We have developed a new version of the code built by Campo Bagatin et al. (1994a,b) and Campo Bagatin (1998) to model the collisional evolution of the asteroid size distribution. The new code distinguishes between ``intact'', unfractured asteroids that did not undergo catastrophic collisions, and asteroids converted by energetic collisions into reaccumulated bodies, or ``rubble--piles''. The distinction can also be made on a physical ground, by assigning different collisional parameters to the two kinds of objects, with the object of simulating the different responses to energetic impacts that rubble--piles may have -- due to their different structure -- with respect to unshattered bodies. Rubble--piles abundance turns out to be generally higher when those objects are supposed to transfer less kinetic energy to the fragments than monolithic asteroids. We have run a number of simulations of the collisional evolution process to assess the size range where reaccumulated bodies should be expected to be abundant in the main asteroid belt. We find that this diameter range goes from about 10 to 100 km, but may extend to smaller or larger bodies depending on the prevailing collisional response parameters, such as the strength of the material, the strength scaling law, the anelasticity coefficient of the impacts and the reaccumulation model. In particular, the simulations that take into account the derived ``hydrocode'' scaling laws (Davis et al., 1994) show that nearly 100\% of the main belt asteroids larger than a few kms should be reaccumulated objects. On the other hand the present code shows that the scaling--law recently proposed by Durda et al. (1998) to match the actual population of asteroids, does not fit it if collisional processes are accounted for in a self--consistent way.