Statistical Analysis for Prediction of Distribution Coefficient of Soil-Contaminant System

The soil-contaminant interaction is significantly influenced by various attributes of soil and contaminant, and it can be quantified by employing a term known as distribution coefficient (k(d)). However, because of the involvement of many parameters related to soil and contaminant, determination of k(d) becomes a tedious and time-consuming exercise that requires extensive experimental investigations. In this context, application of the Taguchi method for minimizing the number of experiments and establishing the most critical parameters which influence k(d) (obtained from the linear isotherm) has been shown to be quite useful by the researchers in the recent past. Keeping this point in view, application of L25 orthogonal array (OA) used in the Taguchi method has been employed to design experiments which would facilitate establishment of the influence of various parameters concomitantly on k(d) and its prediction as well. The k(d) has been predicted by employing two methods viz, the Taguchi method and the statistical regression model (SRM). With this in view, four parameters (viz, cation exchange capacity of the soil, ionic species, temperature, and interaction time of the soil-contaminant system) were selected for developing the L25OA, and it has been clearly demonstrated that these methodologies help in easily (1) identifying the most significant parameter on which k(d) is dependent and (2) predicting the precise k(d) by inputting the four parameters. Moreover, based on this study, the sorption mechanism that yields a maximum k(d) for a given soil-contaminant system (SCS) by employing the most optimized parameters has also been proposed. (C) 2015 American Society of Civil Engineers