I am currently working on validating experimental results from the Hanchak ballistic impact tests using the Holmquist–Johnson–Cook (HJC) model, implemented in Abaqus via a VUMAT subroutine. At higher impact velocities, the simulated residual velocities remain consistent and show good agreement with experimental data. However, as the impact velocity decreases, I observe minimal variation in residual velocity. According to the reference study, the ballistic limit is reported at 301 m/s, but in my case, I still obtain a residual velocity of 205 m/s, even after refining the mesh to 1 mm.

Moreover, when the total simulation time is increased, the model exhibits rapid and excessive element deletion, which appears unrealistic and significantly affects the results.

Despite reviewing multiple studies that utilize the HJC model, I have not found a clear resolution to this issue. I’m reaching out to see if others have encountered similar challenges and to seek suggestions on corrective measures that could improve simulation accuracy—particularly at lower impact velocities.

A parameter that appears to play a critical role is the Failure Strain (FS) threshold used for element deletion. While damage in the HJC model is primarily governed by D1 and D2, I am unsure why FS is influencing the results so strongly. According to the Abaqus documentation, an FS value of 0.2 is suggested for SiC, whereas other papers have used values as low as 0.004—yet many studies fail to report this parameter altogether.

If anyone can provide guidance on selecting an appropriate FS value or share methods for determining it experimentally or analytically, I would greatly appreciate your insights. Thank you!