SIF Accuracy - Abaqus XFEM

Hello,

 I understand from word of mouth, as well as an extensive study/Master'sThesis (http://publications.lib.chalmers.se/records/fulltext/164269.pdf), that Abaqus XFEM formulation works, but that processing of results to obtain reported SIF (KI, KII, KIII, not sure about J, others) are in error due to apparent bugs in Abaqus's algorithm to calculate the SIFs.

Does anyone know of open source and/or commercial add-ons/plugins/postprocessing tools that can take XFEM output and process properly for accurate SIF (and J, C*, etc)?  Ideally, that code would be able to provide such calculations for stationary cracks as well as propagating cracks, as well as be able to feedback such information to Abaqus to modify the crack front for evolution of the crack. 

Thanks!

mcgrete