Among the numerous moisture sorption models, the GAB, an expanded version of the BET equation originally derived for inert gases adsorption on solid surfaces, is the most prominent. Both models are based on distinction between the adsorbate molecules settling on the bare surface in a monolayer (Langmuir adsorption) and those settling on already absorbed molecules forming multilayers. The BET theory has been successfully used to determine porous catalysts’ and fine powders’ specific surface area to this day. In contrast, applying the BET and GAB equations to water vapor sorption by solid foods (and non-foods) has at least three major problems: The calculated food powders’ specific surface area is independent of their particle size; the expected shoulder in foods’ enthalpy vs. moisture plot is absent; and there is a huge discrepancy between the specific surface areas calculated from moisture sorption isotherms and those of nitrogen. An alternative modeling approach posits the non-existence of a water monolayer, and suggests that moisture sorption is governed by at least two simultaneous mechanisms having different scaling exponents. Mathematical analysis and comparison of the resulting sorption models with the GAB equation show that they produce practically indistinguishable moisture sorption isotherms even with the same number of adjustable parameters. They also demonstrate that the sigmoid shape of a moisture sorption isotherm does not contain enough information to identify and quantify the underlying sorption mechanisms, and that a model’s good fit by statistical criteria in itself does not validate mechanistic assumptions.

在众多的水分吸收模型中，GAB是BET方程的扩展版本，最初是针对惰性气体在固体表面的吸附而衍生的BET方程，是最突出的。两种模型均基于沉降在单层裸表面上的吸附物分子（Langmuir吸附）和沉降在已吸收的分子上而形成多层的分子之间的区别。迄今为止，BET理论已成功用于确定多孔催化剂和细粉的比表面积。相比之下，将BET和GAB方程应用于固体食品（和非食品）对水蒸气的吸附至少有三个主要问题：计算出的食品粉末的比表面积与它们的粒径无关；在食品的焓值与水分关系图上没有预期的肩峰；由水分吸附等温线计算得出的比表面积与氮的表面积之间存在巨大差异。一种替代的建模方法假定不存在水单层，并表明水分的吸收是由至少两种同时发生的，具有不同水垢指数的机制控制的。对所得吸附模型与GAB方程的数学分析和比较表明，即使使用相同数量的可调参数，它们也能产生几乎无法区分的水分吸附等温线。他们还证明了湿气吸附等温线的S型形状没有包含足够的信息来识别和量化潜在的吸附机理，并且模型本身通过统计标准的良好拟合本身并不能验证机理假设。