X-ray imaging using photon-counting detectors (PCDs) allows for the calculation of quantitative images such as the effective atomic number (Zeff). However, physical phenomena such as characteristic X-ray emission and charge sharing can cause incomplete total absorption events during signal generation, which reduces the accuracy of X-ray penetration analysis. To solve the problem, an anti-coincidence mode (ACM) has been developed, but complete correction has not been established. We aimed to investigate the accuracy of Zeff image when the proposed software-based corrections, namely beam hardening and response function corrections, are added to the hardware-based correction of the ACM. The response function was calculated using the Monte-Carlo simulation code. In a simulation study, we analyzed a phantom composed of virtual materials with Zeff values of 4–16. While sufficient accuracy cannot be achieved by applying only the ACM, it was demonstrated that low-noise Zeff images can be obtained by applying our correction. Furthermore, it was demonstrated that Zeff images of food samples can be generated by using actual non-destructive testing equipment with a 10 m/min transportation speed. In conclusion, our correction procedure can maximize the performance of PCDs, and our findings are essential for promoting imaging techniques concerning quantitative images.
The increasing deployment of nuclear technologies necessitates the rapid and reliable localization of lost or unauthorized radioactive sources to ensure nuclear safety. However, existing robotic systems often struggle to adapt to unfamiliar and complex environments. This work introduces SACCL, a reinforcement learning framework that integrates Soft Actor-Critic with Curriculum Learning to enhance source localization stability and transferability in complex radiation environments. SACCL employs a selective state-encoding scheme to minimize computational complexity and combines curriculum-based task partitioning with prioritized experience replay to improve learning efficiency and generalization performance. Furthermore, a radiation-aware adaptive speed mechanism is introduced, which accelerates exploration in low-intensity regions while enabling precise localization near high-intensity sources. SACCL is evaluated in obstacle-free radiation fields, environments with fixed obstacles, and scenarios featuring unknown obstacles. In obstacle-free settings, SACCL achieves a 100% success rate with an average localization error of 0.17 m. In more challenging contexts with fixed and unknown obstacles, success rates of 98% and 93% are attained, with corresponding localization errors of 0.38 m and 0.69 m. Compared to baselines including SAC, SAC_CL, SAC_LSTM, TD3, PF-PPO, and Infotaxis, SACCL demonstrates the best overall balance of success rate, localization accuracy, search efficiency, and robustness across the evaluated radiation environments.
Aiming to balance efficient Californium-252 (Cf-252) production with reactor safety, this study investigates the Ultra-high Flux Lead Bismuth Reactor (UFLBR) and develops six progressive core schemes: no target, targets only, targets with moderator, layered target-moderator arrangement, spectrum-optimized layered arrangement, and spectrum-optimized arrangement based on neutron unidirectional flow. A neutronics-thermal-hydraulics-structural mechanics coupling framework is used to evaluate Cf-252 yield, heavy-nuclide conversion, temperature distribution, stress, and deformation. Results indicate that targets alone produce negligible Cf-252 in the fast spectrum, whereas moderation and layered spectrum regulation greatly enhance production. The spectrum-optimized layered scheme achieves the highest conversion rate of 1.67%, but also induces severe local power and thermal-mechanical risks. By introducing neutron unidirectional flow, thermal neutron backflow is effectively suppressed while favorable target-region spectra are retained. This scheme yields 11.7 g Cf-252 per fuel cycle with a 1.22% conversion rate, about 3.94 times that of the HFIR C78 campaign. Meanwhile, maximum fuel, cladding, coolant, and target temperatures remain below safety limits, and deformation stays within engineering margins. The proposed iterative optimization strategy provides a general reference for isotope-production scheme design and feasibility assessment.
The uncertainty analysis results of key nuclear design parameters for the VHTR-350 core were presented in this work. The uncertainty analysis code system applies a two-step approach, integrating the Generalized Perturbation Theory (GPT) and the random sampling method. In the lattice calculation step, DeCART2D_HTR was employed to generate block-homogenized few-group cross-sections and their corresponding generalized adjoint fluxes. Subsequently, MUSAD quantified the uncertainties for these few-group cross-sections and produced randomly sampled cross-section sets. In the core analysis stage, CAPP performed the 3-D core diffusion analysis using the sampled data sets. The results indicate that the uncertainties for the multiplication factor were 0.547% for the fresh 3-D core and 0.595% for the depleted state. Notably, the uncertainty of the isothermal temperature coefficient exhibited a significant increase with burnup, rising from 4.8% to 16.5%. The uncertainties for the 3-D power peaking factor were evaluated as 0.32% at fresh state and 0.58% at depleted state, while the maximum uncertainty in the radial power distribution increased from 0.53% to 0.63% during the depletion process.
This paper reports on the effect of gamma-ray irradiation on single-mode optical fibers, including measurement of the relaxation process. The study was performed using gamma-irradiation facilities with 60Co sources at doses of 67.9 kGy and 170.0 kGy, with irradiation rates of 0.97 kGy/h and 1.05 kGy/h, respectively. We measured changes in optical losses for seven types of optical fibers, including those with germanium-doped silica core and pure silica glass cladding, as well as four optical fibers with pure silica glass and fluorine-doped silica cladding. Measurements of optical losses and radiation-induced attenuation were performed at four wavelengths: 1310 nm, 1490 nm, 1550 nm, and 1625 nm. The gamma irradiation of the fibers and optical loss measurements were conducted at room temperature, and our measurements showed that in harsh environments where gamma rays can irradiate optical fibers, a wavelength of 1310 nm (O-band, 1260-1360 nm) is better suited for data communications than 1550 nm (C-band, 1530-1565 nm).
Aiming at the multivariable coupling and variable operating condition difficulties in the once-through steam generator (OTSG) feedwater system, this paper investigates system modeling and decoupler design. Firstly, based on the similarity law of feedwater pumps, the flow characteristics of feedwater regulating valves, and the pipeline momentum equation, a nonlinear simulation model of the feedwater system is established. The transfer function model is then derived via linearization, and the accuracy of the linear model is verified through simulation. Secondly, the relative gain array (RGA) method is applied to quantify the coupling degree among system variables. If the system exhibits strong coupling, a feedforward compensation decoupler is designed to eliminate the interaction between variables by introducing compensation elements. Meanwhile, a pre-compensation matrix is constructed based on the inverse Nyquist array (INA) method. Furthermore, this paper proposes a weighting strategy based on membership functions, which realizes smooth interpolation of full-condition decoupling matrices relying on limited typical operating points and overcomes the low engineering operability of traditional single-condition decoupling schemes. Simulation results verify that the designed decouplers can effectively weaken the coupling effects among variables, which is beneficial to the design of subsequent controllers.
In October 2023 an underground chemical explosive test, Physics Experiment 1-A, was executed in a tunnel at the Nevada National Security Site. This test was part of a series of multi-physics experiments focused on enhancing nuclear explosion monitoring techniques. Part of the experiment included monitoring ga movement. As part of the facility operations, there were two events during the first five days of the experiment where the ventilation system was operated to purge gases from the facility. These events afforded an opportunity to evaluate the mass of gases emitted to the atmosphere during these ventilation events and compare to the total mass produced during the experiment. This provides a basis for estimating the fraction of gas produced in an underground test that might be expected to be released through natural and engineered pathways. Using several approaches, an estimate of between 1 and 9% of the total gas produced by the experiment was captured and discharged to the atmosphere during these events. The percentage of gas captured was nominally equal to the percentage of the plume volume that was bisected by the facility tunnels and represents a similar fraction to that assumed as potential releases for underground nuclear tests.
Due to increasing concerns regarding the toxicity, high density, and environmental limitations of lead-based shielding materials, heavy rare-earth elements such as lutetium (Lu) are gaining attention as potential alternatives for radiation protection applications. In this study, the interaction of 10 MeV alpha particles with pure lutetium and two engineered lutetium-based alloys (Lu–B and Lu–W) was investigated using SRIM simulation. Key shielding parameters, including electronic and nuclear stopping powers, projected range, longitudinal straggle, and lattice displacement damage, were evaluated and compared with those of pure lead (Pb). The Lu–W alloy exhibited a projected range of 36.1 μm, slightly lower than that of Pb (36.5 μm), indicating superior stopping efficiency against 10 MeV alpha particles. In contrast, the Lu–B alloy achieved approximately 21.5% weight reduction while maintaining acceptable shielding performance. Radiation damage analysis further revealed that Lu–W produced the highest vacancy concentration due to enhanced collision cascades, whereas Lu–B showed comparatively reduced defect formation, suggesting improved radiation tolerance. Despite its lower atomic number compared with lead, lutetium demonstrated excellent attenuation behavior and structural stability, highlighting its potential as a lightweight and environmentally safer shielding material. These findings suggest that lutetium-based composites are promising candidates for advanced aerospace systems and next-generation nuclear shielding technologies.
To establish direct experimental evidence for vibration fatigue assessment of 316L stainless-steel small-bore piping butt-welded joints in nuclear power plants, full-scale TIG/GTAW butt-weld specimens (Φ16 mm × 3 mm) were tested using an electrodynamic shaker resonance excitation system. A total of 39 specimens were subjected to uniaxial constant-amplitude vibration fatigue tests at six nominal stress levels (130–210 MPa) with approximately fully reversed loading (R ≈ −1) and first-mode resonance frequencies of 55–83 Hz. Digital image correlation (DIC) was employed to measure the weld-toe strain field, X-ray diffraction (XRD) was used to determine residual stresses, scanning electron microscopy (SEM) was performed for fracture-surface analysis, and equivalent structural stress (ESS) was calculated by extracting nodal forces from ANSYS. The influence mechanism of weld reinforcement height on local mechanical states and fatigue life was systematically investigated. Specimens were classified into three groups according to weld reinforcement height: high reinforcement (h ≥ 1.0 mm), medium reinforcement (0.5 mm < h < 1.0 mm), and low reinforcement (h ≤ 0.5 mm). SEM observations on one failed specimen tested at 210 MPa indicated that crack propagation accounted for only ∼0.8% of the total life, supporting the approximation that leak-detection life was initiation-dominated. DIC measurements yielded weld-toe stress concentration factors (SCFs) of 1.26–1.47, which increased monotonically with reinforcement height. XRD results showed that reducing reinforcement height generally corresponds to a larger magnitude of compressive residual stress on the outer surface. Finite-element calculations gave bending ratios r ranging from 0.453 to 0.508 (mean 0.475); the mean r values for the high-, medium-, and low-reinforcement groups were 0.498, 0.472, and 0.458, respectively, indicating that reinforcement height affects fatigue performance through three coupled pathways: SCF, the load mode at the control section, and residual stress. Based on ESS S–N regression using 30 failed specimens, the fitted parameters were m = 7.88, A = 25.91, SlogN = 0.331, and R2 = 0.618; the P97.7% engineering design-line intercept was Ad = 25.25. A supplementary right-censored maximum-likelihood analysis showed that, at Nf = 106, the mean-line ΔSs increased by only 4.2% compared with the baseline result, indicating local stability of the fitted curve within the observed life range. Compared with the ASME VIII Div. 2 Annex 3.F master S–N curve, the two mean lines intersect at ΔSs ≈ 417.4 MPa and Nf ≈ 1.80 × 105. At Nf = 106 and 107, the present mean line is higher than the ASME mean line by 39.2% and 116.9%, respectively, suggesting that the ASME master curve is conservative in the medium-to-high-cycle vibration fatigue regime covered herein. The scenario-specific ESS S–N design line developed in this work provides direct experimental support for vibration fatigue evaluation of 316L small-bore piping butt-welded joints under comparable material compositions, pipe dimensions, and resonance-bending loading conditions.
This study evaluates the effect of LiOH and KOH as pH-control agents on the corrosion behavior of 304 stainless steel in simulated pressurized water reactor primary water chemistry. Corrosion tests were performed in LiOH–H3BO3 and KOH–H3BO3 solutions at 593 K, and the exposed specimens were characterized in terms of corrosion rate, oxide morphology, phase constitution, and surface chemical states. In addition, the wettability and electrochemical behavior of 304 stainless steel were evaluated by contact angle measurement and potentiodynamic polarization tests. 304 stainless steel showed a slightly higher initial corrosion rate in the LiOH–H3BO3 solution than in the KOH–H3BO3 solution. However, the difference between the two solutions became smaller with increasing exposure time. Duplex oxide layers consisting of Fe-rich oxide particles and a Cr-rich inner oxide layer were formed in both solutions. No clear difference was observed in the overall oxide characteristics after long-term exposure. These results indicate that the type of pH-control agent slightly affected the initial corrosion behavior of 304 stainless steel, whereas its influence on the long-term oxide layer characteristics was minor under the present conditions.
Motion blur in X-ray radiography can obscure fine anatomical structures and lead to repeat imaging. This study presents a dual-acquisition residual-domain framework that combines a normal-dose motion-blurred image with an auxiliary low-dose fast-acquisition image for acquisition-specific blur-kernel estimation and restoration. The estimated kernel and residual-domain input are used for residual-domain Richardson–Lucy (RL) restoration, followed by gain-controlled restoration and detail recovery. The method was evaluated through simulation studies using line-pair and ex vivo human dental phantoms and controlled C-arm experiments using an anthropomorphic chest phantom. Across 20 repeated ROI-based evaluations, dental phantom SSIM increased from 0.87 ± 0.02 to 0.94 ± 0.02 and PSNR from 33.81 ± 0.22 to 35.70 ± 0.26 dB, with higher SSIM and PSNR than Wiener, non-blind RL, and total variation–regularized RL deconvolution. In C-arm experiments, BRISQUE decreased from 19.40 to 9.51 in the upper-spine case and from 31.03 to 19.98 in the lower-spine case, while SNR remained near normal-dose levels. CNR remained comparable to the normal-dose level in the upper-spine case and increased in the lower-spine case. The framework improved visibility under small-amplitude, approximately linear motion blur, although sequential acquisition introduces registration and additional-exposure considerations.
For the leakage detection and localization in the vacuum chambers of future fusion reactors, this paper proposes a design method based on the coupling of remote operation and laser spectroscopy. By using a pulsed laser to ablate solid targets, the design excites leaked gases in the environment and generates a spectrum. Spectrometers and detectors are used to collect information on atomic spectral lines in the area, enabling leak detection. A remote-operated robot is then used to transport the spectrometer to different areas of the vacuum chamber for continuous sampling. The precise location of the leak can be determined based on the distribution trends of the leak gas concentration gradient. To fully validate the feasibility of leak location, this paper analysis the diffusion behaviour of the leaking gas under different gas types and background pressures using finite element software. The results show that the mass fraction distribution of the leaking gas within the vacuum chamber exhibits a monotonically decreasing spatial gradient along the primary direction of diffusion. This gradient exhibits a distinguishable decay pattern under various leakage gas conditions (air and water vapor) and at different background pressures, demonstrating the feasibility of using concentration gradients to locate leaks.
This study investigates the neutronic performance, depletion behavior, and safety characteristics of a 5 MWth Sejong TRISO-fueled molten salt filled heat pipe cooled reactor. It employs tristructural-isotropic (TRISO) uranium-oxycarbide fuel, LiF-BeF2 (FLiBe) molten salt coolant, and a beryllium-oxide (BeO) reflector. The Monte Carlo code MCS is used for three-dimensional burnup analysis over a 5 year cycle. The core adopts a pebble bed with 21,600 TRISO-embedded pebbles and 665 sodium heat pipes, achieving passive heat removal and a thermal spectrum that enhances neutron economy and control drum effectiveness. Depletion under all-control-drums-out shows smooth reactivity decrease from k-eff starting at 1.06075 at beginning-of-cycle and reaching 1.01138 at end-of-cycle. This is accompanied by limited spectral hardening and flattened radial power distributions, indicating benign fuel utilization and power peaking. Parametric studies clarify the roles of FLiBe, BeO, and carbon in shaping the spectrum, demonstrating pronounced BeO and carbon moderation and F-19 resonance absorption effects on the intermediate energy flux. Safety analyses of drum rotation, drum seizure, and shutdown rod insertion failure confirm sufficient control margin and fault tolerance. The combination of passive heat pipes, integrity fuel and coolant, and redundant shutdown systems supports a safety-driven microreactor concept.
Li4SiO4 is a promising solid-state tritium breeder material due to its high lithium density and low tritium retention. Under irradiation, lithium vacancies are the predominant defects formed, influencing tritium release behavior. This study systematically investigates tritium adsorption and diffusion in different lithium vacancy configurations within Li4SiO4 using first-principles calculations. The investigation covers three stages: first, adsorption and migration characteristics within vacancies; second, adsorption sites and migration behaviors after escape; finally, diffusion coefficients for both processes. The results show maximum adsorption energies of −1.597 eV (within vacancies) and −1.638 eV (after escape). The minimum migration energy barriers within the three vacancy configurations are 0.231 eV, 0.129 eV, and 0.091 eV, with corresponding optimal escape barriers of 0.236 eV, 0.373 eV, and 0.766 eV. The optimal migration pathways were identified as TⅣ→TⅢ→O37, PⅤ→PⅢ→PⅡ→O19, and OⅤ→OⅣ→O30. At 900 K, diffusion coefficients are 1.63 × 10−7 m2/s (within vacancies) and 2.96 × 10−8 m2/s (escape), with the former being approximately one order of magnitude higher, indicating greater mobility within the vacancies. The results provide an atomistic understanding of the individual stages of tritium migration and can indirectly inform subsequent breeder material design and fabrication.
This study provides empirical evidence on community acceptance of small modular reactor (SMR) siting among urban residents using a double-bounded dichotomous choice contingent valuation survey of 2,456 respondents in South Korea. Among respondents with a finite reservation price, the median annual Willingness-to-Accept (WTA) was KRW 111,000 per person (95% bootstrap CI: KRW 99,000 to 124,000). Respondents were classified into three tiers according to their reactions to two sequential bids: bid acceptors (52.6%), price-conditional acceptors (25.4%), and non-compensatory rejectors (22.0%), who reject SMR siting regardless of the amount offered. Latent class analysis identified five policy-relevant groups based on their perception profiles. The classes differed substantially in their tier compositions and perceptions of risk, benefits, trust, fairness, and negative affect, whereas demographic characteristics had limited explanatory power. The findings show that community responses to urban SMR siting cannot be adequately represented by a simple binary distinction between acceptance and opposition. By integrating compensation-based valuation, behavioral response classification, and perception-based segmentation, this study provides evidence for differentiated community engagement and benefit-sharing strategies that extend beyond uniform policy approaches.
The lubrication of charging pump is critical in the RCV system in nuclear power plants and keeping the oil level stable is essential for safe plant operation. In this study, the reasons for the low-level alarm activation and recovery in the RCV system are investigated by the combination of both theoretical analysis and experimental testing. The results show that two flow characteristics cause fluctuations of liquid level, the maximum amplitudes respectively reach 0.92 mm and 5.73 mm for the initial liquid level of 290 mm. And temperature is a vital factor to the level of the tank, one is ambient temperature which causes lubricating oil to undergo thermal expansion and contraction, changing the level up to 2.6 mm with temperature difference of 10 °C. The other is oil temperature, lower temperatures result in higher lubricant viscosity at the beginning of charging pump startup, prolonging oil return time, which causes level to remain low for an extended period and the maximum level difference up to 3.24 mm between 15 and 25 °C for a D/4 back flow, increasing the likelihood of triggering the low-level alarm. As temperature rises, the liquid is more quickly back to the tank, making the level recovered.
Packed spherical pebble beds are widely used in several engineering applications, including Generation-IV Pebble Bed Reactors (PBRs), where the randomly packed core geometry produces highly complex flow structures within the inter-sphere void regions. Accurate characterization of turbulence structure and coherent flow dynamics in such configurations remains limited due to experimental accessibility constraints. The present experimental investigation examines the spatial correlation characteristics, spectral behavior of velocity fluctuations, and vortex structure dynamics in a randomly packed pebble bed using high-resolution Time-Resolved Particle Image Velocimetry (TR-PIV) measurements. A Matched Index of Refraction (MIR) facility enabled non-intrusive optical access to the void regions between spheres, allowing detailed measurements of turbulence structure within the packed geometry. Experiments were performed under isothermal conditions at two modified Reynolds numbers (Re1 = 306 and Re2 = 1473), representing transition and fully turbulent flow regimes. Two-point spatial cross-correlation analysis of spanwise and streamwise velocity fluctuations at representative near-wall and center-region locations quantified the spatial extents of coherent turbulence structures across the measurement domain. Persistence power spectral density analysis further identified the dominant frequency content of velocity fluctuations and revealed the spectral distribution of turbulent interactions within the inter-sphere flow regions. Vortex identification analysis was performed to characterize the spatial occurrence, circulation strength, and size distribution of coherent vortex structures in both near-wall and central regions of the randomly packed bed. The results provide detailed insight into the evolution of turbulence structures and coherent flow dynamics in randomly packed pebble beds and establish a high-fidelity experimental dataset for validation of computational fluid dynamics models relevant to thermal-hydraulic analysis and performance assessment of pebble bed reactor core geometries.
The integration of Solid Oxide Electrolysis Cell (SOEC) technology and nuclear power systems is investigated in this study, with a focus on operational flexibility and efficiency during peak shaving periods. As renewable energy increasingly penetrate the power grid, the proposed Nuclear Power Plant-Solid Oxide Electrolysis Cell (NPP-SOEC) system addresses the need for flexible baseload operations by repurposing surplus electricity and thermal energy into hydrogen. The system is modeled and simulated under varying load conditions using Aspen Plus software, establishing two distinct thermal integration strategies. Findings reveal a profound thermodynamic contrast between the two configurations. While the Low-Grade Recovery Configuration maintains a slightly higher first-law thermal efficiency reaching up to approximately 80%, the High-Pressure Extraction Configuration demonstrates higher second-law performance. Specifically, the high-pressure strategy achieves a peak exergy efficiency of approximately 75% at a 50% generation load, compared to approximately 67% attained by the low-grade alternative. This highlights the necessity of matching high-grade thermal energy with the sensible heating requirements of the electrolysis process to minimize exergy destruction. Under these optimal conditions, the high-pressure configuration exhibits a hydrogen production capacity of 2711 kmol/h. Comprehensive techno-economic evaluations incorporating operation, maintenance, and stack replacement costs indicate financial viability, with the return on investment reaching up to approximately 340% over a ten-year lifespan. Furthermore, while continuous full-power reactor operation mitigates thermal-hydraulic fatigue risks, the safety assessment integrates quantitative regulatory constraints. It emphasizes that enforcing a safe stand-off distance against hydrogen detonation hazards, such as the 1 psi overpressure threshold, constitutes a primary source of economic uncertainty. Overall, the integration of SOEC technology with nuclear power offers a systematic framework for evaluating nuclear-SOEC coupling configurations, while acknowledging that site-specific factors—including electricity market structures, hydrogen storage economics, and seasonal demand variations—will determine the practical viability of deployment.
This paper presents an equivalent source response matrix method (ESRMM) for solving the neutron diffusion equation. Starting from Neumann boundary conditions, the prescribed boundary-source problem in a finite homogeneous node is reformulated, through equivalent source theory, as a superposition of source–response problems in an infinite homogeneous medium. The resulting source–response relation is expressed in a parametric form, enabling efficient response matrix generation via precomputed interpolation tables. The boundary-source-to-flux response matrix Rs is directly compatible with discontinuity factors from generalized equivalence theory and response correction factors from the recently proposed response equivalent theory (RET). Numerical results for the IAEA 2D benchmark, the IAEA 3D benchmark, and a small PWR problem with discontinuity factors show that ESRMM with one node per assembly is more accurate than a 1 × 1 nodal discretization and approaches the accuracy of a 2 × 2 subdivision. The two-dimensional problems complete in milliseconds, and the IAEA 3D benchmark completes in 0.12 s.
This study evaluates the applicability of Regulatory Guide 1.99 Rev. 2 (RG 1.99) and ASTM E900-15—candidate models for future regulatory adoption—to SA508 Gr.3 Class 2 high-strength steel, a reactor pressure vessel material under consideration for innovative small modular reactors. As a regulatory-basis pre-assessment, materials from a combined Korean and U.S. surveillance database were classified as Class 1- or Class 2-equivalent based on the unirradiated yield strength and ultimate tensile strength, and the residuals of each model were statistically analyzed by product form. Under RG 1.99 Position 1.1, the forging residuals of Class 2 were significantly lower than those of Class 1. This result reflects a conservative overprediction of the transition temperature shift and is therefore acceptable from a regulatory safety standpoint. Under RG 1.99 Position 2.1, the unit-specific chemistry factor recalculation eliminated all significant class differences. ASTM E900-15 yielded similar conclusions while resolving the high-fluence underprediction bias of RG 1.99. No combination of model and product form exhibited a nonconservative bias for Class 2 within the analyzed surveillance database. These conclusions are limited to retrospectively classified Class 2-equivalent materials. Therefore, direct validation using actual SA508 Gr.3 Class 2 materials remains necessary.
