this looks quite familiar to me. What you observe can have basically two main reasons:
if you initialize the spheres with a dense packing without any small gap between them, because of the discretization and the used contact law, there might be a numerical instability that causes an “explosion”
if you choose a too large time step (either manual or a too large time step scaling), the collisions can not be computed correctly and you will observe some kind of chaotic behavior
In your case it seems the second reason applied. Perhaps you could switch to the automatic time stepping with a scale factor of 1 see if this works better (just as a trial run that only runs 1-2sec).
Additionally sometimes there are numerical instabilities caused by the belt movement. If this is the reason to the “blast”, try to keep the belt at rest until the filling reached it and later started moving it.
Another hint: if you want to visualize the particles with their correct size, you can use the glyph filter, select as glyph type spheres, radius as 1, scalars as radii, scale mode scalar and scale factor 1.
I hope that helps!