In order to investigate basic properties of galaxies, such as the star formation rate and the masses of baryonic components, it is important to account for dust reprocessing. Dust particles absorb and scatter the stars' optical/UV emission, and they re-radiate thermally in the infrared. A combination of simulations and post-processing radiative transfer computations can produce mock data, which can be compared directly to observations. Until now, however, dust properties have only been included in our simulations by means of post-processing assumptions, leaving room for uncertainties, particularly significant at wavelengths shorter than 100 microns. To reduce these uncertainties, we implemented a state-of-the-art treatment of the production and evolution of dust grains within our simulation code, P-GADGET3. This model traces the creation, evolution, and destruction of dust through various processes. It accounts for the diameter of dust particles with a two-grain-size approximation proposed by H. Hirashita. We will present a first result of our new code applied to zoom-in simulations of massive ($M_{200} > 3 \times 10^4 M_{\odot}$) galaxy clusters, focusing in particular to the early stages of assembly of the cluster at high redshift, around $z = 2$, where the SF activity is at its maximum and the proto-cluster regions are rich of cold, dust-polluted gas. This publication has received funding from the European Union's Horizon 2020 research and innovation programme under grant agreement No 730562 [RadioNet]