Uncertainty analysis and properties scaling of kinetics uranium sorption/desorption in contaminated sediment

Rates of U(VI) release from two field-aggregated and contaminated composite sediments were collected from the seasonally saturated lower vadose zone of the US Hanford 300 Area were examined in stirred flow-cell reactors. Kinetic U(VI) desorption which is rate-limited by diffusive mass transfer to and from intragranular sorption sites in subsurface sediments was descried using a multirate surface complexation reaction (SCR) model. In this research, the first sediment was used to access the uncertainty of multirate SCR model and U(VI) desorption under variable chemical conditions. A Bayesian-based, Differential Evolution Markov Chain method was used to assess the parameter uncertainty of the multirate SCR model. The rate constants in the multirate SCR model were estimated with and without assumption of a specified log-normal distribution to test the lognormal assumption typically used to minimize the number of the rate constants in the multirate model. The second sediment and its individual grain size fractions were used to develop a model to scale U(VI) sorption/desorption properties based on sediment grain size distribution. The scaling model assumes that the mean and variance of U(VI) sorption/desorption properties in a sediment including labile U(VI) concentration, sorption site concentration, equilibrium surface complexation reaction constants, and sorption/desorption rate constants can be predicted from the corresponding properties in individual grain size fractions in the sediment. The first sediment results showed that the estimated rate constants without a specified lognormal assumption approximately followed a lognormal distribution, indicating that the lognormal is an effective assumption for the rate constants in the multirate SCR model. However, those rate constants with their corresponding half-lives longer than the experimental durations for model characterization had larger uncertainties and could not be reliably estimated. The uncertainty analysis revealed that the time-scale of the experiments for calibrating the multirate SCR model, the assumption for the rate constant distribution, the geochemical conditions involved in predicting U(VI) desorption, and equilibrium U(VI) speciation reaction constants were the major factors contributing to the extrapolation uncertainties of the multirate SCR model. For the secondary sediment, the scaling U(VI) sorption/desorption properties calculated from scaling model in a composite sediment <2mm are all close to the measurement values and the relative errors are smaller than the measurement errors. Meanwhile, the predicted 95% credible interval of U(VI) concentration using scaling rate constants includes about 78% of the measured U(VI) desorption in <2mm composite sediment. These results indicates the scaling model is an effective method to predict U(VI) desorption in sediment. The developed approach can be used to upscale U(VI) sorption/desorption properties from laboratory to field using field-scale grain size distribution.