Highly stressed volume approach in ANSYS apdl
Highly stressed volume approach in ANSYS apdl
(OP)
Hi all at Eng-tips
I i'm getting better and better at the apdl-language, even though i still think its a bit daunting. I've started to work on a technique to extract the highly stressed volume in a component for fatigue life assessment. In short, the highly stressed volume of the component (minimum ∽95* maximum of 1.st principal stress ) should (i'm going to investigate this further) correlate well with the fatigue strength of the component. (R. Kuegel first described the method as far as i know).
The code is currently as follows:
And the following nodes is selected and calculated the volume of:

The current methods accuracy is highly dependent on the mesh size. To improve performance some form of interpolation is needed.
Any questions or ideas for improvement are very welcome :)
Best, BTHS
I i'm getting better and better at the apdl-language, even though i still think its a bit daunting. I've started to work on a technique to extract the highly stressed volume in a component for fatigue life assessment. In short, the highly stressed volume of the component (minimum ∽95* maximum of 1.st principal stress ) should (i'm going to investigate this further) correlate well with the fatigue strength of the component. (R. Kuegel first described the method as far as i know).
The code is currently as follows:
CODE --> apdl
!Select Named selection cmsel,s,Area1,node !Get Maximum pricipal stress in selection and the number of the stressed node NSORT,S,1,0,0,1 *GET,NS_SMAX,SORT,,MAX *GET,NS_NMAX,SORT,,IMAX !Select nodes that has a stress of at least 0.9*NS_SMAX nsel,s,s,1,0.9*NS_SMAX !Select elements that contain at least one of the selected nodes esln,r,0,all !Get number of selected nodes *get,n_elem,elem,,count !State variables ENUM=0 tot_vol=0 !Sum up the volume of elements *do,i,1,n_elem ENUM=elnext(ENUM) *get,el_vol,elem,enum,volu tot_vol = tot_vol+el_vol *enddo
And the following nodes is selected and calculated the volume of:

The current methods accuracy is highly dependent on the mesh size. To improve performance some form of interpolation is needed.
Any questions or ideas for improvement are very welcome :)
Best, BTHS




