SIF semi-elliptical crack in pipelines

Dear all,

I am working in my under-graduate thesis. Basically, i'm trying to calculate the SIF and J-Integral for semi-elliptical surface crack using CINT command in ANSYS APDL (all of this in a pipeline model). I have seen one paper shows SIF's distribution along the crack front like the following picture (fig. 8):

In my analysis I have obtained the following distribution (I have used a/c=0.5 with a/tc and tc/Rc equal to the fig. 8):

My issue is that I don't know why the curve I've obtained has that behaviour. In the course of my study I have calibrated the mesh characteristics with a plate model obtaining good aproximations respect to the Raju-Newman Model, but using the same characteristics in my pipe model I obtained the results shown above:

I will be very gratefully if someone can help me with some advices to solve this problem.

Br.