User login


You are here

Crack propagation simulations

Hi all,

Apologize if a similar thread was discussed elsewhere. I wanted to know as to currently, what is the

best method to simulate large geometries with cracks as in 3 or 4 point bending test and get factors like CTOD

etc. from such simulation in a physical manner.

I am also interested in crack propagation. As I understand FE requires a prior knowledge of the crack propagation.

What are the disadvantages of XFEM in terms of accuracy? Is there any review on discussing various methods including

phasefields applicable to large scale problems. I am interested in experimental strain rates and not the strain rates used

in Molecular dynamics. Also anyone has experience in using the commerical tools like Abaqus, or even the open source codes

that can be used?



marco.paggi's picture

The phase field model of fracture has seen significant progresses since the early stages, and nowadays it becomes a quantitative methodology to simulate crack growth without the a priori knowledge of the crack path, see e.g.:

Large scale computations are still expensive and generalization to dynamic crack growth still in progress. For a better understanding of the role of the length scales involved in the approach, and therefore CPU limitations in the case of large problems, please see a thoroughly discussion in the articles mentioned in my previous blog entry.

As ever,

Marco Paggi

Dear Professor Paggi, Thank you for the reply. I will check out the link and follow the publications.

The phase field method seems to be used in research more and more now. But if possible can you also shed light on methods to simulate fracture in industry? 

Really appreciate your reply.


marco.paggi's picture


So far we can classify fracture mechanics theories according to the following main categories:

- brittle fracture (LEFM)

- cohesive fracture (NLFM)

- ductile (elasto-plastic) fracture

NLFM can include LEFM as a limit case when the process zone size tends to zero.

Computationally, if you know the crack path in advance (an interface with adhesive, for instance) then you may use interface elements, which are like the segment-to-segment contact elements with fixed pairing. They have also been extended to node-to-segment or node-to-surface finite elements, which are useful to couple different mesh discretizations at the interface:

If the crack path is unknown a priori, then you may opt for remeshing techniques and singular FEs at the crack tip for LEFM, see FRANC2D by Cornell fracture group, which is freely available and I also used in the past. However, the issue of crack nucleation, branching and similar phenomena are not included in that code. 3D simulations are also complex, not so straightforward for industrial research.

XFEM and phase field are powerful and the phase field, in my opinion, opens new perspectives for 3D simulations, and it can be coupled with interface elements for cohesive fracture. As a limitation, the fine mesh you need to use, which can be problematic for big industrial simulations.

Dynamic fracture deserves a specific discussion, if it is interesting to you.







Subscribe to Comments for "Crack propagation simulations"

More comments


Subscribe to Syndicate