Rapid prediction of the activated carbon adsorption ratio by a regression model.

Regression analysis of micropore volume and particle size of powdered activated carbon (PAC) is applied to develop a model to predict the adsorption ratio of a non-polar compound, 2-methylisorneol (2-MIB), onto PACs. This model likewise predicts the adsorption ratio of the same PACs and other PACs in background water containing similar natural organic matter (NOM). When this model is used to predict the same PACs adsorption ratios at 30 and 50 mg L-1, the respective percent sample deviations standard error (SDEV) is of 30% SDEV and 12% were obtained. Further, the model is also employed for the prediction of 2-MIB adsorption capacities for 12 different PACs in water with similar NOM at the same dosages, with average SDEV values of 44% and 28%, respectively. Results indicate that 2-MIB adsorption occurrs mainly through the micropore filling mechanism. Nevertheless, when this model is expanded to predict PAC adsorption of NOM with different properties in water, the results exhibited rather large errors. Though this model cannot be applied to waters containing NOM with different properties, it provides information for water utilities themselves or the ones using similar source water to predict the PAC dosage without any adsorption experiment when change of PAC is needed.