Purpose Most previous studies focused on the metal extraction efficiency of chelant-enhanced soil washing under various operational conditions; however, one of the keys to cost-effective field application is to maintain a high throughput rate (i.e., short washing time) while achieving the treatment objectives. Therefore, this study presented a conceptual model for simulating the kinetic extraction of heavy metals and evaluated its sensitivity analysis to the predetermined parameter values in five soils with different initial metal distributions. Materials and methods The proposed conceptual model considered the initial metal distribution and the respective fast and slow first-order extraction rates as the predetermined input parameters. The sensitivity analysis was conducted by combined and individual variation of the parameter values by fixed percentages. An implicit assumption is that there is no interaction between the metal fractions and the parameters are independent of each other. All other parameters were kept constant. Such systematic testing of the model behaviour in reaction to changes in the input parameters helped us identify the parameters that are more sensitive and require particular attention in the modelling process.
Results and discussionIn the sensitivity analysis of combined parameter groups, the fraction of fast extraction ( f ) was linearly and inversely correlated to the required washing time in two stages, in which the correlation was more significant at the lower range of f values. The first-order rate coefficients of fast extraction (k′) and slow extraction (k″) showed a nonlinear and negative relationship with the required washing time. The individual contributions of f i and k i ′ of the three fraction groups were then analysed. Although the fraction group 1 (exchangeable and carbonate) had the highest f i and k i ′ values, their importance was limited by relatively small proportion in the field-contaminated soils. The required washing time for Pb extraction was most sensitive to the fraction group 2 (Fe/Mn oxides).Conclusions The results of this study suggested that model simulations are less sensitive to f 1 and k 1 ′ values of exchangeable and carbonate fractions, which can therefore be considered constant. However, f 2 and k 2 ′ values of the oxide fraction are sensitive parameters and should be prudently justified, or even recalibrated, at different sites.