I am validating a hyperelastic constitutive model implemented in Abaqus against an independent JAX/Python implementation.

Verification status

• Uniaxial tension: JAX and Abaqus match almost perfectly.
• Equibiaxial tension: There is a noticeable mismatch in the middle of the loading curve.

Abaqus model

• Element: C3D8H
• incompressible hyperelastic material
• Square specimen with dimensions 1 × 1 × 0.04
• Symmetric biaxial displacement loading:
  • Left edge: prescribed (U_1)
  • Right edge: prescribed (U_1)
  • Bottom edge: prescribed (U_2)
  • Top edge: prescribed (U_2)
  • Bottom-left corner fixed only in (U_3) to prevent rigid body motion.

Reaction force extraction

I extract the sum of RF1 on the right edge and compare it with the analytical/JAX force. I also extracted the sum of RF2 on the top edge, same mismatch.

My questions

1. Is the sum of RF1 on the right edge the correct reaction force to compare with an analytical equibiaxial constitutive response?
2. Is there any reason why I should instead use another reaction force quantity or combine reactions from multiple boundaries?