Microstructural control is a key aspect in producing ceramics with tailored properties and is often achieved by using dopants in a rather empirical fashion. Atomic scale simulations could provide much needed insight but the long-standing challenge of linking simulation results on isolated grain boundaries to those measured in real ceramics needs to be resolved. Here a novel Monte-Carlo simulation method based on a microstructural model in combination with energies obtained from atomic scale energy minimization is presented. This approach allows, for the first time, the prediction of the nominal solubility of dopants (Y, La and Mg) in a ceramic purely from theory. Results compare well with segregation/precipitation data as a function of grain size, found in the literature. The method can therefore be used in developing experimental guidelines for the effective use of dopants in ceramic production, thus accelerating the development of novel materials required for innovative applications.