Tshibangu, Theophile Mbuyi, Ngoma, Guyh Dituba et Gagnon, Martin.
2025.
« Estimating the fatigue reliability of hydraulic turbine runner blades using frailty models and censored maintenance data ».
In Proceedings of the CSME-CFDSC-CSR 2025 International Congress (Montreal, QC, Canada, May 25-28, 2025)
Coll. « Progress in Canadian Mechanical Engineering », vol. 8.
Prévisualisation |
PDF
596 - Estimating the fatigue reliability.pdf - Version publiée Licence d'utilisation : Tous les droits réservés aux détenteurs du droit d'auteur. Télécharger (660kB) | Prévisualisation |
Résumé
Hydropower plays a critical role in the global energy transition. The hydraulic turbine runner, a key component of the turbine-generator unit, is susceptible to two forms of damage: cavita-tion and fatigue. Despite continuous monitoring systems equipped with sensors to measure key operating parameters, these systems are unable to detect cracks in turbine runner blades. Consequently, such cracks are often unexpectedly discovered during planned mainte-nance. Turbine blade cracking is a rare event leading to right-censored observations during inspections. Hence survival frailty models can be used to account for both censored data and variability observed among individuals within the same family. Maintenance data analyzed in this study are collected from inspections and repairs of 18 turbine runners. By incorporating the frailty effect to account for heterogeneity within a family, the results suggest that some turbines exhibit similar reliability patterns. Model fit is assessed using the chi-square test, confirming acceptability at a 95% significance level. Additionally, the root mean squared error is used to evaluate model accuracy and to compare the cumulative hazard function to a non-parametric Nelson-Aalen estimator.The confidence interval for the cumulative hazard function is evaluated at a 95% confidence level, using the delta method to quantify the uncertainty resulting from the estimation of model parameters. Despite the limited data, the frailty model outperforms the Nelson-Aalen estimator. Optimizing model parameters with a robust algorithm using right-censored data suggests that crack events in hydraulic turbine runner blades appear to follow a frailty model.
| Type de document: | Compte rendu de conférence |
|---|---|
| Éditeurs: | Éditeurs ORCID Hof, Lucas A. NON SPÉCIFIÉ Di Labbio, Giuseppe NON SPÉCIFIÉ Tahan, Antoine NON SPÉCIFIÉ Sanjosé, Marlène NON SPÉCIFIÉ Lalonde, Sébastien NON SPÉCIFIÉ Demarquette, Nicole R. NON SPÉCIFIÉ |
| Date de dépôt: | 18 déc. 2025 15:33 |
| Dernière modification: | 18 déc. 2025 15:33 |
| URI: | https://espace2.etsmtl.ca/id/eprint/32517 |
Actions (Authentification requise)
 |
Dernière vérification avant le dépôt |