Pub Date : 2026-05-01Epub Date: 2025-12-19DOI: 10.1016/j.ijfatigue.2025.109445
Peng Liu, Haoyuan Li, Hailong Tian, Lai Wei, Yunshenghao Qiu
Dynamic load spectra for electric-drive assemblies are difficult to estimate from road tests because the signals are non-stationary, non-Gaussian, and noisy. We propose a pseudo-damage-constrained data–model fusion framework that reconstructs torque/load histories while preserving rainflow counting and fatigue consistency. The approach combines trend extraction with nonlinear state estimation and an innovation-based adaptive step that enforces pseudo-damage equivalence to the raw signal within a controlled tolerance. Extreme-value fits are used only as tail diagnostics to verify that rare high-load behavior is preserved; they are not involved in cycle counting. On representative road data, the method achieved a Peak–Valley Preservation Rate ≈93% and the lowest weighted-MAPE (26.2%) among EKF, PF, KalmanNet, and LSTM baselines, with clear gains in fatigue-critical mid–high levels and no inflation of the spectrum tail. The results indicate that the proposed framework yields higher-fidelity spectra for durability analysis and test-bench replay while keeping established fatigue rules (four-point rainflow with Goodman correction) unchanged.
{"title":"A pseudo-damage-constrained data–model fusion method for dynamic load spectrum estimation in electric-drive assemblies","authors":"Peng Liu, Haoyuan Li, Hailong Tian, Lai Wei, Yunshenghao Qiu","doi":"10.1016/j.ijfatigue.2025.109445","DOIUrl":"10.1016/j.ijfatigue.2025.109445","url":null,"abstract":"<div><div>Dynamic load spectra for electric-drive assemblies are difficult to estimate from road tests because the signals are non-stationary, non-Gaussian, and noisy. We propose a pseudo-damage-constrained data–model fusion framework that reconstructs torque/load histories while preserving rainflow counting and fatigue consistency. The approach combines trend extraction with nonlinear state estimation and an innovation-based adaptive step that enforces pseudo-damage equivalence to the raw signal within a controlled tolerance. Extreme-value fits are used only as tail diagnostics to verify that rare high-load behavior is preserved; they are not involved in cycle counting. On representative road data, the method achieved a Peak–Valley Preservation Rate ≈93% and the lowest weighted-MAPE (26.2%) among EKF, PF, KalmanNet, and LSTM baselines, with clear gains in fatigue-critical mid–high levels and no inflation of the spectrum tail. The results indicate that the proposed framework yields higher-fidelity spectra for durability analysis and test-bench replay while keeping established fatigue rules (four-point rainflow with Goodman correction) unchanged.</div></div>","PeriodicalId":14112,"journal":{"name":"International Journal of Fatigue","volume":"206 ","pages":"Article 109445"},"PeriodicalIF":6.8,"publicationDate":"2026-05-01","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"145784901","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":2,"RegionCategory":"材料科学","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}
Pub Date : 2026-05-01Epub Date: 2025-12-29DOI: 10.1016/j.ijfatigue.2025.109469
Luohuan Zou , Yu Gong , Dingli Tian , Sizhuo Hao , Jianyu Zhang , Libin Zhao , Ning Hu
Delamination usually occurs and grows in composite laminates under fatigue loading. The stress ratio is an important factor, while its influence law has no consensus yet. In this paper, to fully investigate the influence of fiber bridging and stress ratio on the fatigue delamination behavior, mode I fatigue delamination tests under two stress ratios (0.1 and 0.5) are conducted. Test results reveal that, the initial and steady-state values of the fatigue R-curve are consistent with those of quasi-static ones, while there are significant differences in the growth stage of fiber bridging. Furthermore, it is found that, the slope and intercept of the da/dN-Gmax curves vary under different stress ratios. A novel four-parameter fatigue model considering fiber bridging and stress ratio effects is proposed. The proposed model is compared with other classical models in literatures using the fatigue data of two stress ratios (0.1 and 0.5). It is found that the proposed model can well characterize fatigue delamination behavior. To further verify the model applicability, fatigue tests under stress ratio of 0.3 are supplemented. The predicted da/dN-Gmax curves by the model and experimental results are compared with a 95% confidence interval, which indicates that the proposed model has good applicability and can provide an effective method for fatigue delamination prediction.
{"title":"A new empirical model for mode I fatigue delamination of composite laminates considering fiber bridging and stress ratio effects","authors":"Luohuan Zou , Yu Gong , Dingli Tian , Sizhuo Hao , Jianyu Zhang , Libin Zhao , Ning Hu","doi":"10.1016/j.ijfatigue.2025.109469","DOIUrl":"10.1016/j.ijfatigue.2025.109469","url":null,"abstract":"<div><div>Delamination usually occurs and grows in composite laminates under fatigue loading. The stress ratio is an important factor, while its influence law has no consensus yet. In this paper, to fully investigate the influence of fiber bridging and stress ratio on the fatigue delamination behavior, mode I fatigue delamination tests under two stress ratios (0.1 and 0.5) are conducted. Test results reveal that, the initial and steady-state values of the fatigue R-curve are consistent with those of quasi-static ones, while there are significant differences in the growth stage of fiber bridging. Furthermore, it is found that, the slope and intercept of the d<em>a</em>/d<em>N</em>-<em>G<sub>max</sub></em> curves vary under different stress ratios. A novel four-parameter fatigue model considering fiber bridging and stress ratio effects is proposed. The proposed model is compared with other classical models in literatures using the fatigue data of two stress ratios (0.1 and 0.5). It is found that the proposed model can well characterize fatigue delamination behavior. To further verify the model applicability, fatigue tests under stress ratio of 0.3 are supplemented. The predicted d<em>a</em>/d<em>N</em>-<em>G<sub>max</sub></em> curves by the model and experimental results are compared with a 95% confidence interval, which indicates that the proposed model has good applicability and can provide an effective method for fatigue delamination prediction.</div></div>","PeriodicalId":14112,"journal":{"name":"International Journal of Fatigue","volume":"206 ","pages":"Article 109469"},"PeriodicalIF":6.8,"publicationDate":"2026-05-01","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"145895491","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":2,"RegionCategory":"材料科学","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}
Pub Date : 2026-05-01Epub Date: 2025-12-12DOI: 10.1016/j.ijfatigue.2025.109442
Liheng Tian , Yu Gong , Hao Liu , Sizhuo Hao , Jianyu Zhang , Libin Zhao , Ning Hu
This study proposes a physics-informed machine learning framework to predict the delamination onset life of unidirectional carbon fiber-reinforced plastic laminates under mode I fatigue loading. Traditional methods are limited by the simplified assumptions of physical models or the risks of underfitting caused by purely data-driven models, making it challenging to achieve high-precision predictions under small-sample conditions. By combining experimental data obtained from quasi-static test and fatigue test, this study develops a normalized G-N curve model considering fiber bridging effect, which is incorporated into the loss function of the neural networks to improve predictive robustness and model transparency. Predictions demonstrate that the physics-informed machine learning model exhibits significantly lower prediction deviations in the low-life region (around N = 103) compared to both purely data-driven machine learning model and the proposed physical model, while maintaining stable predictive capability in the high-life region (N > 104). Compared to the pure data-driven model (R2 = 0.8462), the R2 value of the physics-informed machine learning increases to 0.9823, demonstrating the validity of the physics-data fusion approach. This method provides a high-precision solution for fatigue life prediction of unidirectional CFRP laminates, particularly suitable for rapid evaluation in engineering scenarios with insufficient data.
{"title":"A physics-informed machine learning approach for onset life prediction of mode I fatigue delamination in CFRP laminates","authors":"Liheng Tian , Yu Gong , Hao Liu , Sizhuo Hao , Jianyu Zhang , Libin Zhao , Ning Hu","doi":"10.1016/j.ijfatigue.2025.109442","DOIUrl":"10.1016/j.ijfatigue.2025.109442","url":null,"abstract":"<div><div>This study proposes a physics-informed machine learning framework to predict the delamination onset life of unidirectional carbon fiber-reinforced plastic laminates under mode I fatigue loading. Traditional methods are limited by the simplified assumptions of physical models or the risks of underfitting caused by purely data-driven models, making it challenging to achieve high-precision predictions under small-sample conditions. By combining experimental data obtained from quasi-static test and fatigue test, this study develops a normalized <em>G-N</em> curve model considering fiber bridging effect, which is incorporated into the loss function of the neural networks to improve predictive robustness and model transparency. Predictions demonstrate that the physics-informed machine learning model exhibits significantly lower prediction deviations in the low-life region (around <em>N</em> = 10<sup>3</sup>) compared to both purely data-driven machine learning model and the proposed physical model, while maintaining stable predictive capability in the high-life region (<em>N</em> > 10<sup>4</sup>). Compared to the pure data-driven model (R<sup>2</sup> = 0.8462), the R<sup>2</sup> value of the physics-informed machine learning increases to 0.9823, demonstrating the validity of the physics-data fusion approach. This method provides a high-precision solution for fatigue life prediction of unidirectional CFRP laminates, particularly suitable for rapid evaluation in engineering scenarios with insufficient data.</div></div>","PeriodicalId":14112,"journal":{"name":"International Journal of Fatigue","volume":"206 ","pages":"Article 109442"},"PeriodicalIF":6.8,"publicationDate":"2026-05-01","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"145731769","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":2,"RegionCategory":"材料科学","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}
Pub Date : 2026-05-01Epub Date: 2025-12-29DOI: 10.1016/j.ijfatigue.2025.109470
Kangnan Zhu, Jiajun Shi, Anji Wang, Guijun Xian, Chenggao Li
Carbon/glass hybrid fiber reinforced polymer (C/GFRP) tubes, which offer both high performance and cost-effectiveness, are often subjected to the synergistic effects of fatigue and creep during their service life as transportation carriers, which reduces the safety of the structure. This study investigates the tension–tension fatigue behavior of C/GFRP tubes under constant stress ratio at different stress levels. The influence of a hygrothermal environment on fatigue failure modes, fatigue life, and stiffness degradation was examined via laboratory accelerated aging (150 days of immersion in distilled water at 60 °C). The creep displacement evolution was investigated by experimental and analytical means. Finally, a modified fatigue stiffness degradation model accounting for creep effects was proposed based on the creep growth curve. During fatigue loading, the primary load-bearing responsibility gradually shifts from the resin to the fibers as the resin deforms. This transition alters the material’s viscoelastic behavior, evolving from resin-dominated viscoelasticity toward fiber-dominated elasticity. Consequently, the total energy dissipated per loading cycle significantly decreases. Hygrothermal aging alters the failure mode, causing irregular serrated matrix fractures due to interface degradation, and significantly reduces fatigue life. After 150 days of accelerated aging, the fatigue life retention rates of the C/GFRP tubes at stress levels of 0.50, 0.45, 0.40, and 0.38 were 16.3 %, 61.6 %, 57.1 %, and 45.8 %, respectively. Creep effects lead to increased stiffness during fatigue in tubes. The modified stiffness degradation model effectively characterizes the actual stiffness evolution of C/GFRP tubes during fatigue process by separating the cyclic creep.
{"title":"Creep-fatigue interaction and hygrothermal aging effect on the fatigue behavior of carbon/glass hybrid fiber filament-wound tubes","authors":"Kangnan Zhu, Jiajun Shi, Anji Wang, Guijun Xian, Chenggao Li","doi":"10.1016/j.ijfatigue.2025.109470","DOIUrl":"10.1016/j.ijfatigue.2025.109470","url":null,"abstract":"<div><div>Carbon/glass hybrid fiber reinforced polymer (C/GFRP) tubes, which offer both high performance and cost-effectiveness, are often subjected to the synergistic effects of fatigue and creep during their service life as transportation carriers, which reduces the safety of the structure. This study investigates the tension–tension fatigue behavior of C/GFRP tubes under constant stress ratio at different stress levels. The influence of a hygrothermal environment on fatigue failure modes, fatigue life, and stiffness degradation was examined via laboratory accelerated aging (150 days of immersion in distilled water at 60 °C). The creep displacement evolution was investigated by experimental and analytical means. Finally, a modified fatigue stiffness degradation model accounting for creep effects was proposed based on the creep growth curve. During fatigue loading, the primary load-bearing responsibility gradually shifts from the resin to the fibers as the resin deforms. This transition alters the material’s viscoelastic behavior, evolving from resin-dominated viscoelasticity toward fiber-dominated elasticity. Consequently, the total energy dissipated per loading cycle significantly decreases. Hygrothermal aging alters the failure mode, causing irregular serrated matrix fractures due to interface degradation, and significantly reduces fatigue life. After 150 days of accelerated aging, the fatigue life retention rates of the C/GFRP tubes at stress levels of 0.50, 0.45, 0.40, and 0.38 were 16.3 %, 61.6 %, 57.1 %, and 45.8 %, respectively. Creep effects lead to increased stiffness during fatigue in tubes. The modified stiffness degradation model effectively characterizes the actual stiffness evolution of C/GFRP tubes during fatigue process by separating the cyclic creep.</div></div>","PeriodicalId":14112,"journal":{"name":"International Journal of Fatigue","volume":"206 ","pages":"Article 109470"},"PeriodicalIF":6.8,"publicationDate":"2026-05-01","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"145881308","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":2,"RegionCategory":"材料科学","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}
Pub Date : 2026-05-01Epub Date: 2026-01-03DOI: 10.1016/j.ijfatigue.2025.109478
Tian Xu , Huwei Qiu , Wentao Huang , Delong He , Yao Chen , Chong Wang , Qingyuan Wang , Jinbo Bai , Fulin Liu , Yongjie Liu
To meet the application requirements for aero-engine Blisks, very high cycle fatigue (VHCF) behavior of Ti60/TC17 linear friction welded (LFW) joints was investigated at room temperature (RT) and high temperatures (HT). The results indicate that fatigue strength decreases with increasing temperature. Fatigue fractures predominantly occur in the weaker Ti60 base material at all test temperatures. Nearly all fatigue crack initiation sites are characterized by facet morphologies formed through the cleavage of α grains. Specifically, subsurface failures originate from an oversized facet, whereas internal failures arise from facet clusters. Microstructural analysis reveals that cracks primarily nucleate at the α/β phase interface and along slip bands within equiaxed α grains. Notably, high temperature significantly influences the crack initiation mechanism, causing a transition in facet formation from basal slip dominance at RT to synergistic basal and prismatic slip at HT. Furthermore, for subsurface crack initiation at HT, the synergistic effect of prolonged high-temperature exposure and dislocation-assisted oxygen diffusion facilitates brittle oxide formation at the crack tips, thereby accelerating fatigue failure.
{"title":"Crack initiation mechanisms of linear friction welded dissimilar Ti60/TC17 joint in very high cycle fatigue regime at different temperatures","authors":"Tian Xu , Huwei Qiu , Wentao Huang , Delong He , Yao Chen , Chong Wang , Qingyuan Wang , Jinbo Bai , Fulin Liu , Yongjie Liu","doi":"10.1016/j.ijfatigue.2025.109478","DOIUrl":"10.1016/j.ijfatigue.2025.109478","url":null,"abstract":"<div><div>To meet the application requirements for aero-engine Blisks, very high cycle fatigue (VHCF) behavior of Ti60/TC17 linear friction welded (LFW) joints was investigated at room temperature (RT) and high temperatures (HT). The results indicate that fatigue strength decreases with increasing temperature. Fatigue fractures predominantly occur in the weaker Ti60 base material at all test temperatures. Nearly all fatigue crack initiation sites are characterized by facet morphologies formed through the cleavage of α grains. Specifically, subsurface failures originate from an oversized facet, whereas internal failures arise from facet clusters. Microstructural analysis reveals that cracks primarily nucleate at the α/β phase interface and along slip bands within equiaxed α grains. Notably, high temperature significantly influences the crack initiation mechanism, causing a transition in facet formation from basal slip dominance at RT to synergistic basal and prismatic slip at HT. Furthermore, for subsurface crack initiation at HT, the synergistic effect of prolonged high-temperature exposure and dislocation-assisted oxygen diffusion facilitates brittle oxide formation at the crack tips, thereby accelerating fatigue failure.</div></div>","PeriodicalId":14112,"journal":{"name":"International Journal of Fatigue","volume":"206 ","pages":"Article 109478"},"PeriodicalIF":6.8,"publicationDate":"2026-05-01","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"145894531","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":2,"RegionCategory":"材料科学","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}
Pub Date : 2026-05-01Epub Date: 2025-12-31DOI: 10.1016/j.ijfatigue.2025.109474
Manjiang Yu , Ye Wang , Fangli Duan , Chaofeng Lü
The service failure evaluation of the damaged rail can provide theoretical guidance for the routine maintenance of existing railway lines. In this work, U71MnG rail that failed under traffic loading is selected to investigate the fatigue damage mechanism of pearlitic rail steel. In addition to the conventional surface crack, the branch crack characterized by the ‘Y’ at the subsurface inclusion is also observed. Among them, the leading crack aligned with the rolling direction forms and propagates first, followed by the two trailing cracks propagating in the reverse rolling direction. Driven by the plastic ratcheting at the surface layer and the deeper bending deformation, these two trailing cracks may converge with other adjacent leading cracks, resulting in the formation of the spalling pit hundreds of μm deep. Moreover, the sliding friction in the transverse direction of the rail promotes the formation of wavy pearlite, which reduces the fracture toughness of the outside rail. In the predictive maintenance of rails, it is essential to promptly identify and address potential hazards of surface spalling and transverse fracture.
{"title":"Fatigue damage development and mechanical property degradation of pearlitic rail steel under loaded traffic","authors":"Manjiang Yu , Ye Wang , Fangli Duan , Chaofeng Lü","doi":"10.1016/j.ijfatigue.2025.109474","DOIUrl":"10.1016/j.ijfatigue.2025.109474","url":null,"abstract":"<div><div>The service failure evaluation of the damaged rail can provide theoretical guidance for the routine maintenance of existing railway lines. In this work, U71MnG rail that failed under traffic loading is selected to investigate the fatigue damage mechanism of pearlitic rail steel. In addition to the conventional surface crack, the branch crack characterized by the ‘Y’ at the subsurface inclusion is also observed. Among them, the leading crack aligned with the rolling direction forms and propagates first, followed by the two trailing cracks propagating in the reverse rolling direction. Driven by the plastic ratcheting at the surface layer and the deeper bending deformation, these two trailing cracks may converge with other adjacent leading cracks, resulting in the formation of the spalling pit hundreds of μm deep. Moreover, the sliding friction in the transverse direction of the rail promotes the formation of wavy pearlite, which reduces the fracture toughness of the outside rail. In the predictive maintenance of rails, it is essential to promptly identify and address potential hazards of surface spalling and transverse fracture.</div></div>","PeriodicalId":14112,"journal":{"name":"International Journal of Fatigue","volume":"206 ","pages":"Article 109474"},"PeriodicalIF":6.8,"publicationDate":"2026-05-01","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"145894535","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":2,"RegionCategory":"材料科学","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}
Pub Date : 2026-05-01Epub Date: 2025-12-13DOI: 10.1016/j.ijfatigue.2025.109438
Pinduo Zhu , Feng Dai , Zelin Yan , Dingran Song , Hao Tan , Pan Zhou
In major underground engineering projects, rocks are frequently subjected to cyclic loadings, causing various engineering disasters. Therefore, accurately predicting the rock fatigue behaviors is important for rock engineering safety. Current methodologies like laboratory testing and numerical analysis all have limits like excessive test costs, limited applicability etc., demonstrating insufficient capability in predicting rock fatigue life. While artificial intelligence (AI) has brought new opportunities to such researches. In this study, six machine learning (ML) models including gradient boosting machine (GBM), extreme gradient boosting (XGBoost), along with other four models combined with bayesian optimization (BO) method are used to predict rock fatigue life. Initially, a dataset with 213 rock fatigue life experimental data is built with 8 parameters including maximum cyclic stress and its ratio, cyclic loading amplitude and its ratio, uniaxial compressive strength (UCS), frequency, waveform and rock type. Meanwhile, three error correlation coefficients including the mean absolute error (MAE), the root mean square error (RMSE) and the coefficient of determination (R2) are selected. The results indicate that prediction accuracy and feature importance are significantly influenced by algorithm types. Specifically, XGBoost and GBM demonstrate the superior predictive accuracy than other algorithms, with mean R2 values of 0.9 and 0.88. For the eight parameters, maximum cyclic stress and its ratio exert substantial influence with mean importance factors of 0.22 and 0.3, then follows cyclic loading amplitude (0.14) and UCS (0.17). SHapley Additive exPlanations analysis indicates that ML models can correctly capture the complex relationship between major parameters and fatigue life. This study shows that, after carefully considering the algorithms and input parameters, ML can effectively predict the fatigue life of rocks.
{"title":"Fatigue life prediction of rocks under cyclic uniaxial compression using XGBoost and GBM based on Bayesian optimization","authors":"Pinduo Zhu , Feng Dai , Zelin Yan , Dingran Song , Hao Tan , Pan Zhou","doi":"10.1016/j.ijfatigue.2025.109438","DOIUrl":"10.1016/j.ijfatigue.2025.109438","url":null,"abstract":"<div><div>In major underground engineering projects, rocks are frequently subjected to cyclic loadings, causing various engineering disasters. Therefore, accurately predicting the rock fatigue behaviors is important for rock engineering safety. Current methodologies like laboratory testing and numerical analysis all have limits like excessive test costs, limited applicability etc., demonstrating insufficient capability in predicting rock fatigue life. While artificial intelligence (AI) has brought new opportunities to such researches. In this study, six machine learning (ML) models including gradient boosting machine (GBM), extreme gradient boosting (XGBoost), along with other four models combined with bayesian optimization (BO) method are used to predict rock fatigue life. Initially, a dataset with 213 rock fatigue life experimental data is built with 8 parameters including maximum cyclic stress and its ratio, cyclic loading amplitude and its ratio, uniaxial compressive strength (UCS), frequency, waveform and rock type. Meanwhile, three error correlation coefficients including the mean absolute error (MAE), the root mean square error (RMSE) and the coefficient of determination (R<sup>2</sup>) are selected. The results indicate that prediction accuracy and feature importance are significantly influenced by algorithm types. Specifically, XGBoost and GBM demonstrate the superior predictive accuracy than other algorithms, with mean R<sup>2</sup> values of 0.9 and 0.88. For the eight parameters, maximum cyclic stress and its ratio exert substantial influence with mean importance factors of 0.22 and 0.3, then follows cyclic loading amplitude (0.14) and UCS (0.17). SHapley Additive exPlanations analysis indicates that ML models can correctly capture the complex relationship between major parameters and fatigue life. This study shows that, after carefully considering the algorithms and input parameters, ML can effectively predict the fatigue life of rocks.</div></div>","PeriodicalId":14112,"journal":{"name":"International Journal of Fatigue","volume":"206 ","pages":"Article 109438"},"PeriodicalIF":6.8,"publicationDate":"2026-05-01","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"145749975","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":2,"RegionCategory":"材料科学","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}
Pub Date : 2026-05-01Epub Date: 2025-12-18DOI: 10.1016/j.ijfatigue.2025.109451
Lu Zhang
The estimation of fatigue life is vital for ensuring structural durability and safety in engineering applications. To address the limitations of current nonlinear fatigue cumulative damage models, which often overlook the interplay between load sequence and material properties, a novel nonlinear fatigue damage accumulation model is developed in this work. By comprehensively reviewing and analyzing nonlinear action coefficients in prior enhanced models, a new function for the action coefficient is formulated, incorporating three key elements: adjacent stress ratio, material S-N curve slope, and equivalent fatigue damage. The key parameters of the model are determined based on two-level stress test data of multiple materials. Furthermore, using fatigue test data of various metal materials under two-level to five-level stress spectra, the prediction accuracy of the new model is compared and verified against 8 typical models. The results show that the new model exhibits better adaptability and prediction accuracy under different stress levels and load sequences, demonstrating good potential for engineering applications.
{"title":"Novel nonlinear fatigue damage model based on dynamic action coefficient with three factors","authors":"Lu Zhang","doi":"10.1016/j.ijfatigue.2025.109451","DOIUrl":"10.1016/j.ijfatigue.2025.109451","url":null,"abstract":"<div><div>The estimation of fatigue life is vital for ensuring structural durability and safety in engineering applications. To address the limitations of current nonlinear fatigue cumulative damage models, which often overlook the interplay between load sequence and material properties, a novel nonlinear fatigue damage accumulation model is developed in this work. By comprehensively reviewing and analyzing nonlinear action coefficients in prior enhanced models, a new function for the action coefficient is formulated, incorporating three key elements: adjacent stress ratio, material S-N curve slope, and equivalent fatigue damage. The key parameters of the model are determined based on two-level stress test data of multiple materials. Furthermore, using fatigue test data of various metal materials under two-level to five-level stress spectra, the prediction accuracy of the new model is compared and verified against 8 typical models. The results show that the new model exhibits better adaptability and prediction accuracy under different stress levels and load sequences, demonstrating good potential for engineering applications.</div></div>","PeriodicalId":14112,"journal":{"name":"International Journal of Fatigue","volume":"206 ","pages":"Article 109451"},"PeriodicalIF":6.8,"publicationDate":"2026-05-01","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"145784903","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":2,"RegionCategory":"材料科学","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}
Pub Date : 2026-05-01Epub Date: 2025-12-13DOI: 10.1016/j.ijfatigue.2025.109432
Jumei Lu , Jianhui Liu , Youtang Li , Juntai Hu , Shengchuan Wu
This study explores the multiaxial fatigue performance of ER8C wheel steel, with a focus on the influence of phase shift on its multiaxial fatigue behavior. The test results show that as the phase shift increases, the fatigue life of the ER8C wheel steel exhibits a significant growth trend. A fracture analysis was conducted on it, revealing that the phase shift is negatively correlated with the number of crack sources and the area of instantaneous fracture zone. Furthermore, the effect of the phase shift is the first discussed from the perspective of the stress/strain control mode in fatigue tests, aiming to incorporate the direct impact of the phase shift into the damage coefficient considering non-proportional loading paths. Finally, by combining the pure tensile and pure torsion fatigue curves, a set of multiaxial fatigue life prediction models based on MWCM (Modified Wöhler Curve Method) with good scalability is established. The validity of these models is verified using five types of multiaxial materials respectively. In particular, for the method where phase shift is positively correlated with fatigue life, the test results of ER8C wheel steel are used for verification, and the fatigue life of ER8C wheel steel is estimated.
{"title":"Multiaxial fatigue performance of ER8C wheel steel and fatigue life prediction based on MWCM","authors":"Jumei Lu , Jianhui Liu , Youtang Li , Juntai Hu , Shengchuan Wu","doi":"10.1016/j.ijfatigue.2025.109432","DOIUrl":"10.1016/j.ijfatigue.2025.109432","url":null,"abstract":"<div><div>This study explores the multiaxial fatigue performance of ER8C wheel steel, with a focus on the influence of phase shift on its multiaxial fatigue behavior. The test results show that as the phase shift increases, the fatigue life of the ER8C wheel steel exhibits a significant growth trend. A fracture analysis was conducted on it, revealing that the phase shift is negatively correlated with the number of crack sources and the area of instantaneous fracture zone. Furthermore, the effect of the phase shift is the first discussed from the perspective of the stress/strain control mode in fatigue tests, aiming to incorporate the direct impact of the phase shift into the damage coefficient considering non-proportional loading paths. Finally, by combining the pure tensile and pure torsion fatigue curves, a set of multiaxial fatigue life prediction models based on MWCM (Modified Wöhler Curve Method) with good scalability is established. The validity of these models is verified using five types of multiaxial materials respectively. In particular, for the method where phase shift is positively correlated with fatigue life, the test results of ER8C wheel steel are used for verification, and the fatigue life of ER8C wheel steel is estimated.</div></div>","PeriodicalId":14112,"journal":{"name":"International Journal of Fatigue","volume":"206 ","pages":"Article 109432"},"PeriodicalIF":6.8,"publicationDate":"2026-05-01","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"145753416","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":2,"RegionCategory":"材料科学","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}
Pub Date : 2026-05-01Epub Date: 2026-01-06DOI: 10.1016/j.ijfatigue.2026.109483
Matheus Garcia do Vale , Thiago Roberto Felisardo Cavalcante , Gualter Silva Pereira , Julián Arnaldo Ávila Díaz , José Luiz Boldrini , Marco Lúcio Bittencourt
Fatigue crack growth critically influences the lifespan of structural components in high-demanding engineering applications. Despite advances in phase-field fracture models, cycle-by-cycle simulations remain computationally prohibitive and often rely on extrapolation techniques. This work introduces a novel energy-based fatigue degradation evolution equation within a phase-field framework, enabling direct recovery of the Paris-law behavior without the need for explicit cycle-jumping algorithms. We implement a staggered solution scheme and employ a constant- loading procedure to compute crack growth rates in selected stress intensity ranges. The proposed strategy is effective in calibrating simulation parameters, resulting in a reduction of up to 97% in simulated cycles. Additionally, we utilize an automatic crack length measurement algorithm based on the A* pathfinder heuristic, which minimizes user intervention and mesh dependence. Validation with experimental data for the WE43 and AA7050 alloys shows excellent agreement in the Paris plots, while reducing computational costs. The proposed methodology offers a robust and efficient tool for material characterization and fatigue analysis in brittle-to-ductile materials.
{"title":"Fitting of an FCG test of the WE43C and AA7050 alloys using two phase-fields and a Constant-ΔK approach","authors":"Matheus Garcia do Vale , Thiago Roberto Felisardo Cavalcante , Gualter Silva Pereira , Julián Arnaldo Ávila Díaz , José Luiz Boldrini , Marco Lúcio Bittencourt","doi":"10.1016/j.ijfatigue.2026.109483","DOIUrl":"10.1016/j.ijfatigue.2026.109483","url":null,"abstract":"<div><div>Fatigue crack growth critically influences the lifespan of structural components in high-demanding engineering applications. Despite advances in phase-field fracture models, cycle-by-cycle simulations remain computationally prohibitive and often rely on extrapolation techniques. This work introduces a novel energy-based fatigue degradation evolution equation within a phase-field framework, enabling direct recovery of the Paris-law behavior without the need for explicit cycle-jumping algorithms. We implement a staggered solution scheme and employ a constant-<span><math><mrow><mi>Δ</mi><mi>K</mi></mrow></math></span> loading procedure to compute crack growth rates in selected stress intensity ranges. The proposed strategy is effective in calibrating simulation parameters, resulting in a reduction of up to 97% in simulated cycles. Additionally, we utilize an automatic crack length measurement algorithm based on the A* pathfinder heuristic, which minimizes user intervention and mesh dependence. Validation with experimental data for the WE43 and AA7050 alloys shows excellent agreement in the Paris plots, while reducing computational costs. The proposed methodology offers a robust and efficient tool for material characterization and fatigue analysis in brittle-to-ductile materials.</div></div>","PeriodicalId":14112,"journal":{"name":"International Journal of Fatigue","volume":"206 ","pages":"Article 109483"},"PeriodicalIF":6.8,"publicationDate":"2026-05-01","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"145922150","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":2,"RegionCategory":"材料科学","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