NLGEOM not converge (XFEM)

[RachelClinton]

Dear all?I'm using XFEM to model crack propagation. There is convergence problem whenI turned on NLGEOM. MSG file reminds excessive distorting at crack region,or sometimes reminds cracked elment might not be calculated correctly,or too many attempts. Itworks well if turned off NLGEOM. Moreover, if I change input elastic-plasticproperties to linearly elastic properties, it works well even when NLGEOM isturned on. Anybody has some idea of this? I tried stabilization function at*step but failed to resolve this problem.
Attached is my .inp file.

THANKS for any help.

RACHEL