A Computational Fluid Dynamic model of oxy-coal devolatilization experiments in a pilot-scale entrained flow reactor is proposed to gain insight into the thermal histories of the cloud of solid fuel particles and to investigate sources of uncertainties that may affect the analysis of devolatilization. Solid fuel particles experience different paths and are characterized by a temperature which is lower than the reactor nominal one at some of the sampling locations used for investigating devolatilization rates. Indeed, neglecting the quick heating up of the particles by assuming their temperature to be equal to the reactor one for calculating kinetics leads to an underestimation of devolatilization rates. A simple hypothesis on a linear dependence of the particle temperature on residence time through an average heating rate, with residence time and heating rate estimated from the numerical model, was found to largely improve the reliability of devolatilization kinetics even with a simple Single First Order Reaction model. An analysis of uncertainties on the kinetic parameters due to the cloud of particles is performed.