I know there was a recent question on precision, but I didn't want to hijack that thread.

I am running into precision issues with displacement boundary conditions created via analytical expression fields. For example: I have a square domain with two displacement boundary conditions on adjacent edges (shown graphically in the attached figure). At the common corner node, both expression fields mathematically resolve to v=0. However, sometimes, the boundary condition for one expression at this node resolves to a "numerical zero" (2.22E-16). This discrepancy between one displacement BC of 0, and another of 2.22E-16 at the same node leads to an ErrNodeBCRedundantDof error message, obviously (Abaqus is powerful, but it can't split atoms). 

Currently, I create analytical expression fields with python by forming strings of my relevant constants and driving variables. I have tried using python 'round' to limit precision to 5 digits and using the Decimal data type (as suggested in the previous thread), but to no avail. My current workaround consists of manually editing the .inp file after creation, and while that suffices, I was wondering if there was something that I could do differently to prevent having to edit the input file.

Thanks in advance.

Patrick

​​​​​​​q