The problems is that since the templates for each of the monomers have
the same coordinates (modelling each of the monomers of my dimer) i do
obtain a model of my dimer with the 2 monomer chains basically
inter-twisted (so there is no any kind of interface in between the
chains because the chains are nested one in another).
Modeller only uses internal coordinates for the restraints, so is
independent of the relative orientations of your templates, but your
initial model (.ini file) is probably bad to start with. One solution
would be to manually translate one monomer in the .ini file so they
don't overlap, and then provide that as your starting model.
Also i can’t understand why, but one of the chains has secondary
structure while the other doesn’t.
This sounds like a problem with your alignment (most likely the "bad"
chain isn't aligned with any template), but I'd need to see the real
alignment file to be sure.