A numerical model based on computational fluid dynamics is presented to predict the efficiency of a semi-dry flue gas desulfurization process. The Euler-Lagrange particle tracking method is used to predict both evaporation and mass transfer between the flue gas and the lime slurry droplets. The overall desulfurization process is split into two periods, i.e. the constant-rate and the falling-rate period, during which the fundamental chemical reaction between sulfur dioxide and calcium hydroxide is modeled. The absorption efficiency is calculated for different spray characteristics, taking into account different size spectra of the lime slurry droplets characterized by different standard deviations for a constant mass mean diameter d(50) of 25 mu m. The results are compared with published experiments and show excellent agreement as well as a significant role of the spray characteristics in the FGD efficiency.