Comparative study on retardation behavior of Cs in crushed and intact rocks: two potential repository host rocks in the Taiwan area

Developing a machine learning-based predictive model for cesium sorption distribution coefficient on crushed granite

The sorption of radionuclides on granite has been extensively studied over the past few decades due to its significance in the safety assessment of geological disposal for high-level radioactive waste (HLW). The sorption properties of granite for radionuclides exhibit considerable variability under different experimental conditions. To reduce the time and cost associated with traditional experiments, this study developed a data-driven approach utilizing machine learning (ML) algorithms to predict the sorption distribution coefficients of cesium (Cs) on crushed granite efficiently. Four ML algorithms, namely AdaBoost, GBDT, LightGBM, and XGBoost, were employed to construct predictive models using a dataset of 384 data points. All models demonstrated strong performance, with R2 values exceeding 0.8 for both the training and test sets. Comparative analysis of evaluation metrics indicated that the XGBoost model exhibited the best predictive performance and generalization ability. An explanation analysis of the XGBoost model further revealed the importance and influence of each input feature in predicting the distribution coefficient of Cs on crushed granite. The features affecting radionuclide sorption on granite were ranked by importance as follows: solid/liquid ratio, ion strength, pH, contact time, initial concentration, and maximum particle size. The underlying sorption mechanisms by which different input features affect the sorption coefficient, as derived from shapley additive explanations (SHAP) analysis, correspond with experimental observations. The approach proposed in this study can serve as a supplement to resource-intensive experimental methods, providing new insights into predicting the sorption behavior of radionuclides on crushed granite for the safety assessment of HLW geological disposal.

Enhancing Europium Adsorption Effect of Fe on Several Geological Materials by Applying XANES, EXAFS, and Wavelet Transform Techniques

This study conducted adsorption experiments using Europium (Eu(III)) on geological materials collected from Taiwan. Batch tests on argillite, basalt, granite, and biotite showed that argillite and basalt exhibited strong adsorption reactions with Eu. X-ray diffraction (XRD) analysis also clearly indicated differences before and after adsorption. By combining X-ray absorption near-edge structure (XANES), extended X-ray absorption fine structure (EXAFS), and wavelet transform (WT) analyses, we observed that the Fe2O3 content significantly affects the Eu-Fe distance in the inner-sphere layer during the Eu adsorption process. The wavelet transform analysis for two-dimensional information helps differentiate two distances of Eu-O, which are difficult to analyze, with hydrated outer-sphere Eu-O distances ranging from 2.42 to 2.52 Å and inner-sphere Eu-O distances from 2.27 to 2.32 Å. The EXAFS results for Fe2O3 and SiO2 in argillite and basalt reveal different adsorption mechanisms. Fe2O3 exhibits inner-sphere surface complexation in the order of basalt, argillite, and granite, while SiO2 forms outer-sphere ion exchange with basalt and argillite. Wavelet transform analysis also highlights the differences among these materials.

Scaling effect on cesium diffusion in compacted MX-80 bentonite for buffer materials in HLW repository

In this study, radionuclide behavior in high-level radioactive waste (HLW) disposal repositories is complicated because of the spatial heterogeneity of porous media, coupled flow-transport mechanisms, and multiple chemical reaction processes. Discrepancies in the diffusion behavior of a non-sorbing tracer (HTO) and a reactive tracer (137Cs) in porous media have long been recognized but are not yet fully understood, which hinders effective assessment of the capabilities of buffer materials. This paper was dedicated to exploring and explaining the discrepancies in the transport behavior of non-sorbing and reactive tracers through laboratory experiments and an investigation of contributing mechanisms. Our results showed that for a bentonite sample of the same thickness, 137Cs has smaller apparent and less effective diffusion coefficients than those for HTO. These discrepancies can be attributed to the negative surface electric effects, atomic properties, and chemical reactions. In the case of bentonite samples with different thicknesses (0.5, 0.75, 2.0, 2.5 cm), the apparent and effective diffusion coefficients show an increasing trend with bentonite thickness. According to the experimental data and fitting results, the apparent and effective diffusion coefficients are highly related to bentonite thickness. Thus, scaling effects on transport parameters were proposed to explain the results, which were attributed to the nonuniform distribution of the pore space in the bentonite sample. The scale effect behavior of radionuclide was quantified through a regression analysis. The results can be used to improve buffer designs for radionuclides diffusion.

A sensitive improved method for analyzing diffusion coefficients of Cs in compacted bentonite with different lengths

Based on the one-dimensional diffusion theory, the diffusion parameters were obtained from numerical fitting by the Cyclic Initial Value (CIV) program written by MATLAB. Taking the through-diffusion experimental of cesium (stable isotope 133Cs) as an example, on the premise of ensuring accuracy, fitting calculation was used to obtain the diffusion equilibrium time of Cs in different lengths bentonite column. The fitting results of diffusion test for tritium water (HTO) and Cs are both very well. The calculation results of the equilibrium time for Cs diffusion show that the equilibrium time obtained by CIV is less than the experimental period in both groundwater (GW) and seawater (SW). In GW environment, when the sensitivity is at the maximum setting value, the diffusion coefficient of Cs in 1.5 cm bentonite column could be calculated in a shorter period of time. Compared with the experimental period, the time was shortened by 110 days. The main purpose is to verify the feasibility of CIV through the experimental data of Cs in different column lengths. The CIV program can also be used to fit and calculate the experimental data of other ongoing diffusion experiments and get the equilibrium time required for diffusion experiments. It shows that the proposed models offer the advantages of saving experimental time and reducing experimental waste.

Novel method for analyzing transport parameters in through-diffusion tests

Buffer materials such as bentonite are vital for absorbing radionuclide leakage and retarding migration from radioactive waste canisters. The diffusion coefficient and the retardation factor are the predominant properties controlling the diffusion-reaction process in a buffer material. Diffusion experiments combined with Crank's graphical method are a well-established process for determining asymptotic diffusion coefficients. However, the inaccuracy of the diffusion coefficient that results from the subjective judgement of the late-time linear part of the cumulative concentration data in Crank's graphical method will deteriorate the estimate of the retardation factor. A novel parameter identification process based on an iterative and analytical method (PIPIAM) is proposed here to obtain the diffusion coefficients and porosity of bentonite using concentration data. The results of PIPIAM and the graphical method are compared through an error analysis of concentration. The results show that PIPIAM outperforms the graphical method in terms of the error reduction of the concentration and the uncertainty decrease of the estimated parameters. The proposed method is thus a good alternative for acquiring transport parameters for use in safety assessments of nuclear waste repositories.