Due to the high investment cost of the wet scrubbing, the dry desulphurisation process remains an attractive way to remove sulphur dioxide when reduction level should not be higher than 90%. In this technique, the SO2 is removed from flue gases by direct injection of small calcium based sorbents such as calcium carbonate (CaCO3) or calcium hydroxide (Ca(OH)2). The sorbent is first transformed into calcium oxide (CaO) before reacting with SO2 to form a stable calcium sulphate (CaSO4). Although the global reactions are rather simple, a full description of the mechanism requires the knowledge of both, the initial physical properties of the sorbent and the evolution of these properties during the reaction. Furthermore, the way how particles are injected and dispersed into the SO2 containing gases is also a major issue of the dry desulphurisation process. Both aspects are considered in this paper since a specialised routine was build to describe the calcination-sulphation process and coupled with the KIVA code to perform 30 calculations under real industrial conditions. The desulphurisation model was first tested alone using the kinetic data of the International Flame Research Foundation and then, coupled with KIVA to simulate the desulphurisation process in a new type of boiler which is developed by BABCOCK Entreprise and the Institut Francais du Petrole (the AUDE boiler). In this last case, the objective was to determine the sorbent injection locations leading to the highest desulphurisation efficiencies. As the dispersion of particles is an important parameter of the desulphurisation process, an example of validation of the KIVA dispersion model is also presented in this paper. For the selected test case, i.e: particle dispersion in a plane mixing layer, a good agreement was found between the experimental results of Ando et al (1990) and the calculations performed with KIVA.