure inside the structure, so I need to determine the pressure it exerts to the confining wall (CFD this will be done in openFOAM). Then the second part would be the response (strain and stresses to be in deal.ii) of the walls including cracks,damage, fractures or failures (if there are any, due to the high pressures) using FEA.  So basically, I wanted to couple this two procedures, CFDxFEA in order to have a more exact simulation and result which results to a dynamic remeshing or anything like this for every time step until the explosion fades. I would like to ask if deal.ii is capable of simulating failure due to separation of the elements? This will be done in blastFOAM-preCICE-deal.ii. Hoping for your response at your convenient time. Thank you and more power!

[Wolfgang Bangerth]

On 4/10/23 00:51, jessie wrote:
> The idea that is on my mind is basically two parts. The first part is
> that that an explosive charge (say TNT) inside a structure will be
> detonated that will cause shockwave and gas pressure inside the
> structure, so I need to determine the pressure it exerts to the
> confining wall (CFD this will be done in openFOAM). Then the second part
> would be the response (strain and stresses to be in deal.ii) of the
> walls including cracks,damage, fractures or failures (if there are any,
> due to the high pressures) using FEA.  So basically, I wanted to couple
> this two procedures, CFDxFEA in order to have a more exact simulation
> and result which results to a dynamic remeshing or anything like this
> for every time step until the explosion fades. I would like to ask if
> deal.ii is capable of simulating failure due to separation of the
> elements? This will be done in blastFOAM-preCICE-deal.ii.

As I already said in the other email, "capable" can mean many things.

People have spent their whole careers trying to come up with models that
accurately describe cracks, damage, and fractures. It is a very
difficult field. By itself, this would make for a good PhD topic. Of
course, if you're not interested in accuracy and only want to produce
something that looks reasonably realistic for the purposes of movies or
computer games, then that's a separate issue.

Either way, coupling this with a CFD solver on a domain that changes
dynamically in response to the failure of the solid is, also, quite a
difficult area.

So yes, it's *possible* to do what you want. But it will likely take
several years of work to come up with accurate model equations and
parameters, and an efficient implementation, to do what you describe.

Best
W.