Modelling of precipitation in concentrated binary alloys requires to revisit the fundamentals of precipitation. While no experimental approach can provide information accurate enough for this purpose, Atomic Kinetic Monte Carlo simulations (AKMC) can generate large amounts of data produced in well-controlled conditions. Recently, the extensive analysis of such data has revealed new features about the evolution of clusters during precipitation. In particular, it has been shown that their free energy evolves linearly with the chemical potential of monomers µ1. Their volume is another complex function which evolves with µ1. This quantity can be easily extracted from AKMC simulations: its variable part depends on C1, the concentration of monomers and M1, the concentration of pure matrix sites, neglected in classical approaches. Cluster dynamics (CD) behaves as a bridge between atomistic simulations and phenomenological models. CD can be applied to concentrated alloys provided one can describe the evolution of the free energy and the volume of clusters as functions of µ1. Therefore, modelling of M1 is the last obstacle to build a complete CD scheme, no longer limited by the concentration or the supersaturation of alloys. For this purpose, two methods are examined and to predict the evolution of M1 during precipitation and compared with the results of AKMC simulations. One is derived from a generalisation of the exact solution in 1D while the other extrapolates the results of AKMC simulations.