ABSTRACT This study develops a physics‑guided machine learning framework to predict sodium (Na+) diffusion in compacted bentonite and to examine whether classical transport relationships can be recovered directly from archival data. A curated subset of Na+ diffusion experiments is extracted from the JAEA Diffusion Database, covering a broad range of dry densities, porosities, and temperatures. The model is evaluated under a Solid+Reference Group K‑fold cross‑validation scheme in which all measurements from a given material-reference combination are kept either in training or in testing, so that interpolation within individual experimental series is clearly separated from cross‑material generalization. Under this leakage‑resistant design, the coefficient of determination (R2) is markedly lower than in standard random K‑fold validation, underscoring the impact of validation strategy on the perceived predictive performance. Despite the absence of explicit constitutive equations in the learning algorithm, the inferred dependence of effective diffusivity on porosity follows an Archie‑type power law with a cementation exponent within reported ranges for compacted clays. Density‑response analysis further shows that the logarithmic slopes of effective and apparent diffusion coefficients with respect to dry density are nearly parallel, while the rock capacity factor remains approximately constant over the examined compaction range, a pattern consistent with a compensatory interlayer‑diffusion pathway that helps maintain surface‑mediated flux as macroporosity decreases. Two specific contributions emerge from this framework. First, the leakage‑resistant Group K‑fold validation reveals that conventional random splitting substantially overestimates predictive skill for unseen bentonite compositions, a finding with direct implications for the reliability of diffusivity surrogates in repository safety assessment. Second, despite the absence of imposed constitutive equations, the model autonomously recovers an Archie‑type porosity scaling and an interlayer‑diffusion signature consistent with a nearly constant rock capacity factor, demonstrating that carefully validated machine learning can function as a hypothesis‑oriented analytical tool for probing transport mechanisms in engineered barrier systems. Together, these results provide a calibrated benchmark for extending physics‑guided machine learning approaches to chemically more complex radionuclides in geological disposal safety assessments.

ABSTRACT Diffusion and sorption of radionuclides in potential host rocks are key processes controlling the safe geological disposal of radioactive waste. To set reliable parameters of diffusion and sorption relevant to safety assessment, it is necessary to understand these processes under laboratory and in situ conditions, and to confirm the consistency or the relationship of these parameters (D e and K d ) between both conditions. This paper focused on comparative experiments and modeling of diffusion and sorption of Cs for both in situ and laboratory conditions in siliceous mudstone in the 350 m Gallery at the Horonobe Underground Research Laboratory (URL) in northern Hokkaido, Japan. The D e values for Cs were consistent between in situ and laboratory results, indicating typical cation excess and porewater salinity effects. The effect of the diffusion direction, parallel or perpendicular to the bedding plane was not significant. The K d values for Cs were also consistent between in situ and laboratory results, indicating the porewater salinity effect. The clay-based modeling approach, which accounts for the major contributions of clay minerals through cation excess diffusion by electrostatic interaction at the clay surface, and the component additive cation exchange reactions, could interpret the diffusion and sorption behaviors of Cs under different salinity conditions and for both in situ and laboratory conditions.

ABSTRACT The structural, optical, fluorescence, dielectric, and electrical changes that PM-355 nuclear track detector films undergo after being exposed to gamma-ray radiation dosages between 30 and 270 kGy are examined in this work. X-ray diffraction examination showed widening, a shift toward higher diffraction angles, and an increase in peak intensity, all of which were indicative of increased structural disorder and crosslinking together with chain scission. Through the weakening of distinctive C–H, C–O–C, and carbonate groups, as well as the creation of novel oxygenated species, FTIR spectra verified radiation-induced degradation. The creation of conjugated structures, color centers, and carbon-rich clusters was linked to a redshift in the absorption edge and a notable decrease in the direct bandgap (4.15 → 3.40 eV) and indirect bandgap (3.60 → 2.60 eV), according to UV – Vis research. Fluorescence analyses confirmed spectrum shifts linked to defect formation and improved non-radiative recombination pathways, as well as a dose-dependent drop in emission intensity (hypochromic impact). Dielectric tests revealed that as the dose rose, the dielectric constant, dielectric loss, and relaxation time increased. These findings were compatible with defect development, interfacial polarization, and increasing dipolar contributions. The associated barrier-hopping mechanism of charge transport was validated by the drop in the frequency exponent and the increase in AC conductivity with both frequency and radiation dose. Under gamma-ray irradiation, PM-355 films generally display variable structural and optoelectronic characteristics, underscoring their promise for sophisticated photonic, sensing, and radiation-dosimetry applications.

ABSTRACT This study aims to reduce waste derived from extractant-impregnated adsorbents by assessing the reusability of spent adsorbents generated during extraction chromatography for the separation of minor actinides (MA(III)) from high-level liquid waste (HLLW). Regeneration was investigated via (i) stripping of the extractant with organic solvents and (ii) re-impregnation of the support. n-dodecane and 1-octanol were examined as candidate solvents. In addition to evaluating stripping performance and the adsorption/elution performance of re-impregnated adsorbents, vacuum distillation was applied to the post-stripping solutions to separate and recover the extractant and solvent. The results demonstrate that the extractant can be stripped from the adsorbent by solvent treatment and that the adsorption and elution performance can be restored by re-impregnation. Furthermore, efficient separation of solvent and extractant from the stripping solution was achieved when n-dodecane was used.

ABSTRACT A large amount of radioactive waste resin will be produced in the production and operation of the nuclear industry, which needs to be treated and disposed safely. The microwave-assisted Fenton reaction is used in this study for the degradation of radioactive waste resin, and the response surface method is adopted to optimize and establish the high-efficiency degradation process. The four-factor (pH value, the dosage of FeSO4 and H2O2 reagents, and the microwave power) and three-level response surface method was used to analyze the influence degree of various factors on the experimental results, screening the key factors affecting the degradation effect, and exploring the best process conditions for the experiment. The research results show that based on the response surface optimization scheme, the optimal process conditions for degrading cationic resin are: pH = 1.637, 41.3 mL of 0.02 mol/L FeSO4, 116.1 mL of (30%) H2O2, 438.31 W of microwave power. Under the optimal condition, the cation exchange resin is rapidly degraded within 1 hour, and the COD of degradation solution is as low as 1.04 g/L. The findings indicate that the microwave-assisted Fenton reaction is an effective method to treat the cation exchange resin, and the response surface optimization method is reliable to the research of cationic resin degradation, which is conducive to design the treatment process of radioactive waste resin.

ABSTRACT This study addresses the challenge of accurately simulating reactor core behavior in the presence of strong heterogeneities using nodal methods. These methods, commonly employed in reactor simulations due to their computational efficiency, rely on the diffusion approximation, which simplifies neutron transport but may introduce errors in heterogeneous configurations. To mitigate these errors, Generalized Equivalence Theory (GET) introduces Assembly Discontinuity Factors (ADFs) that correct for discrepancies between homogeneous and heterogeneous fluxes at node interfaces. Traditional methods of computing ADFs often rely on infinite lattice assumptions, which can be inaccurate near boundaries or in highly heterogeneous cores. To enhance the accuracy of nodal diffusion calculations, this study introduces a full 3D equivalence approach, utilizing the Serpent 2 Monte Carlo code to generate homogenized nuclear data compatible with an NEM-based diffusion solver. A custom subroutine was developed to compute discontinuity factors in all spatial directions, ensuring consistency with the NEM framework. Validation through several benchmark models demonstrated that the incorporation of ADFs significantly improves the estimation of axial power distributions, particularly near boundaries where flux discontinuities are most pronounced. Furthermore, the multiplication factor showed strong agreement with reference Serpent calculations, with notable improvements when DFs were applied. These results highlight the value of the proposed approach in enhancing the fidelity of nodal simulations, especially for complex, highly heterogeneous reactor systems.

ABSTRACT This study investigates the removal of low-concentration iodine vapor from air using aqueous sodium hydroxide ( NaOH ) in a laboratory-scale packed column, employing both experimental and modelling approaches. Experiments were conducted to determine the overall volumetric mass transfer coefficient ( K G a eff ) and assess the influence of gas flow rate, liquid flow rate, NaOH concentration, and packing type. Results showed that K G a eff and the effective interfacial area were strongly dependent on the gas-phase Reynolds number ( R e G ), while liquid-phase parameters had minimal impact under most conditions. However, at high gas Reynolds numbers and low NaOH concentrations, the liquid flow rate progressively affected K G a eff , indicating a transition in the controlling mass transfer regime. A correlation for K G a eff was developed and integrated into a one-dimensional finite difference solver to simulate the absorption process. The predicted outlet iodine concentrations agreed with experimental results, with deviations within 10%. Phase resistance analysis revealed a shift from gas-phase-controlled to mixed gas-liquid controlled mass transfer at R e G ≥ 25 and [ OH ] ≤ 0.25 M. Additionally, porosity was found to inversely affect K G a eff , underscoring the importance of selecting packing materials that minimize gas-phase resistance. These findings highlight the critical role of gas-phase hydrodynamics and hydroxide concentration in optimizing iodine removal. The insights gained are valuable for the design and operation of caustic scrubbers intended to control radioactive iodine emissions in nuclear and radiochemical facilities.

ABSTRACT Image restoration technology has made significant progress in the fields of signal processing, computer vision, and remote sensing. However, the application of Unmanned Aerial Vehicle-borne gamma-ray spectrometry data processing is relatively limited. The aim of this study is to fully utilize the wide applicability of image restoration technology and explore its potential application in the processing of UAV-borne gamma-ray spectrum data. By using methods such as median filtering and edge influence compensation, targeted noise in UAV-borne gamma-ray spectrum data is processed, and data fusion is achieved through image restoration technology, significantly improving the accuracy and visualization effect of data reconstruction. The experimental results show that in practical applications, by fitting the mean vector of the original and restored images, the element content or count rate values can be successfully solved, and the generated contour map is basically consistent with the map trend. The useful information in the restored image is significantly increased, while the average variation value of interference information decreased by 41.47%.

ABSTRACT A fast burnup calculation method based on neutron spectrum reconstruction is proposed. The method employs a reduced-order model (ROM), constructed using proper orthogonal decomposition (POD) and regression models, to estimate neutron spectra experienced by fuel during burnup. The ROM is built from snapshot data generated through detailed burnup and neutron transport simulations under various conditions. During burnup calculations, the ROM is used to rapidly reconstruct neutron spectra at each burnup step. These reconstructed spectra are then used to compute one-group cross sections from multi-group effective cross sections derived using background cross sections. The proposed method significantly reduces computational time by avoiding repeated neutron transport simulations. Its performance is demonstrated using a PWR UO2 fuel pin model. Results show that, with the 6th-order POD, the method predicts nuclide inventories with an average error within ± 5% compared to reference Monte Carlo calculations. Error analysis indicates that prediction accuracy is primarily limited by the regression models, rather than by the POD truncation or the multi-group cross section calculations.

ABSTRACT The 129I neutron capture cross-section was measured at the Accurate Neutron-Nucleus Reaction Measurement Instrument (ANNRI) beamline in the Materials and Life Science (MLF) facility of the Japan Proton Accelerator Research Complex (J-PARC). To obtain accurate results in the keV neutron region, the recently developed neutron filtering technique was employed with a 20-cm-thick nat Si filter to tailor the incident neutron flux into quasi-monoenergetic neutron peaks with averaged neutron energies of 51.5 and 127.7 keV and, at the same time, hugely increase the signal-to-noise ratio. The present cross-section results were determined from the capture yield ratio of 129I to that of 197Au and using the evaluated nuclear data from JENDL-5 as a reference. The 129I neutron capture cross-section was found to be 0.278 ± 0.034 b at 51.5 keV and 0.141 ± 0.015 b at 127.7 keV, validating the JENDL-5 and JEFF-3.3 evaluations, as well as the results from Noguere et al. in contrast to the larger evaluations of JENDL-4.0 and ENDF/B-VIII.1 and the data from Macklin et al.