Xiaojing Liu
Current Organizations / Institutes
The organizations/institutes from the scholar's latest article
Previous Organizations / Institutes
The organizations / institutes from the scholar's other articles in the past ten years
Research Subjects
Xiaojing Liu has a total of 164 co-authors. Here are the top co-authors by number of shared publications:
Xiang Chai
School of Mechanical Engineering, Shanghai Jiao Tong University, Shanghai, China
15 shared publications- Hui He
School of Mechanical Engineering, Shanghai Jiao Tong University, Shanghai, China
11 shared publications Xu Cheng
Karlsruhe Inst Technol KIT, Inst Fus & Reactor Technol, Kaiserstr 12, D-76131 Karlsruhe, Germany
11 shared publicationsTengfei Zhang
School of Mechanical Engineering, Shanghai Jiao Tong University, Shanghai, China
10 shared publications- Tengfei Zhang
Shanghai Jiao Tong University
8 shared publications - Jinbiao Xiong
School of Nuclear Science and Engineering, Shanghai Jiao Tong University, Shanghai, China
5 shared publications
Latest Publications
- Open Access
- 12 October 2026
- Published by Elsevier in SSRN Electronic Journal
Experimental and Numerical Study of Saturated Pool Boiling under Ultrasonic Excitation
Enhanced boiling heat transfer is important for improving thermal performance and ensuring the safe operation of high-heat-flux systems. Ultrasound is a promising active enhancement technique because it can modify bubble dynamics and disturb near-wall liquid flow. However, the mechanisms by which ultrasonic-induced flow affects bubble dynamics and near-wall transport during saturated pool boiling remain insufficiently understood. In this study, saturated pool boiling experiments were conducted under ultrasonic excitation, followed by numerical simulations using a two-dimensional acoustics-fluid-thermal coupled model to elucidate the underlying enhancement mechanisms. The experimental results show that ultrasonic excitation induces pronounced lateral bubble migration and promotes earlier bubble departure. The average bubble departure period decreases from approximately 0.075 s to 0.045 s, corresponding to a reduction of about 40%, while the experimental heat transfer enhancement factor reaches approximately 1.4-1.6. Numerical results reveal that acoustic streaming modifies the near-wall velocity field, generates local vortical structures, and enhances liquid disturbance around the bubble, thereby promoting lateral bubble migration and earlier detachment and improving local heat transfer. The maximum numerical heat transfer enhancement of 11.82% is obtained at 28 kHz with an acoustic amplitude of 1 μm. Within the investigated frequency range of 28-40 kHz, the enhancement decreases with increasing frequency, which may be associated with frequency-dependent changes in the acoustic field, acoustic streaming, and the resulting near-wall flow structures. These findings provide a mechanistic understanding of ultrasonic enhancement of saturated pool boiling and highlight the important role of acoustic streaming in coupling bubble dynamics with near-wall heat transfer.
- 12 October 2026
- Published by Elsevier in SSRN Electronic Journal
Open AccessExperimental and Numerical Study of Saturated Pool Boiling under Ultrasonic Excitation
Enhanced boiling heat transfer is important for improving thermal performance and ensuring the safe operation of high-heat-flux systems. Ultrasound is a promising active enhancement technique because it can modify bubble dynamics and disturb near-wall liquid flow. However, the mechanisms by which ultrasonic-induced flow affects bubble dynamics and near-wall transport during saturated pool boiling remain insufficiently understood. In this study, saturated pool boiling experiments were conducted under ultrasonic excitation, followed by numerical simulations using a two-dimensional acoustics-fluid-thermal coupled model to elucidate the underlying enhancement mechanisms. The experimental results show that ultrasonic excitation induces pronounced lateral bubble migration and promotes earlier bubble departure. The average bubble departure period decreases from approximately 0.075 s to 0.045 s, corresponding to a reduction of about 40%, while the experimental heat transfer enhancement factor reaches approximately 1.4-1.6. Numerical results reveal that acoustic streaming modifies the near-wall velocity field, generates local vortical structures, and enhances liquid disturbance around the bubble, thereby promoting lateral bubble migration and earlier detachment and improving local heat transfer. The maximum numerical heat transfer enhancement of 11.82% is obtained at 28 kHz with an acoustic amplitude of 1 μm. Within the investigated frequency range of 28-40 kHz, the enhancement decreases with increasing frequency, which may be associated with frequency-dependent changes in the acoustic field, acoustic streaming, and the resulting near-wall flow structures. These findings provide a mechanistic understanding of ultrasonic enhancement of saturated pool boiling and highlight the important role of acoustic streaming in coupling bubble dynamics with near-wall heat transfer.
- Open Access
- 28 September 2026
- Published by Elsevier in SSRN Electronic Journal
Accelerating the Stiffness Confinement Method through inter-step slope continuity
The Stiffness Confinement Method (SCM) uses a frequency transformation to reformulate the space-time neutron kinetics equations as a transient eigenvalue problem (TEVP), where the neutron flux frequency represents its relative growth rate. Macroscopic cross-section modifications enable SCM to simulate transient problems with an unchanged steady-state eigenvalue solver. At each time step the frequency is adjusted iteratively until the dynamic eigenvalue reaches unity. Each trial requires an eigenvalue solve, making this frequency search the main computational cost of SCM. In this work, first-order perturbation theory gives a local linear relation between the frequency and the dynamic eigenvalue, and explains the error-propagation mechanism between neighboring time steps. On this basis, an acceleration method is proposed that reuses the converged slope of the frequency--eigenvalue relation from the preceding time step to initialize the current secant search. It stores only a few scalar values and requires no modification to the eigenvalue solver. The method is tested on the LMW transient, the NEACRP rod-ejection benchmark, and two micro modular reactor (MMR) cases with different control-rod histories. At the two convergence tolerances considered here, it reduces the number of eigenvalue solves to 36.5%--57.5% of that required by Conventional SCM, with corresponding single-core wall-clock speedups of 1.70--2.57$\times$. The reference-backed cases retain power accuracy comparable to that of the baseline at the $10^{-7}$ tolerance.- 28 September 2026
- Published by Elsevier in SSRN Electronic Journal
Open AccessAccelerating the Stiffness Confinement Method through inter-step slope continuity
The Stiffness Confinement Method (SCM) uses a frequency transformation to reformulate the space-time neutron kinetics equations as a transient eigenvalue problem (TEVP), where the neutron flux frequency represents its relative growth rate. Macroscopic cross-section modifications enable SCM to simulate transient problems with an unchanged steady-state eigenvalue solver. At each time step the frequency is adjusted iteratively until the dynamic eigenvalue reaches unity. Each trial requires an eigenvalue solve, making this frequency search the main computational cost of SCM. In this work, first-order perturbation theory gives a local linear relation between the frequency and the dynamic eigenvalue, and explains the error-propagation mechanism between neighboring time steps. On this basis, an acceleration method is proposed that reuses the converged slope of the frequency--eigenvalue relation from the preceding time step to initialize the current secant search. It stores only a few scalar values and requires no modification to the eigenvalue solver. The method is tested on the LMW transient, the NEACRP rod-ejection benchmark, and two micro modular reactor (MMR) cases with different control-rod histories. At the two convergence tolerances considered here, it reduces the number of eigenvalue solves to 36.5%--57.5% of that required by Conventional SCM, with corresponding single-core wall-clock speedups of 1.70--2.57$\times$. The reference-backed cases retain power accuracy comparable to that of the baseline at the $10^{-7}$ tolerance. - Open Access
- 10 September 2026
- Published by Elsevier in SSRN Electronic Journal
Interfacial Bonding Mechanism of Nife2o4 On M-Zro2 And Α-Cr2O3 Surfaces for CRUD-Resistant Coatings
The deposition and growth of NiFe2O4-rich corrosion-related unidentified deposits (CRUD) on nuclear fuel cladding can compromise nuclear power plant economics and safety. Preliminary experiments suggest that Cr-coated cladding can suppress CRUD relative to conventional Zr-alloy cladding. However, the underlying suppression mechanism remains unknown due to lack of understanding on the interfaces formed by NiFe2O4 and cladding surface oxide (ZrO2 or α-Cr2O3). Here, first-principles calculations were performed to compare the adhesion and bonding of NiFe2O4(111) with monoclinic ZrO2 and α-Cr2O3. The calculated work of adhesion (Wad) is 15.93–31.83 J m−2 for the ZrO2 interfaces, substantially higher than 4.41–5.92 J m−2 for the Cr2O3 interfaces. For the highest-Wad models, the ZrO2 interface contains more bridging O atoms per unit area than the Cr2O3 interface (0.093 versus 0.071 Å−2) and shows larger charge-density changes. Analysis of the charge-density difference and projected density of states shows that bonding at the ZrO2 interface occurs mainly through bridging O atoms, whereas the Cr2O3 interface also contains a weaker direct Fe–Cr interaction, consistent with its lower Wad. The NiFe2O4/Cr2O3 interface therefore requires less work for separation and may contribute to lower retention of the first NiFe2O4-rich particles on oxidized Cr coatings.
- 10 September 2026
- Published by Elsevier in SSRN Electronic Journal
Open AccessInterfacial Bonding Mechanism of Nife2o4 On M-Zro2 And Α-Cr2O3 Surfaces for CRUD-Resistant Coatings
The deposition and growth of NiFe2O4-rich corrosion-related unidentified deposits (CRUD) on nuclear fuel cladding can compromise nuclear power plant economics and safety. Preliminary experiments suggest that Cr-coated cladding can suppress CRUD relative to conventional Zr-alloy cladding. However, the underlying suppression mechanism remains unknown due to lack of understanding on the interfaces formed by NiFe2O4 and cladding surface oxide (ZrO2 or α-Cr2O3). Here, first-principles calculations were performed to compare the adhesion and bonding of NiFe2O4(111) with monoclinic ZrO2 and α-Cr2O3. The calculated work of adhesion (Wad) is 15.93–31.83 J m−2 for the ZrO2 interfaces, substantially higher than 4.41–5.92 J m−2 for the Cr2O3 interfaces. For the highest-Wad models, the ZrO2 interface contains more bridging O atoms per unit area than the Cr2O3 interface (0.093 versus 0.071 Å−2) and shows larger charge-density changes. Analysis of the charge-density difference and projected density of states shows that bonding at the ZrO2 interface occurs mainly through bridging O atoms, whereas the Cr2O3 interface also contains a weaker direct Fe–Cr interaction, consistent with its lower Wad. The NiFe2O4/Cr2O3 interface therefore requires less work for separation and may contribute to lower retention of the first NiFe2O4-rich particles on oxidized Cr coatings.
- Open Access
- 8 September 2026
- Published by Springer Nature in Communications Engineering
Reverse design of fluid-solid interfacial bumps for passive heat transfer enhancement
- 8 September 2026
- Published by Springer Nature in Communications Engineering
Open AccessReverse design of fluid-solid interfacial bumps for passive heat transfer enhancement
- Open Access
- 16 July 2026
- Published by Elsevier in SSRN Electronic Journal
A Current-Driven On-the-Fly Environment Correction Method for Fission Response Function Neutronics Calculations and Integration in FLASH
The fission response function (FRF) method provides an efficient two-step, pin-by-pin neutronics framework for reactor core analysis. Because the response functions are generated in uniform assembly-level database models, inter-assembly environmental factor (IaEF) corrections are required in heterogeneous core environments. In the conventional approach, IaEFs are pre-computed by dedicated supercell calculations for every neighboring-assembly combination, which imposes a heavy database-generation burden and becomes infeasible for burnup-dependent core conditions. This work proposes a current-driven on-the-fly (OTF) IaEF generation method and implements it in the FLASH code. The thermal outward currents and current-to-fission response functions are tallied once per assembly type during the existing FRF database generation, and the IaEFs for the actual neighboring environments are reconstructed on line during the core calculation. As a result, the IaEF generation no longer requires the approximately 12-hour offline pre-computation and takes only tens of seconds during the core calculation. Environmental effects from enrichment, burnable absorber loading, burnup, and other state parameters are corrected synchronously in this process. For the simplified BEAVRS full-core benchmark, the OTF method gives a k-eff deviation of 149 pcm and a pin-wise fission reaction rate RMS error of 1.19% relative to the Monte Carlo reference solution, comparable in accuracy to the conventional pre-computed IaEF method.
- 16 July 2026
- Published by Elsevier in SSRN Electronic Journal
Open AccessA Current-Driven On-the-Fly Environment Correction Method for Fission Response Function Neutronics Calculations and Integration in FLASH
The fission response function (FRF) method provides an efficient two-step, pin-by-pin neutronics framework for reactor core analysis. Because the response functions are generated in uniform assembly-level database models, inter-assembly environmental factor (IaEF) corrections are required in heterogeneous core environments. In the conventional approach, IaEFs are pre-computed by dedicated supercell calculations for every neighboring-assembly combination, which imposes a heavy database-generation burden and becomes infeasible for burnup-dependent core conditions. This work proposes a current-driven on-the-fly (OTF) IaEF generation method and implements it in the FLASH code. The thermal outward currents and current-to-fission response functions are tallied once per assembly type during the existing FRF database generation, and the IaEFs for the actual neighboring environments are reconstructed on line during the core calculation. As a result, the IaEF generation no longer requires the approximately 12-hour offline pre-computation and takes only tens of seconds during the core calculation. Environmental effects from enrichment, burnable absorber loading, burnup, and other state parameters are corrected synchronously in this process. For the simplified BEAVRS full-core benchmark, the OTF method gives a k-eff deviation of 149 pcm and a pin-wise fission reaction rate RMS error of 1.19% relative to the Monte Carlo reference solution, comparable in accuracy to the conventional pre-computed IaEF method.
- Open Access
- 16 May 2026
- Published by Elsevier in SSRN Electronic Journal
A Human-in-the-Loop Machine Learning Framework for Online Fault Diagnosis of Space Power Systems
Ensuring the stable operation of space power systems is a critical challenge for long-duration missions. Remote operation and communication latency hinder timely fault detection and make human expert intervention more difficult, while limited historical data restricts the generalizability of machine learning models. This study proposes a human-in-the-loop machine learning diagnostic framework based on a Transformer backbone, with fusion mechanisms at both the feature and decision levels to integrate expert knowledge and data-driven learning. To further characterize the interaction between human and machine intelligence, a multi-stage game-theoretic method is introduced to quantify their respective contributions in the decision-making process. Experiments on the TOPAZ-II power system, involving 11205 noisy samples across 13 fault types, show that the proposed framework improves diagnostic accuracy and robustness compared to both data-driven and human-only approaches. Under 10% data noise and 30% expert error, it achieves 98.19% precision and 98.22% recall. Additional evaluation on the MVTec benchmark demonstrates its superiority over existing models. The framework provides a scalable, interpretable, and effective solution for fault diagnosis in data-limited, high-risk environments.
- 16 May 2026
- Published by Elsevier in SSRN Electronic Journal
Open AccessA Human-in-the-Loop Machine Learning Framework for Online Fault Diagnosis of Space Power Systems
Ensuring the stable operation of space power systems is a critical challenge for long-duration missions. Remote operation and communication latency hinder timely fault detection and make human expert intervention more difficult, while limited historical data restricts the generalizability of machine learning models. This study proposes a human-in-the-loop machine learning diagnostic framework based on a Transformer backbone, with fusion mechanisms at both the feature and decision levels to integrate expert knowledge and data-driven learning. To further characterize the interaction between human and machine intelligence, a multi-stage game-theoretic method is introduced to quantify their respective contributions in the decision-making process. Experiments on the TOPAZ-II power system, involving 11205 noisy samples across 13 fault types, show that the proposed framework improves diagnostic accuracy and robustness compared to both data-driven and human-only approaches. Under 10% data noise and 30% expert error, it achieves 98.19% precision and 98.22% recall. Additional evaluation on the MVTec benchmark demonstrates its superiority over existing models. The framework provides a scalable, interpretable, and effective solution for fault diagnosis in data-limited, high-risk environments.
- Open Access
- 30 April 2026
- Published by Elsevier in SSRN Electronic Journal
Dynamic mode decomposition for multi-cycle burnup distribution prediction in the high-temperature gas-cooled reactor
In the Pebble-Bed High Temperature Gas-cooled Reactor (PB-HTGR), the multi-pass fuel recirculation scheme and the double heterogeneity arising from TRi-structural ISOtropic (TRISO) coated particles and fuel pebbles give rise to a spatially heterogeneous burnup distribution, of which the accurate evaluation across multiple fuel cycling steps requires iterative multi-physics coupled calculations, demanding substantial computational resources and time. To efficiently evaluate the specific characteristics of the burnup distribution in PB-HTGRs, this work presents the application of the dynamic mode decomposition (DMD) and its Koopman-based extension (eDMD) as non-intrusive, data-driven reduced-order models for predicting the multi-cycle spatiotemporal burnup evolution. A dataset of 12,000 parametric cases is generated by varying six operating parameters under a deterministic fuel shuffling scheme, each providing a 12-cycle burnup sequence on a mesh. The DMD retains 16 modes via a 99.999% cumulative energy criterion and achieves satisfactory accuracy in early cycles but exhibits progressive error accumulation with the MAPE rising to 18.74% at cycle 12 and the upper core layers reaching 35-61%, owing to the inability of a single linear operator to capture nonlinear perturbations from cycle-dependent fuel loading. To address this problem, an eDMD employing quadratic polynomial lifting functions maps the 120-dimensional state into a 240-dimensional feature space to approximate the Koopman operator. The optimal configuration from a 48-combination grid search retains 23 modes and reduces the overall MAPE to 3.80% with the cycle-12 upper core MAPE decreasing from 31.2% to 6.8%. These results demonstrate that the DMD effectively captures dominant linear dynamics in early cycles but its limited nonlinear capability leads to error accumulation in later cycles, while the Koopman-based eDMD overcomes this limitation, effectively capturing the nonlinear spatiotemporal dynamics of burnup evolution in the core.
- 30 April 2026
- Published by Elsevier in SSRN Electronic Journal
Open AccessDynamic mode decomposition for multi-cycle burnup distribution prediction in the high-temperature gas-cooled reactor
In the Pebble-Bed High Temperature Gas-cooled Reactor (PB-HTGR), the multi-pass fuel recirculation scheme and the double heterogeneity arising from TRi-structural ISOtropic (TRISO) coated particles and fuel pebbles give rise to a spatially heterogeneous burnup distribution, of which the accurate evaluation across multiple fuel cycling steps requires iterative multi-physics coupled calculations, demanding substantial computational resources and time. To efficiently evaluate the specific characteristics of the burnup distribution in PB-HTGRs, this work presents the application of the dynamic mode decomposition (DMD) and its Koopman-based extension (eDMD) as non-intrusive, data-driven reduced-order models for predicting the multi-cycle spatiotemporal burnup evolution. A dataset of 12,000 parametric cases is generated by varying six operating parameters under a deterministic fuel shuffling scheme, each providing a 12-cycle burnup sequence on a mesh. The DMD retains 16 modes via a 99.999% cumulative energy criterion and achieves satisfactory accuracy in early cycles but exhibits progressive error accumulation with the MAPE rising to 18.74% at cycle 12 and the upper core layers reaching 35-61%, owing to the inability of a single linear operator to capture nonlinear perturbations from cycle-dependent fuel loading. To address this problem, an eDMD employing quadratic polynomial lifting functions maps the 120-dimensional state into a 240-dimensional feature space to approximate the Koopman operator. The optimal configuration from a 48-combination grid search retains 23 modes and reduces the overall MAPE to 3.80% with the cycle-12 upper core MAPE decreasing from 31.2% to 6.8%. These results demonstrate that the DMD effectively captures dominant linear dynamics in early cycles but its limited nonlinear capability leads to error accumulation in later cycles, while the Koopman-based eDMD overcomes this limitation, effectively capturing the nonlinear spatiotemporal dynamics of burnup evolution in the core.
- Open Access
- 27 April 2026
- Published by Elsevier in SSRN Electronic Journal
A highly selective adsorbent for removal of 110mAg based on PP-g-GMA@MoS4
Based on previous operational experience, activated corrosion products account for approximately 80% of the radioactive nuclides during refueling outages. After taking measures to control 58Co and 60Co, 110mAg has emerged as the main contributor to the radiation field in the primary loop. Traditionally, ion exchangers in Chemical and Volume Control Systems (CVCS) are responsible for removing 110mAg. However, the limited adsorption capacity and poor selectivity of these ion exchangers result in the generation of substantial solid radioactive waste, which directly increases the treatment cost.,To address this challenge, we developed PP-g-GMA@MoS4, a MoS42--functionalized PP nonwoven fabric synthesized via irradiation copolymerization and ion exchange. This adsorbent removes >95% of Ag+ across pH 1–5.9. Even at the most challenging pH of 3, where the adsorption performance of PP-g-GMA@MoS4 was the least favorable, its adsorption capacity still reached 623.05 mg/g during experiments, demonstrating better performance than previously documented adsorbents. The adsorption process was better described by the Freundlich isotherm and the pseudo-first-order kinetic model. Under conditions with 80-fold competitive ions, the distribution coefficient of Ag+ still reached 105 mL/g and the separation factors of Ag+ relative to each coexisting metal ion (SFAg,m) were 104 times higher than those of commercial resins. XPS analysis indicated that Ag-S bond formation is involved in the adsorption mechanism. These features—high capacity, outstanding selectivity, and low material cost—make PP-g-GMA@MoS4 a promising candidate for minimizing radioactive waste generation during 110mAg removal in nuclear power plants.
- 27 April 2026
- Published by Elsevier in SSRN Electronic Journal
Open AccessA highly selective adsorbent for removal of 110mAg based on PP-g-GMA@MoS4
Based on previous operational experience, activated corrosion products account for approximately 80% of the radioactive nuclides during refueling outages. After taking measures to control 58Co and 60Co, 110mAg has emerged as the main contributor to the radiation field in the primary loop. Traditionally, ion exchangers in Chemical and Volume Control Systems (CVCS) are responsible for removing 110mAg. However, the limited adsorption capacity and poor selectivity of these ion exchangers result in the generation of substantial solid radioactive waste, which directly increases the treatment cost.,To address this challenge, we developed PP-g-GMA@MoS4, a MoS42--functionalized PP nonwoven fabric synthesized via irradiation copolymerization and ion exchange. This adsorbent removes >95% of Ag+ across pH 1–5.9. Even at the most challenging pH of 3, where the adsorption performance of PP-g-GMA@MoS4 was the least favorable, its adsorption capacity still reached 623.05 mg/g during experiments, demonstrating better performance than previously documented adsorbents. The adsorption process was better described by the Freundlich isotherm and the pseudo-first-order kinetic model. Under conditions with 80-fold competitive ions, the distribution coefficient of Ag+ still reached 105 mL/g and the separation factors of Ag+ relative to each coexisting metal ion (SFAg,m) were 104 times higher than those of commercial resins. XPS analysis indicated that Ag-S bond formation is involved in the adsorption mechanism. These features—high capacity, outstanding selectivity, and low material cost—make PP-g-GMA@MoS4 a promising candidate for minimizing radioactive waste generation during 110mAg removal in nuclear power plants.
- 1 April 2026
- Published by Springer Nature
- p. 261-271
Time Series Forecasting Framework Based on Autocorrelation Feature and Transformer Architecture
- 1 April 2026
- Published by Springer Nature
- p. 261-271
Time Series Forecasting Framework Based on Autocorrelation Feature and Transformer Architecture
- Open Access
- 13 March 2026
- Published by Elsevier in SSRN Electronic Journal
Oxygen Concentration Control Strategies in Liquid Lead-Bismuth Eutectic Alloy-II: Optimization of Mechanical Strength and Heat Transfer
Active oxygen concentration control is a potential approach to mitigate lead-bismuth corrosion and ensure the long-term operation of lead-bismuth fast reactors (LFRs). This paper presents the second part of a two-part study on oxygen concentration control strategies in liquid lead-bismuth eutectic (LBE) alloy. A multi-objective, multi-strategy optimization algorithm, termed HSA-CAM-WOA (Hybrid Simulated Annealing Chaos Adaptive Mutation Whale Optimization Algorithm), is developed by integrating simulated annealing, adaptive weighting, Gaussian-Cauchy mutation, and chaotic map initialization techniques. Utilizing HSA-CAM-WOA, dynamic oxygen concentration control for the long-term operation of LFRs is optimized with respect to mechanical strength and heat transfer performance, and a novel optimal oxygen concentration control strategy containing four phases of initial low oxygen concentration, compensation, dynamic adjustment and final correction is proposed. Whereby, the thickness of magnetite layer in high temperature regions is substantially reduced with peak thickness decreasing by up to 87.2%, and heat transfer performance is accordingly improved by virtue of peak thermal resistance of oxide layer reduced by 67.5%. The uniformity of the cladding interface is also enhanced with the peak thickness difference narrowed by 36.4% to 55.3%, and the continuous presence of the protective spinel layer is guaranteed. The application of optimal oxygen concentration control strategy contributes to the long-term corrosion resistance of cladding in reactor core with non-isothermal LBE.
- 13 March 2026
- Published by Elsevier in SSRN Electronic Journal
Open AccessOxygen Concentration Control Strategies in Liquid Lead-Bismuth Eutectic Alloy-II: Optimization of Mechanical Strength and Heat Transfer
Active oxygen concentration control is a potential approach to mitigate lead-bismuth corrosion and ensure the long-term operation of lead-bismuth fast reactors (LFRs). This paper presents the second part of a two-part study on oxygen concentration control strategies in liquid lead-bismuth eutectic (LBE) alloy. A multi-objective, multi-strategy optimization algorithm, termed HSA-CAM-WOA (Hybrid Simulated Annealing Chaos Adaptive Mutation Whale Optimization Algorithm), is developed by integrating simulated annealing, adaptive weighting, Gaussian-Cauchy mutation, and chaotic map initialization techniques. Utilizing HSA-CAM-WOA, dynamic oxygen concentration control for the long-term operation of LFRs is optimized with respect to mechanical strength and heat transfer performance, and a novel optimal oxygen concentration control strategy containing four phases of initial low oxygen concentration, compensation, dynamic adjustment and final correction is proposed. Whereby, the thickness of magnetite layer in high temperature regions is substantially reduced with peak thickness decreasing by up to 87.2%, and heat transfer performance is accordingly improved by virtue of peak thermal resistance of oxide layer reduced by 67.5%. The uniformity of the cladding interface is also enhanced with the peak thickness difference narrowed by 36.4% to 55.3%, and the continuous presence of the protective spinel layer is guaranteed. The application of optimal oxygen concentration control strategy contributes to the long-term corrosion resistance of cladding in reactor core with non-isothermal LBE.