Oswin and Halsey Regression models are used in food science to adjust isotherm curves in various applications. However, since they are nonlinear models, one can question its asymptotic proprieties which in turn are related to their capacity to become linear. The purpose of this study conducted via Monte Carlo simulation, considering as independent variable levels of water activity (a w) was on obtaining results from the properties of nonlinearity, by measuring the intrinsic (IN) and extrinsic curvature (PE) . Both models, when considering high values of a w behaved similarly, with promising results regarding the complexity of convergence. The results of the PE showed that in all ranges of a w evaluated the models require a parameterization that can guarantee a behavior closer to that of an equivalent linear model.
Biparametric models; measures of curvature; simulation Monte Carlo