3D J-INTEGRAL

[whiteram]

Hi guys,
I am simulating crack propagation in a cylindrical pressure vessel in 3D using our in-house FE code.I need to determine the J-integral using the JVCE (virtual crack extension)option. I seem to have done everything right however, the program reports a warning/error stating that 'the crank flanks have collapsed or are non-planar'. I have viewed the mesh and there doesn't appear to be any form of physical collapse in the regions referred to.

Could someone please tell me what this message means and how to resolve this problem. Your assistance/contributions will be greatly appreciated.

Note:the crack is elliptical

Regards,
whiteram

 

[EdR]

If it's an in-house FE code ask your in-house expert on the code.....There is no way anyone not familiar with the specific code can answer this type of question....

Ed.R.

 

[GeoSmith]

u use abaqus or ansys?

 

[whiteram]

Hi Guys (EdR & GeoSmith),
I am aware that this problem could be peculiar to the programme used. However, I was hoping that someone may have encountered a similar problem using other softwares. I understand that the computation of stress intensity factors along curvilinear crack fronts is generally a challenging task...

I am looking for tips on how to resolve this. Plz help

Whiteram

 

[afeauser]

Well, this is a linear 3D FEA problem, and the virtual crack extension method has been used extensively.
So this is not a difficult task at all for people with
a graduate degree.

Perhaps you should contact the developer of the
in-house FEA program, to resolve this issue:
(i) it may be due to a code bug; (ii) or maybe the
user does not understand the fracture mechanics
algorithm or how to use the code.

 

[xerf]

I think J-integral determination by virtual crack extension is valid only for linear elastic fracture mechanics when J integral has the same meaning as energy release rate G, otherwise a different computational approach should be used. Do you have plasticity involved ??

If the crack region is under compression, the crack surfaces will tend to come in contact. If the crack surfaces are not modeled through a contact approach they will penetrate each other, which will lead to numerical difficulties;it is possible the message you mentioned to be caused by this.