Learning Forum

[Jose Mishael C J]

I am trying to simulate fatigue crack growth on a tubular joint. I was using the fracture analysis reference for setting up the simulation in APDL. The script is running perfectly, but there are no solutions obtained for SIFs, number of cycles and crack growth.

My crack is modeled with MESH200 elements and SOLID185 elements are used for modelling the tubular joint. The crack is defined by following the same procedure explained under MESH200 elements. The crack is supposed to propagate through the thickness of the tubular member.

When I checked my *.err file, I can see multiple warnings. I couldn't understand some of the warnings and I am pretty sure that these warnings are affecting my solutions. The errors are;

1. Fracture parameter calculation issue: Contour integration for crack 1 shows that some of contours are beyond the FE model and are to be modified. Please carefully evaluate the reported contour integral results. - I tried to modify my contour numbers multiple times and still getting the warning/error.
2. Fracture parameters KI on all selected contours are negative at tip node Id 566231, cgrow Id 1 and crack id 1 during XFEM-fatigue crack  growth calculation. DAMN is used for this tip. - This error is repeated over multiple times.

Can someone help me in solving these errors?

[danielshaw]

Why are you using XFEM for crack-growth? I recommend using SMART.
The warning message indicating that some contours are beyond the FE model is to identify an insufficiently refined mesh. The fracture parameters are calculated on contours radially outward from the crack tip. The contours are essentially centered on the nodes along the crack propagation path (contour 1 on 1st node, contour 2 on 2nd node, etc.). For thin-walled models with a coarse mesh, the outer contours can extend outside the FE model, which would produce incorrect K values. The warning message is intended to flag that situation. However, that warning message can also be issued if the first contour occurs on a free face. In that case, the warning message can be ignored.
What is the stress field at the crack tip? Is the stress normal to the crack propagation direction compressive?

[Jose Mishael C J]

The simulation completed without any stress singularity at the crack-tip. The von-Mises stress distribution is as shown in figure. The applied load is compressive and I feel that the load is making a crack closure effect than crack growth effect.
The XFEM module is clearly explained in the Fracture Mechanics Analysis reference and I felt that XFEM implementation is straight forward in APDL especially for multiple cracks. The SMART module features are mainly available only in workbench. Right? Like multiple crack simulations. Is SMART crack growth with multiple cracks feature available in APDL?
Have you done any XFEM simulations with multiple cracks in ANSYS? When I try to simulate the crack growth of a double edge notched specimen (Similar to the one shown in fracture mechanics analysis reference under Fatigue Crack Growth analysis of SEN specimen), ANSYS is returning an error as 'Both life-cycle and cycle-by-cycle methods for fatigue crack growth arefound in the same model.Only one method is allowed in one simulationmodel.Please check input data.' I have specified only life-cycle method. But I have two cracks with different IDs. Can you tell me what is the reason for this error? I couldn't find any examples of XFEM simulations with multiple cracks in ANSYS.

[danielshaw]

XFEM and SMART were both developed in MAPDL. SMART is considered to be the most robust crack growth simulation method so it is exposed in Mechanical. Both methods are supported in MAPDL I recommend using SMART, because most crack growth development is focused on SMART.
SMART supports multiple cracks. It does not support cracks merging. If 2 cracks merge, SMART will fail with a meshing error.
There error message is clear, but it might be issued incorrectly. I recommend running SMART on each crack independently to ensure that the parameters are correctly defined. If both cracks solve independently and you are certain that both cracks use the life-cycle method, then please submit the model to technical support for review.

[Jose Mishael C J]

Can you provide me some insights on the issues I posted on the following link?
/forum/discussion/36448/smart-crack-growth-with-semi-elliptical-crack#latest