Pub Date : 2026-01-08DOI: 10.1016/j.anucene.2026.112126
M. Hadouachi , K. Laazouzi , O. Belhaj , H. El Yaakoubi , A. Arectout , Abdelhamid Nouayti , H. Boukhal , E. Chakir , T. El Bardouni
In nuclear reactor criticality and stability studies, nuclear data uncertainties can significantly influence integral parameters such as the effective multiplication factor and neutron flux, which are directly linked to reactor safety margins and operational performance. It is therefore essential to quantify the impact of nuclear data uncertainties on reactor calculations. In this work, machine learning techniques were applied to identify the nuclear data that have the greatest impact on criticality calculations. For this purpose, sensitivity profiles, combined with other benchmark characteristics, were used as input features for various machine learning algorithms to predict the bias . In order to interpret the model’s predictions, a SHAP (SHapley Additive exPlanations) analysis was applied to determine which reactions had the greatest influence on (k) bias. The results highlight that nuclear data for nuclides such as 239Pu, 235U, 233U, 238U, 12C, and 1H are the most important parameters related to a high .
{"title":"Evaluation of ENDF/B-VIII.0 nuclear data for criticality calculations using machine learning and the SHAP interpretability method","authors":"M. Hadouachi , K. Laazouzi , O. Belhaj , H. El Yaakoubi , A. Arectout , Abdelhamid Nouayti , H. Boukhal , E. Chakir , T. El Bardouni","doi":"10.1016/j.anucene.2026.112126","DOIUrl":"10.1016/j.anucene.2026.112126","url":null,"abstract":"<div><div>In nuclear reactor criticality and stability studies, nuclear data uncertainties can significantly influence integral parameters such as the effective multiplication factor and neutron flux, which are directly linked to reactor safety margins and operational performance. It is therefore essential to quantify the impact of nuclear data uncertainties on reactor calculations. In this work, machine learning techniques were applied to identify the nuclear data that have the greatest impact on criticality calculations. For this purpose, sensitivity profiles, combined with other benchmark characteristics, were used as input features for various machine learning algorithms to predict the bias <span><math><mrow><mi>Δ</mi><msub><mrow><mi>k</mi></mrow><mrow><mi>eff</mi></mrow></msub></mrow></math></span>. In order to interpret the model’s predictions, a SHAP (SHapley Additive exPlanations) analysis was applied to determine which reactions had the greatest influence on (k<span><math><msub><mrow></mrow><mrow><mi>eff</mi></mrow></msub></math></span>) bias. The results highlight that nuclear data for nuclides such as <sup>239</sup>Pu, <sup>235</sup>U, <sup>233</sup>U, <sup>238</sup>U, <sup>12</sup>C, and <sup>1</sup>H are the most important parameters related to a high <span><math><mrow><mi>Δ</mi><msub><mrow><mi>k</mi></mrow><mrow><mi>eff</mi></mrow></msub></mrow></math></span>.</div></div>","PeriodicalId":8006,"journal":{"name":"Annals of Nuclear Energy","volume":"229 ","pages":"Article 112126"},"PeriodicalIF":2.3,"publicationDate":"2026-01-08","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"145922131","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":3,"RegionCategory":"工程技术","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}
The molten Chloride Fast Reactor (MCFR) emerges as one of the advanced nuclear reactor designs for use in a nuclear marine propulsion. This paper delineates the preliminary design of a 200 MWt Marine MCFR (MMCFR) intended as a propulsion for zero-carbon large container ship, focusing on the neutronic analysis and fuel cycle assessment. The MMCFR employs eutectic 66NaCl-34UCl3 as the fuel with 19.55 wt% enriched uranium as the initial fuel contained in a BeO-reflected core, operated as a long-lived core and batchwise refuelling to simplify reactivity control and refuelling mechanism in a constrained space. As the MMCFR is designed with a compact core and large initial reactivity, the innovative Partial Fuel Change scheme is proposed to optimise fuel consumption and reduce the strain in the front-end fuel cycle, with Constant Mol or Constant Replacement scenario. Initial reactivity is suppressed using burnable absorber (BA) rods and control drums are used to control the reactivity and core shutdown. Neutronic and depletion calculations for the MMCFR design were performed using Serpent-2 code and ENDF/B-VII.0 library. The optimum front-end fuel cycle was obtained to be Constant Replacement scenario with lowest uranium consumption. Meanwhile, excess reactivity can be maintained below 5% throughout operational time by using BA and control drum, whilst temperature coefficient of reactivity (TCR) is sufficiently negative, ensuring the MMCFR fulfils the safety criteria.
{"title":"Preliminary design and front-end fuel cycle assessment of 200 MWt marine molten chloride fast reactor","authors":"Andika Putra Dwijayanto , Kenji Nishihara , Tomohiro Okamura , Masahiko Nakase","doi":"10.1016/j.anucene.2026.112124","DOIUrl":"10.1016/j.anucene.2026.112124","url":null,"abstract":"<div><div>The molten Chloride Fast Reactor (MCFR) emerges as one of the advanced nuclear reactor designs for use in a nuclear marine propulsion. This paper delineates the preliminary design of a 200 MWt Marine MCFR (MMCFR) intended as a propulsion for zero-carbon large container ship, focusing on the neutronic analysis and fuel cycle assessment. The MMCFR employs eutectic 66NaCl-34UCl<sub>3</sub> as the fuel with 19.55 wt% enriched uranium as the initial fuel contained in a BeO-reflected core, operated as a long-lived core and batchwise refuelling to simplify reactivity control and refuelling mechanism in a constrained space. As the MMCFR is designed with a compact core and large initial reactivity, the innovative Partial Fuel Change scheme is proposed to optimise fuel consumption and reduce the strain in the front-end fuel cycle, with Constant Mol or Constant Replacement scenario. Initial reactivity is suppressed using burnable absorber (BA) rods and control drums are used to control the reactivity and core shutdown. Neutronic and depletion calculations for the MMCFR design were performed using Serpent-2 code and ENDF/B-VII.0 library. The optimum front-end fuel cycle was obtained to be Constant Replacement scenario with lowest uranium consumption. Meanwhile, excess reactivity can be maintained below 5% throughout operational time by using BA and control drum, whilst temperature coefficient of reactivity (TCR) is sufficiently negative, ensuring the MMCFR fulfils the safety criteria.</div></div>","PeriodicalId":8006,"journal":{"name":"Annals of Nuclear Energy","volume":"229 ","pages":"Article 112124"},"PeriodicalIF":2.3,"publicationDate":"2026-01-08","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"145922130","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":3,"RegionCategory":"工程技术","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}
With the wide range of sea applications of reactors, it is necessary to develop a high-power LBE reactor with a small volume and high-power output. A process of core design is proposed in this paper for conducting rapid iterations. With this process and a series of design criteria and guidelines, the concept of the High-Power Lead-Bismuth Cooled Micro Reactor (HLCMR) is designed. This design adopts low-enriched fuel and two sets of independent shut-down control systems, which are arranged in a triangular pattern to achieve higher power density and lower power peak. After iterative design, the design power is 12 MW, and the lifetime is at least 5 years with the core power density of 76.72 MW/m3. The power distribution shows that the highest power peak is 1.444 in the operating state. The temperature field and reactivity are also calculated to evaluate the safety and reliability of this design. The results show that all parameters of the HLCMR meet the thermal–hydraulic and control design requirements.
{"title":"Design and performance analysis of high-power lead-bismuth cooled micro reactor","authors":"Yiming Xiong, Ren Li, Jilin Sun, Yuandong Zhang, Genglei Xia, Minjun Peng","doi":"10.1016/j.anucene.2026.112115","DOIUrl":"10.1016/j.anucene.2026.112115","url":null,"abstract":"<div><div>With the wide range of sea applications of reactors, it is necessary to develop a high-power LBE reactor with a small volume and high-power output. A process of core design is proposed in this paper for conducting rapid iterations. With this process and a series of design criteria and guidelines, the concept of the High-Power Lead-Bismuth Cooled Micro Reactor (HLCMR) is designed. This design adopts low-enriched fuel and two sets of independent shut-down control systems, which are arranged in a triangular pattern to achieve higher power density and lower power peak. After iterative design, the design power is 12 MW, and the lifetime is at least 5 years with the core power density of 76.72 MW/m<sup>3</sup>. The power distribution shows that the highest power peak is 1.444 in the operating state. The temperature field and reactivity are also calculated to evaluate the safety and reliability of this design. The results show that all parameters of the HLCMR meet the thermal–hydraulic and control design requirements.</div></div>","PeriodicalId":8006,"journal":{"name":"Annals of Nuclear Energy","volume":"230 ","pages":"Article 112115"},"PeriodicalIF":2.3,"publicationDate":"2026-01-08","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"145915168","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":3,"RegionCategory":"工程技术","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}
Pub Date : 2026-01-07DOI: 10.1016/j.anucene.2025.112110
Fengyuan Tian , Minyun Liu , Yanping Huang , Ruohan Zheng , Yangle Wang , Houjun Gong , Yu Tang , Tianzeng Liu , Jinghan Hu , Haohan Yuan , Yuan Zhou
The depressurization accident is one of the key problems during the transport and utilization of S-CO2. In this paper, an analysis model was developed to analyze the blowdown of S-CO2 from a simple vessel. The critical mass flow rate agreed well with experiment results, and the maximum error is less than 10%. As for the depressurization model, the homogeneous model could predict pressure and temperature, and the phase separation model and bubble rise model could predict the mass flow rate. Depressurization under different initial parameters, back pressures, breaks were analyzed. Results indicated that when the initial temperature exceeds a certain temperature, the pressure and mass flow rate would change smoothly. When the initial temperature is lower, the depressurization process could be divided into rapid depressurization, flash vaporization, and slow depressurization. The model developed may reflect the characteristics of depressurization and provide a reference for depressurization accidents of the S-CO2 system.
{"title":"Analysis of the blowdown of supercritical carbon dioxide from simple vessel","authors":"Fengyuan Tian , Minyun Liu , Yanping Huang , Ruohan Zheng , Yangle Wang , Houjun Gong , Yu Tang , Tianzeng Liu , Jinghan Hu , Haohan Yuan , Yuan Zhou","doi":"10.1016/j.anucene.2025.112110","DOIUrl":"10.1016/j.anucene.2025.112110","url":null,"abstract":"<div><div>The depressurization accident is one of the key problems during the transport and utilization of S-CO<sub>2</sub>. In this paper, an analysis model was developed to analyze the blowdown of S-CO<sub>2</sub> from a simple vessel. The critical mass flow rate agreed well with experiment results, and the maximum error is less than 10%. As for the depressurization model, the homogeneous model could predict pressure and temperature, and the phase separation model and bubble rise model could predict the mass flow rate. Depressurization under different initial parameters, back pressures, breaks were analyzed. Results indicated that when the initial temperature exceeds a certain temperature, the pressure and mass flow rate would change smoothly. When the initial temperature is lower, the depressurization process could be divided into rapid depressurization, flash vaporization, and slow depressurization. The model developed may reflect the characteristics of depressurization and provide a reference for depressurization accidents of the S-CO<sub>2</sub> system.</div></div>","PeriodicalId":8006,"journal":{"name":"Annals of Nuclear Energy","volume":"229 ","pages":"Article 112110"},"PeriodicalIF":2.3,"publicationDate":"2026-01-07","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"145922128","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":3,"RegionCategory":"工程技术","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}
A new high-fidelity neutronics analysis tool for fast reactor calculations is built by coupling the lattice code TULIP, the SN code HYDRA and the high-fidelity neutron transport code NECP-X. TULIP is employed to generate multigroup effective cross sections for assemblies based on the 0-D or 1-D cylindrical model. HYDRA utilizes a reactor core-reflector model to perform group condensation for reflector. The cross sections are passed to NECP-X to conduct a whole-core simulation with explicit geometry description. Verifications were conducted based on the Superphénix 2-D, MET-1000 2-D and JOYO MK-I 3-D core problems. The reference solutions were obtained through Monte Carlo calculations with NECP-MCX. The difference of keff was at most 298 pcm for 2-D problems and did not exceed 55 pcm for 3-D problem. The root mean square error of assembly power was a maximum of only 1 %. These results prove the capability of this method in fast reactor calculations.
{"title":"Calculation and analysis of self-shielded cross sections for the high-fidelity neutronics calculation of fast reactors","authors":"Xiang Li, Zhouyu Liu, Wenjie Chen, Liangzhi Cao, Hongchun Wu","doi":"10.1016/j.anucene.2025.112103","DOIUrl":"10.1016/j.anucene.2025.112103","url":null,"abstract":"<div><div>A new high-fidelity neutronics analysis tool for fast reactor calculations is built by coupling the lattice code TULIP, the SN code HYDRA and the high-fidelity neutron transport code NECP-X. TULIP is employed to generate multigroup effective cross sections for assemblies based on the 0-D or 1-D cylindrical model. HYDRA utilizes a reactor core-reflector model to perform group condensation for reflector. The cross sections are passed to NECP-X to conduct a whole-core simulation with explicit geometry description. Verifications were conducted based on the Superphénix 2-D, MET-1000 2-D and JOYO MK-I 3-D core problems. The reference solutions were obtained through Monte Carlo calculations with NECP-MCX. The difference of <em>k</em><sub>eff</sub> was at most 298 pcm for 2-D problems and did not exceed 55 pcm for 3-D problem. The root mean square error of assembly power was a maximum of only 1 %. These results prove the capability of this method in fast reactor calculations.</div></div>","PeriodicalId":8006,"journal":{"name":"Annals of Nuclear Energy","volume":"229 ","pages":"Article 112103"},"PeriodicalIF":2.3,"publicationDate":"2026-01-07","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"145922127","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":3,"RegionCategory":"工程技术","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}
Pub Date : 2026-01-07DOI: 10.1016/j.anucene.2025.112099
NK Maheshwari, Divij Kishal
Steam condensation plays a key role in removing heat from the containment in case of a postulated accident in water-cooled nuclear reactors. Steam released into the containment mixes with air present in that environment and condenses in the presence of noncondensable gases (air, helium, etc.) on the containment wall and other structures present in the containment. Advanced reactors design adopts passive containment cooling systems for long term containment cooling during the design basis and severe accident conditions. In this article, the research performed on free convective condensation in the presence of noncondensable gases on the tube outer surface has been reviewed. In the first part of the article, experimental studies have been covered. It is revealed that both the thermal hydraulic and geometrical parameters affect the condensation heat transfer in the presence of noncondensable gases. In the second part, various correlations developed by researchers are discussed accounting for thermal hydraulic, geometric parameters and nondimensional numbers; an assessment of these correlations is performed. In the third part, the theoretical model developed, results obtained and CFD studies performed by previous authors have been discussed. The effects of various parameters are discussed on the basis of experimental work and theoretical model developed. Finally, based on the review and studies performed, a summary is provided.
{"title":"Free convective condensation in the presence of noncondensable gases − A review with heat transfer studies","authors":"NK Maheshwari, Divij Kishal","doi":"10.1016/j.anucene.2025.112099","DOIUrl":"10.1016/j.anucene.2025.112099","url":null,"abstract":"<div><div>Steam condensation plays a key role in removing heat from the containment in case of a postulated accident in water-cooled nuclear reactors. Steam released into the containment mixes with air present in that environment and condenses in the presence of noncondensable gases (air, helium, etc.) on the containment wall and other structures present in the containment. Advanced reactors design adopts passive containment cooling systems for long term containment cooling during the design basis and severe accident conditions. In this article, the research performed on free convective condensation in the presence of noncondensable gases on the tube outer surface has been reviewed. In the first part of the article, experimental studies have been covered. It is revealed that both the thermal hydraulic and geometrical parameters affect the condensation heat transfer in the presence of noncondensable gases. In the second part, various correlations developed by researchers are discussed accounting for thermal hydraulic, geometric parameters and nondimensional numbers; an assessment of these correlations is performed. In the third part, the theoretical model developed, results obtained and CFD studies performed by previous authors have been discussed. The effects of various parameters are discussed on the basis of experimental work and theoretical model developed. Finally, based on the review and studies performed, a summary is provided.</div></div>","PeriodicalId":8006,"journal":{"name":"Annals of Nuclear Energy","volume":"229 ","pages":"Article 112099"},"PeriodicalIF":2.3,"publicationDate":"2026-01-07","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"145922092","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":3,"RegionCategory":"工程技术","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}
Pub Date : 2026-01-07DOI: 10.1016/j.anucene.2026.112113
Yuan Xu , Yun Long , Long Cai , Zhe Jiao
We investigate unsteady hydrodynamics of a shaft-sealed reactor coolant pump (RCP) across cold/hot states (25 °C/292 °C; 1/15.9 MPa) and tip-clearance variations from the design tip clearance of 0.8 mm (0.5 mm smaller and 1.0 mm and 2.0 mm larger than the design value). Using SST k–ω CFD with rotor–stator coupling, we quantify performance, pressure-pulsation spectra, and radial forces. Dominant components occur at the shaft frequency (1fn≈24.75 Hz) and diffuser blade-passing (5fn≈123.75 Hz), with a leading-peak shift to 5fn near the impeller outlet. Hot conditions intensify pulsations and radial loading and smooth the performance curve; at ∼ 1.2Q the hot-state head is ∼ 5 % higher. The radial force shows periodic modulation (≈3.22–3.42 kN) and a five-lobe pattern. Clearance changes alter head and spectra: +2 mm reduces head and elevates 5fn, whereas − 0.5 mm improves uniformity and still lowers head. These results provide a spectral-shift–load-coupled basis for clearance tolerance and operating-window selection to enhance RCP stability and safety.
{"title":"Unsteady hydrodynamics of an RCP impeller across clearances and hot conditions pressure pulsation, spectral shift, and radial force","authors":"Yuan Xu , Yun Long , Long Cai , Zhe Jiao","doi":"10.1016/j.anucene.2026.112113","DOIUrl":"10.1016/j.anucene.2026.112113","url":null,"abstract":"<div><div>We investigate unsteady hydrodynamics of a shaft-sealed reactor coolant pump (RCP) across cold/hot states (25 °C/292 °C; 1/15.9 MPa) and tip-clearance variations from the design tip clearance of 0.8 mm (0.5 mm smaller and 1.0 mm and 2.0 mm larger than the design value). Using SST k–ω CFD with rotor–stator coupling, we quantify performance, pressure-pulsation spectra, and radial forces. Dominant components occur at the shaft frequency (1fn≈24.75 Hz) and diffuser blade-passing (5fn≈123.75 Hz), with a leading-peak shift to 5fn near the impeller outlet. Hot conditions intensify pulsations and radial loading and smooth the performance curve; at ∼ 1.2Q the hot-state head is ∼ 5 % higher. The radial force shows periodic modulation (≈3.22–3.42 kN) and a five-lobe pattern. Clearance changes alter head and spectra: +2 mm reduces head and elevates 5fn, whereas − 0.5 mm improves uniformity and still lowers head. These results provide a spectral-shift–load-coupled basis for clearance tolerance and operating-window selection to enhance RCP stability and safety.</div></div>","PeriodicalId":8006,"journal":{"name":"Annals of Nuclear Energy","volume":"229 ","pages":"Article 112113"},"PeriodicalIF":2.3,"publicationDate":"2026-01-07","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"145922129","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":3,"RegionCategory":"工程技术","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}
Pub Date : 2026-01-06DOI: 10.1016/j.anucene.2025.112112
Shihao Dong , Junjie Deng , Pengcheng Zhao , Zijing Liu , Wei Li
Current research on the multi-scale coupling of reactors primarily focuses on the development of coupled simulation programs, which suffer from numerous uncertainties. This work establishes an uncertainty quantification (UQ) framework for multi-scale thermal–hydraulic (TH) coupling, which leverages the preCICE open-source platform to integrate the high-fidelity CFD code FLUENT, subchannel code SUBCHANFLOW, and UQ code DAKOTA. A 3 × 3 rod bundle configuration is used as a benchmark to validate the coupled framework under steady-state and transient conditions. Under steady-state conditions, the coupled model consistently predict axial temperature distributions when benchmarked against solvers (FLUENT and SUBCHANFLOW), validating the computational accuracy of multi-scale TH coupling. Under transient conditions with sinusoidal inlet flow variations, the outlet flow response synchronizes the period and phase with input perturbations, confirming the dynamic simulation capability of coupled system. Uncertainty quantification suggests that key parameters, including coolant temperature and peak cladding temperature, exhibit a normal distribution approximately. Sensitivity analysis reveals that inlet mass flow rate, outlet pressure, inlet temperature, and fuel rod heat flux are the dominant parameters influencing the system response. Overall, the proposed system exhibits reliable response characteristics under dynamic conditions.
{"title":"Uncertainty analysis method for the multi-scale coupling program based on preCICE","authors":"Shihao Dong , Junjie Deng , Pengcheng Zhao , Zijing Liu , Wei Li","doi":"10.1016/j.anucene.2025.112112","DOIUrl":"10.1016/j.anucene.2025.112112","url":null,"abstract":"<div><div>Current research on the multi-scale coupling of reactors primarily focuses on the development of coupled simulation programs, which suffer from numerous uncertainties. This work establishes an uncertainty quantification (UQ) framework for multi-scale thermal–hydraulic (TH) coupling, which leverages the preCICE open-source platform to integrate the high-fidelity CFD code FLUENT, subchannel code SUBCHANFLOW, and UQ code DAKOTA. A 3 × 3 rod bundle configuration is used as a benchmark to validate the coupled framework under steady-state and transient conditions. Under steady-state conditions, the coupled model consistently predict axial temperature distributions when benchmarked against solvers (FLUENT and SUBCHANFLOW), validating the computational accuracy of multi-scale TH coupling. Under transient conditions with sinusoidal inlet flow variations, the outlet flow response synchronizes the period and phase with input perturbations, confirming the dynamic simulation capability of coupled system. Uncertainty quantification suggests that key parameters, including coolant temperature and peak cladding temperature, exhibit a normal distribution approximately. Sensitivity analysis reveals that inlet mass flow rate, outlet pressure, inlet temperature, and fuel rod heat flux are the dominant parameters influencing the system response. Overall, the proposed system exhibits reliable response characteristics under dynamic conditions.</div></div>","PeriodicalId":8006,"journal":{"name":"Annals of Nuclear Energy","volume":"229 ","pages":"Article 112112"},"PeriodicalIF":2.3,"publicationDate":"2026-01-06","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"145922118","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":3,"RegionCategory":"工程技术","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}
Pub Date : 2026-01-06DOI: 10.1016/j.anucene.2026.112118
Wenming Yi , Feng Shen , GuoPing Quan , Xubo Ma , Guang Zhao
Accurate and efficient parameterization of assembly homogenized few-group constants is a critical challenge in reactor physics. Traditional methods are either too computationally expensive, like Monte Carlo codes, or struggle with the high-dimensional, non-linear relationships found in reactor data, like interpolation methods, especially when sample sizes are small. To address this, we propose a novel machine learning ensemble method, the Deep Neural Network with Gaussian Process Residual Correction (DNN-GPRC). This hybrid model uses a DNN to capture the primary data trends and a GPR to model and correct the DNN’s prediction residuals, leveraging GPR’s robustness on small datasets. Furthermore, we employ a Yeo–Johnson transformation in feature engineering to effectively mitigate the long-tail data distribution inherent in burnup calculations, significantly enhancing model performance. Tested on a small dataset of 2874 samples, the DNN-GPRC model consistently outperforms both standalone DNN and traditional linear interpolation methods. Crucially, on the test set, our model achieves a Root Mean Square Error of just 128 pcm for the infinite multiplication factor (), a result markedly superior to linear interpolation. This work demonstrates that the DNN-GPRC framework provides a high-accuracy, computationally efficient, and robust tool for few-group constant parameterization. It moves the field forward by enabling rapid and accurate analysis even in low-sample scenarios, which is vital for accelerating new reactor design cycles and improving simulation fidelity.
{"title":"Prediction of homogenized few-group constants for pressurized water reactor assembly using a Deep Neural Network with Gaussian Process Residual Correction","authors":"Wenming Yi , Feng Shen , GuoPing Quan , Xubo Ma , Guang Zhao","doi":"10.1016/j.anucene.2026.112118","DOIUrl":"10.1016/j.anucene.2026.112118","url":null,"abstract":"<div><div>Accurate and efficient parameterization of assembly homogenized few-group constants is a critical challenge in reactor physics. Traditional methods are either too computationally expensive, like Monte Carlo codes, or struggle with the high-dimensional, non-linear relationships found in reactor data, like interpolation methods, especially when sample sizes are small. To address this, we propose a novel machine learning ensemble method, the Deep Neural Network with Gaussian Process Residual Correction (DNN-GPRC). This hybrid model uses a DNN to capture the primary data trends and a GPR to model and correct the DNN’s prediction residuals, leveraging GPR’s robustness on small datasets. Furthermore, we employ a Yeo–Johnson transformation in feature engineering to effectively mitigate the long-tail data distribution inherent in burnup calculations, significantly enhancing model performance. Tested on a small dataset of 2874 samples, the DNN-GPRC model consistently outperforms both standalone DNN and traditional linear interpolation methods. Crucially, on the test set, our model achieves a Root Mean Square Error of just 128 pcm for the infinite multiplication factor (<span><math><msub><mrow><mi>k</mi></mrow><mrow><mi>i</mi><mi>n</mi><mi>f</mi></mrow></msub></math></span>), a result markedly superior to linear interpolation. This work demonstrates that the DNN-GPRC framework provides a high-accuracy, computationally efficient, and robust tool for few-group constant parameterization. It moves the field forward by enabling rapid and accurate analysis even in low-sample scenarios, which is vital for accelerating new reactor design cycles and improving simulation fidelity.</div></div>","PeriodicalId":8006,"journal":{"name":"Annals of Nuclear Energy","volume":"229 ","pages":"Article 112118"},"PeriodicalIF":2.3,"publicationDate":"2026-01-06","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"145922126","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":3,"RegionCategory":"工程技术","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}
Pub Date : 2026-01-06DOI: 10.1016/j.anucene.2025.112060
Peter J. Kriemadis, Adriaan Buijs
The Zero Energy Deuterium (ZED-2) reactor is a zero-power research reactor located at the Chalk River site of Canadian Nuclear Laboratories (CNL). The reactor was built to assist in neutronics code validation efforts for CANadian Deuterium Uranium (CANDU) reactors, but may find further use in the validation of computer codes used in the design of Small Modular Reactors (SMRs). This paper describes the application of the OpenMC and SERPENT 2 codes to two published benchmarks for ZED-2 neutronics experiments. The results were then compared to MCNP and MONK code results on file. Experiments from the International Criticality Safety Benchmark Evaluation Project (ICSBEP) handbook were reviewed to establish the differences one might expect from Monte Carlo code-to-code comparisons. The completed benchmarks were assessed against this review. In this manner, the OpenMC code is validated both against an experiment and against other validated codes.
{"title":"ZED-2 benchmarks performed in OpenMC and Serpent 2: A validation exercise for OpenMC applications","authors":"Peter J. Kriemadis, Adriaan Buijs","doi":"10.1016/j.anucene.2025.112060","DOIUrl":"10.1016/j.anucene.2025.112060","url":null,"abstract":"<div><div>The Zero Energy Deuterium (ZED-2) reactor is a zero-power research reactor located at the Chalk River site of Canadian Nuclear Laboratories (CNL). The reactor was built to assist in neutronics code validation efforts for CANadian Deuterium Uranium (CANDU) reactors, but may find further use in the validation of computer codes used in the design of Small Modular Reactors (SMRs). This paper describes the application of the OpenMC and SERPENT<!--> <!-->2 codes to two published benchmarks for ZED-2 neutronics experiments. The results were then compared to MCNP and MONK code results on file. Experiments from the International Criticality Safety Benchmark Evaluation Project (ICSBEP) handbook were reviewed to establish the differences one might expect from Monte Carlo code-to-code comparisons. The completed benchmarks were assessed against this review. In this manner, the OpenMC code is validated both against an experiment and against other validated codes.</div></div>","PeriodicalId":8006,"journal":{"name":"Annals of Nuclear Energy","volume":"229 ","pages":"Article 112060"},"PeriodicalIF":2.3,"publicationDate":"2026-01-06","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"145922117","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":3,"RegionCategory":"工程技术","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}
扫码关注我们
求助内容：
应助结果提醒方式：
京公网安备 11010802042870号