PurposeTo describe water absorption by the rice grains over time, diffusion and empirical models were used. Also, an optimization software was developed in this study to determine parameters and their uncertainties for the diffusion models (LS Optimizer, for partial differential equations). Parameters (and their uncertainties) for empirical models were determined by LAB Fit Curve Fitting Software.Design/methodology/approachHeat and mass diffusion phenomena are found in various processes of technological interest, including pasteurization, drying and water immersion of agricultural products, among others. The objective of this work was to study the process of water absorption by rice grains with and without husk, using diffusion and empirical models to describe the absorption kinetics. Rice grains were immersed (approximately 10 g for each experiment) in drinking water maintained at constant temperatures of 28, 40 and 50 C. In the experiments, the water contents absorbed by rice grains over time were obtained by the gravimetric method.FindingsAmong empirical models, Peleg was the most satisfactory to describe the kinetics of water absorption by rice without husk, while the Silva et alii model had the best statistical indicators for rice with husk. It was also verified that a diffusion model with boundary condition of the first kind showed the best (or equivalent) results in the description of all processes of kinetics of water absorption by rice grains, with and without husk. For grains without husk, the effective mass diffusivities were (1.186 ± 0.045) × 10−9, (1.312 ± 0.024) × 10−9 and (2.133 ± 0.028) × 10−9 m2 min−1, for the immersion temperatures of 28, 40 and 50C, respectively. For grains with husk, diffusivities were (0.675 ± 0.011) × 10−9 and (1.269 ± 0.017) × 10−9 m2 min−1, for temperatures of 28 and 50 C, respectively.Originality/valueThis work developed a solver for the diffusion equation in cylindrical geometry and presented the LS Optimizer software developed to determine differential equation parameters through experimental data sets.