Starting from the grain itself, the process of roasting coffee (Coffea arabica) is essential for a quality coffee. Since the movement of water is essential, drying kinetics allows modeling the process and obtain the effective diffusion coefficient, Deff, key parameter to understand the water transport during the roasting process. Two approaches were compared considering that this coefficient is a function of particle temperature, Tp, and low humidity, MR. For the first model, adjustable model parameters were used, while for the second, a Mamdani fuzzy type of model was proposed. Each model was included in the Fick's second law to estimate theoretical values ​​of MR. These values were compared with the experimental ones obtained during roasting coffee in a spouted bed at three different levels of air temperature, TA (275, 300 and 325°C). Even though the adjustable parameters model showed a good fit, the Mamdani fuzzy model was much better (R2 > 0.999 and SEE < 0.001), but plots showed anomalous values ​​for those regionswithout experimental data.

.