How big a crack or void can be in phase field fracture model?

[Anil Kunwar]

For the accuracy of the phase field fracture model, what can be the thickness (w) of a crack of length a/2 where a is the length of the square representing SENB (single edge notch bend) test specimen?
How about the radius(R) of the void for the similar model?
Yours Sincerely,
Anil Kunwar

[Wen Jiang]

Hi Anil,

Crack should not have "thickness". In phase field fracture, the length scale parameter is a material property that will affect the maximum tensile stress and  also introduces non-localicity to the formulation.

 Wen

[Anil Kunwar]

Hi Wen,
This is an exodus geometry/mesh i.e. crack_mesh.e of https://github.com/idaholab/moose/tree/devel/modules/combined/tests/phase_field_fracture  as visualized by peacock.
How was the crack thickness defined in the mesh? It is shown like a line therein Fig -->> Peacock-danphe-opt_015.jpg and Peacock-danphe-opt_016.jpg , which are the exodus mesh files within the tutorial.

Now, i have made a UNV-format mesh file with an initial crack of finite width (w=0.01 mm) for a square geometry of length a=1 mm.i.e. a/w = 100 (Figures: Unv_format_Peacock-danphe-opt_015.jpg and Unv_mesh_Peacock-danphe-opt_015.jpg ) . Will it produce singularity?


Yours Sincerely,
Anil Kunwar

Yours Sincerely,
Anil Kunwar

[Anil Kunwar]

Hi Wen,
maybe i could have used the term delta or deflection instead of w.

Yours Sincerely,
Anil Kunwar

On Tuesday, November 15, 2016 at 11:45:56 PM UTC+8, Wen Jiang wrote:

[Anil Kunwar]

The plot of damage variable c for the unv mesh with delta = 0.01 mm and a = 1 mm.

[Anil Kunwar]

The pic of damage variable mentioned above is uploaded here.

[Anil Kunwar]

Hi Wen,
Do we have a style on how we draw sizeless(w = delta ~ 0) crack in geometry and mesh?

Yours Sincerely,
Anil Kunwar

On Tuesday, November 15, 2016 at 11:45:56 PM UTC+8, Wen Jiang wrote:

[Anil Kunwar]

Hi Wen,
What i mean by crack here is the "pre-crack" of given width w or delta.

Yours Sincerely,
Anil Kunwar

On Tuesday, November 15, 2016 at 11:45:56 PM UTC+8, Wen Jiang wrote:

[Benjamin Spencer]

Anil,

  The width of your initial crack will affect the singularity somewhat, but I wouldn't expect that to affect your simulation results that much, especially because it looks like you plan to use a phase field fracture model to represent the propagating crack (since phase field doesn't represent the crack as a strong discontinuity anyway). The mesh you showed isn't refined enough to capture the singularity that well regardless of how wide the crack is. My rough guess is that as long as your crack width is a fraction of the size of an element at the tip, that finite crack width won't have an appreciable effect on the singularity.

-Ben

--
You received this message because you are subscribed to the Google Groups "moose-users" group.
To unsubscribe from this group and stop receiving emails from it, send an email to moose-users+unsubscribe@googlegroups.com.
Visit this group at https://groups.google.com/group/moose-users.
To view this discussion on the web visit https://groups.google.com/d/msgid/moose-users/26428475-69c6-445b-94be-ed32764a7197%40googlegroups.com.

For more options, visit https://groups.google.com/d/optout.

[Anil Kunwar]

Hi Ben,

The mesh you showed isn't refined enough to capture the singularity that well regardless of how wide the crack is. My rough guess is that as long as your crack width is a fraction of the size of an element at the tip, that finite crack width won't have an appreciable effect on the singularity.

Thank you for the explanation regarding the pre-crack width (delta ) for phase field fracture simulation. Cheers.

Yours Sincerely,
Anil Kunwar

On Thursday, November 17, 2016 at 1:13:49 AM UTC+8, benjamin.spencer wrote:

Anil,

  The width of your initial crack will affect the singularity somewhat, but I wouldn't expect that to affect your simulation results that much, especially because it looks like you plan to use a phase field fracture model to represent the propagating crack (since phase field doesn't represent the crack as a strong discontinuity anyway). The mesh you showed isn't refined enough to capture the singularity that well regardless of how wide the crack is. My rough guess is that as long as your crack width is a fraction of the size of an element at the tip, that finite crack width won't have an appreciable effect on the singularity.

-Ben

On Wed, Nov 16, 2016 at 1:34 AM, Anil Kunwar <romagu...@gmail.com> wrote:

Hi Wen,
What i mean by crack here is the "pre-crack" of given width w or delta.

Yours Sincerely,
Anil Kunwar

On Tuesday, November 15, 2016 at 11:45:56 PM UTC+8, Wen Jiang wrote:

Hi Anil,

Crack should not have "thickness". In phase field fracture, the length scale parameter is a material property that will affect the maximum tensile stress and  also introduces non-localicity to the formulation.

 Wen

On Sunday, November 13, 2016 at 11:41:59 PM UTC-7, Anil Kunwar wrote:

For the accuracy of the phase field fracture model, what can be the thickness (w) of a crack of length a/2 where a is the length of the square representing SENB (single edge notch bend) test specimen?
How about the radius(R) of the void for the similar model?
Yours Sincerely,
Anil Kunwar

--
You received this message because you are subscribed to the Google Groups "moose-users" group.

To unsubscribe from this group and stop receiving emails from it, send an email to moose-users...@googlegroups.com.

Visit this group at https://groups.google.com/group/moose-users.

[Anil Kunwar]

Hi Ben and All,
For the examples https://github.com/idaholab/moose/tree/devel/modules/combined/tests/phase_field_fracture and https://github.com/idaholab/moose/tree/devel/modules/combined/tests/phase_field_fracture_viscoplastic, if we draw a filemesh (in unv file format) with geometry identical to the example but of a variable pre-crack dimension, what would be the relation of the mesh element size with the convergence criteria? I am getting some convergence issues.

Yours Sincerely,
Anil Kunwar

On Thursday, November 17, 2016 at 1:13:49 AM UTC+8, benjamin.spencer wrote:

Anil,

  The width of your initial crack will affect the singularity somewhat, but I wouldn't expect that to affect your simulation results that much, especially because it looks like you plan to use a phase field fracture model to represent the propagating crack (since phase field doesn't represent the crack as a strong discontinuity anyway). The mesh you showed isn't refined enough to capture the singularity that well regardless of how wide the crack is. My rough guess is that as long as your crack width is a fraction of the size of an element at the tip, that finite crack width won't have an appreciable effect on the singularity.

-Ben

On Wed, Nov 16, 2016 at 1:34 AM, Anil Kunwar <romagu...@gmail.com> wrote:

Hi Wen,
What i mean by crack here is the "pre-crack" of given width w or delta.

Yours Sincerely,
Anil Kunwar

On Tuesday, November 15, 2016 at 11:45:56 PM UTC+8, Wen Jiang wrote:

Hi Anil,

Crack should not have "thickness". In phase field fracture, the length scale parameter is a material property that will affect the maximum tensile stress and  also introduces non-localicity to the formulation.

 Wen

On Sunday, November 13, 2016 at 11:41:59 PM UTC-7, Anil Kunwar wrote:

For the accuracy of the phase field fracture model, what can be the thickness (w) of a crack of length a/2 where a is the length of the square representing SENB (single edge notch bend) test specimen?
How about the radius(R) of the void for the similar model?
Yours Sincerely,
Anil Kunwar

--
You received this message because you are subscribed to the Google Groups "moose-users" group.

To unsubscribe from this group and stop receiving emails from it, send an email to moose-users...@googlegroups.com.

Visit this group at https://groups.google.com/group/moose-users.

[Wen Jiang]

For phase field fracture, you need at least 2~3 elements within the steep transition zone. So the element size will depend on the pf parameter you choose. The convergence might not only be related to the mesh size, but also other pf fracture parameters, like the viscosity.

[Anil Kunwar]

Wen,
Your suggestion regarding the requirement of 2~3 mesh elements has proved to be an important design criteria for mesh in UNV format. This has really worked in the realization of evolution of phase field based damage variable. Previously, i could not find any change in the variables namely, c, b, disp_x,disp_y. This design criteria requires me to build a robust mesh from which i can realize the evolution of the variables. Thank you for it.
I tried with viscosity and found that it did not affect so much in this context. The major factor was the time step size i.e.  when i reduced it from 0.1 to 1.0e-8; the solve converged.

Yours Sincerely,
Anil Kunwar