The calculated fitting parameters and Akaike values for the different
models are provided in Table 1, together with the Akaike criterion
differences. Since a lower Akaike criterion value suggests a
model with higher statistical quality, positive numbers favour
the linear model while negative numbers favour the pore inactivation
model. On average, the linear model describes the experimental
data slightly better, due to the fact that the linear model has
only one fitting parameter. Another factor is that most data points
are in the linear part of the oil uptake vs. water loss, which is best
described by the linear model. Indeed, the linear model seems to
especially fit the 8.5 mm data sets very well, and, as mentioned
before, the maximum evaporation rate is reached sooner with thinner
fries and higher temperature. Since the sampling times are the
same, this also means that more data points are in the linear part of
the graph. However, the pore inactivation model better explains
the initially slow increase of oil uptake and its fitting parameter (K) has physical meaning, which is not so for the Smax in the linear
model.