In CFD simulation of CFB combustion, computation of the local instantaneous conversion rate of a single fuel particle should be fast. Since fuel conversion is limited by mass and heat transfer and the rates of chemical reactions inside the particle, the optimal model would produce the internal conditions as a function of location. As rigorous 1D, 2D and 3D fuel conversion models presented in the literature are often too time-consuming for CFD environment, simplified 0D approaches have been suggested. To reduce the amount of simplification, the present study introduces an alternative approach that converts a rigorous fuel conversion model into correlations that can be implemented in a CFD code. As a demonstration example, a 1D shrinking particle model for coal combustion is converted into a correlation based description. The single particle model is used to produce a data set of net reaction rates averaged over particle volume in a wide range of fluidization, temperature and composition conditions. The data is then used to train neural network models that are fast to compute and easy to implement in a CFD code.