Due to the nature of stress-based criteria (stress measure = strength limit), damage initiation tends to be over all areas of a structure where the criteria is met.  For example, if you conduct FEM of a dog-bone specimen with idealized material structure and spatially constant strength properties, damage initiation will be predicted over the entire gage section. However, in practice strength distribution will be heterogeneous and/or stochastic ( manufacturing induced defects, void/microstructure variability). Certain locations will have stress concentrations as well as lower strength. The damage initiation will be localized.  Therefore, damage initiation and crack location analyses can be meaningless without a model for spatial distribution of strength (both deterministic and stochastic descriptions)  and effects of the defects on the computed stress fields. 

Thank you very much Mr. Kishore,

You meant flaws that presented  in Glass surface. However it will be difficult to assume exact flaw size and its direction in XFE analyses. Even if we assumed, the results will be different when compared with experimental value. Therefore it is better to avoid those kind of approximation (in my point of view). I am using brittle material (Glass).

Thank you.

Pandian ...

Hello, Pandian

I know it is very late but may i learn if you could solve that problem? In my simulation i am facing similar problem.

Regards

Berkay

Hi Berkay,

By selecting appropriate damage initiation stress and fracture energy for the considered materials, you may avoid the multiple damage initiation in the XFEM model. In addition, the element size also plays an important role to define the damage initiation. I would recommend you to refer "K.P. Marimuthu et al. / Journal of the European Ceramic Society 37 (2017) 381–391" for further clarifications.

Good luck!

Regards,

Pandian

Hi Pandian,

I have read your XFEM simulation of indentation on glass, which I have found very insightful.

I am trying to reproduce the results of this paper. I have used the same material properties and a mesh size of 10 micron. I apply a displacement of 0.1 mm to the indenter semi-circular rigid surface.

In my simulation one crack develops outside the contact radius but then it stops to propagate since, as the indentation proceeds, the stress at the initial crack becomes compressive. Moreover, I don't have any new cracks than initiate at the new point of maximum tensile stress outside the current contact area.

I see from Figure 5 of your paper that you have many cracks that initiate, and only the last one that propagates. I was wondering if you could suggest what I should do to get the same results. 

Best regards,

Francesco

 Hi Francesco,

Using only one enriched zone (EZ) for the inner region we observed that damage propagation ceases during the analysis and that damage does not initiate in other regions although stresses are well beyond the damage initiating threshold. However, by dividing the inner region into several EZ, which are arranged such that elements of one and the same EZ do not share nodes (Figure), we observe smooth damage propagation and even merging. Eight EZs (in 3D case) and four EZs (in 2D case) are necessary.

Hope it will helpful to you.

Best regards,

Pandian

Hi Pandian,

thanks for your reply and very good job!

If it can help I managed to get similar results by introducing initial small cracks at different locations within the same domain. 

Best regards,

Francesco