Correcting stress–strain curves from large-deformation V-notched rail shear tests

Abstract

Stress–strain curves obtained from V-notched rail (VNR) shear tests can become increasingly inaccurate at large deformation as the principal directions evolve and the ligament deformation becomes non-uniform. This study combines PEEK VNR experiments at 180 °C, digital image correlation (DIC), simple-shear kinematics and finite element analysis to quantify and correct this error. The comparison shows that the main discrepancy arises from stress rather than strain calculation: the calculated strain remains close to the ligament-centre response, whereas the stress error increases to about 12% at an effective strain of approximately 0.10. The main advance is to correct the full large-deformation VNR stress–strain response before constitutive identification. This extends the correction beyond individual test properties such as modulus or strength. A two-step stress correction is therefore developed. The first step uses an ROI-averaged reference to address the contribution associated with principal-direction change. The second uses a ligament-centre reference to address the remaining effect associated with non-uniform ligament deformation. After correction, the numerical curves closely follow the prescribed material response and the experimental VNR curve shows substantially improved agreement with the tensile reference. The correction also reduces the apparent hardening identified from the VNR response and improves subsequent strain prediction. Additional numerical cases for PEEK at 240 °C and PA6 at 160 °C show a similar reduction in stress error.