Diverged when the loading of particle is changed

Hi, everyone

There is a confusing problem. As shown below, the only difference between two cases is the loading of particles in the initial conditions. For the case where the internal is half submerged in particle phase at the beginning, the run was diverged after a short time. For another case, it ran well and the internal is also immersed in particle phase when particles were fully fluidized. I checked the residuals of last several time step and found that pressure calculation is hard to converged. The source file is also attached. If you know the solution to this problem, please tell me. Thank you!

source_file.zip (11.2 KB)