Biofuels are key products for the sustainability of the world energy consumption in the next years. Biodiesel in particular is a non-toxic, biodegradable, environmentally friendly alternative diesel fuel. Nowadays, the main problem for the commercialization of biodiesel is its final cost, that is strongly dependent (about 85 % of the total) by the feedstock used. A possible way to lower the biodiesel production costs is using raw oils, which contain a higher amount of Free Fatty Acids (FFA) that should be eliminated before the transesterification reaction to avoid soaps formation. In this work the deacidification of raw sunflower oil in both batch and PBR reactor was conducted. Heterogeneous catalyst Amberlyst 46 was used in its dry form and all the experiments were executed using an amount of methanol higher than the stoichiometric but preserving the mono-phasic liquid phase. Different operative conditions, i.e. different methanol/FFA ratios and temperatures in both the reactor configurations, were tested. The robust techniques for kinetic parameters estimation and simultaneous discrimination of rival models were adopted and combined with a dedicated differential-algebraic equation (DAE) model that characterizes the PBR reactor. The main kinetic parameters were regressed using kinetic models other than literature models (a pseudo-homogeneous and an adsorption based one) in considering the components molar fractions (ideal system). The kinetic parameters were obtained using equilibrated resins, i.e. using the catalyst after having let it to adsorb reactants and products at the operative conditions. From the regression results, it could be concluded that the pseudo-homogeneous model is slightly better for fitting the experimental data.