Artificial Intelligence-Based Stress Prediction for Rotor Blisk in Gas Turbine Engines

Öz

Gas turbine engines are critical components in aerospace, power generation, and industrial applications, consisting of complex rotating and stationary parts subject to extreme mechanical, thermal, and aerodynamic loads. A key component in modern gas turbines is the rotor blisk, which combines the blades and disk into a single unit. Due to its complex geometry and harsh operating conditions, the rotor blisk experiences significant mechanical stresses that must be accurately calculated to ensure reliability, safety, and optimal performance. Traditional methods, such as finite element analysis (FEA), are widely used to calculate stress distributions under various loading conditions. However, FEA is computationally expensive, especially when analyzing multiple scenarios for different operating conditions. This computational cost can become a bottleneck in iterative design studies and real-time decision making. To address this challenge, this study proposes a novel approach that uses deep learning to predict stresses in rotor blisks under varying loads. A deep neural network (DNN) was trained on FEA-generated stress data to learn the relationships between input parameters and resulting stress distributions. The AI-based model was validated using unseen load scenarios for radial, axial, and tangential stress distributions and maximum-minimum stress results, with a maximum deviation of 6% to 15% from FEA results. In addition, the Artificial Intelligence (AI) approach reduced the computational cost by 13,000 times faster than FEA by predicting results instead of solving complex equations. The AI approach enables rapid stress predictions and facilitates real-time design iteration and optimization. These results highlight the transformative potential of AI in engineering simulation, enabling faster, more efficient structural assessments and advancing the optimization of gas turbine components in the aerospace and energy industries.

Anahtar Kelimeler

• Artificial Intelligence
• Deep Learning
• Gas Turbine Engine
• Rotor Blisk
• Finite Element Analysis

Gaz Türbinli Motorlarda Rotor Bliskleri için Yapay Zeka Tabanlı Stres Tahmini

Öz

Gaz türbinli motorlar; havacılık, enerji üretimi ve endüstriyel uygulamalarda kritik bileşenler olup, aşırı mekanik, termal ve aerodinamik yüklere maruz kalan karmaşık döner ve sabit parçalardan oluşmaktadır. Modern gaz türbinlerinin temel bileşenlerinden biri, kanatları ve diski tek bir bütün halinde birleştiren rotor blisktir. Karmaşık geometrisi ve zorlu çalışma koşulları nedeniyle rotor blisk, güvenilirliğin, emniyetin ve optimal performansın sağlanabilmesi için doğru bir şekilde hesaplanması gereken önemli mekanik gerilmelere maruz kalmaktadır. Sonlu elemanlar analizi (SEA) gibi geleneksel yöntemler, farklı yükleme koşullarında gerilme dağılımlarını hesaplamak için yaygın şekilde kullanılmaktadır. Ancak, SEA özellikle farklı çalışma koşulları için çoklu senaryoların analizinde hesaplama açısından maliyetli olup, bu hesaplama yükü tekrarlamalı tasarım çalışmalarında ve gerçek zamanlı karar vermede bir darboğaz hâline gelebilmektedir. Bu zorluğun üstesinden gelmek amacıyla, bu çalışma rotor blisklerde farklı yükler altında gerilmeleri tahmin etmek için derin öğrenme kullanan yeni bir yaklaşım önermektedir. Bir derin sinir ağı (DSA), giriş parametreleri ile ortaya çıkan gerilme dağılımları arasındaki ilişkileri öğrenebilmek için SEA tarafından üretilmiş gerilme verileri üzerinde eğitilmiştir. Yapay zeka tabanlı model, radyal, eksenel ve teğetsel gerilme dağılımları ile maksimum-minimum gerilme sonuçları için görülmemiş yük senaryoları kullanılarak doğrulanmış ve SEA sonuçlarına kıyasla %6 ila %15 arasında maksimum sapma göstermiştir. Ayrıca, yapay zeka yaklaşımı karmaşık denklemleri çözmek yerine sonuçları tahmin ederek SEA’ya kıyasla hesaplama maliyetini 13.000 kat azaltmıştır. Yapay zeka yaklaşımı, hızlı gerilme tahminleri yapılmasını mümkün kılmakta ve gerçek zamanlı tasarım yinelemelerini ve optimizasyonu kolaylaştırmaktadır. Bu sonuçlar, mühendislik simülasyonunda yapay zekânın dönüştürücü potansiyelini vurgulamakta, daha hızlı ve daha verimli yapısal değerlendirmeleri mümkün kılmakta ve havacılık ile enerji endüstrilerinde gaz türbini bileşenlerinin optimizasyonunu ilerletmektedir.

Anahtar Kelimeler

• Yapay Zeka
• Derin Öğrenme
• Gaz Türbinli Motor
• Rotor Blisk
• Sonlu Elemanlar Analizi

Kaynakça

1. Al-Mahasneh, A. J., Anavatti, S. G., & Garratt, M. A. (2018). The development of neural networks applications from perceptron to deep learning. International Journal of Computer Theory and Engineering, 10(1), 23–28.
2. ANSYS, Inc. (2024). ANSYS Mechanical 24.2 manual. ANSYS, Inc.
3. Bandini, A., Cascino, A., Meli, E., Pinelli, L., & Marconcini, M. (2024). Improving aeromechanical performance of compressor rotor blisk with topology optimization. Energies, 17(8), 1883. https://doi.org/10.3390/en17081883
4. Bunyan, S. T., Khan, Z. H., Al Haddad, L. A., Dhahad, H. A., Al Karkhi, M. I., Ogaili, A. A. F., & Al Sharify, Z. T. (2025). Intelligent thermal condition monitoring for predictive maintenance of gas turbines using machine learning. Machines, 13(5), 401. https://doi.org/10.3390/machines13050401
5. Chen, T., Wang, Z., & Liu, S. (2022). Fault diagnosis using extreme learning machines with single hidden layer for gas turbines. Journal of Vibration and Acoustics, 144(2), 021004. https://doi.org/10.1115/1.4051361
6. Elhefny, A., & Megahed, M. (2018). Design and life estimation of blisk in gas turbines. International Research Journal of Engineering and Technology, 5(2), 2312–2317. https://www.irjet.net/archives/V5/i2/IRJET-V5I2231.pdf
7. Fei, C. W., Han, Y. J., Wen, J. R., Li, C., Han, L., & Choy, Y. S. (2024). Deep learning-based modeling method for probabilistic LCF life prediction of turbine blisk. Propulsion and Power Research, 13(1), 12–25. https://doi.org/10.1016/j.jppr.2023.08.005
8. Guo, W., Li, J., & Zhao, Y. (2021). Hybrid temporal convolutional network–autoencoder for fault detection in gas turbines. Mechanical Systems and Signal Processing, 150, 107294. https://doi.org/10.1016/j.ymssp.2021.107294

9. Joachim, A., Pérez, E., & Reitz, S. (2025). Reduced order modeling of unsteady pressure on turbine rotor blades using deep learning. arXiv Preprint, arXiv:2503.00013. https://arxiv.org/abs/2503.00013
10. Kumar, B. V. R. R. (2013). A review on blisk technology. International Journal of Innovative Research in Science, Engineering and Technology, 2(5).
11. Li, P., Xu, Y., & Zheng, W. (2024). Reinforced symbolic learning with logical constraints for predicting turbine blade fatigue life. arXiv Preprint, arXiv:2412.03580. https://arxiv.org/abs/2412.03580
12. Li, X., Chen, Z., & Zhao, H. (2024). Review of machine learning methods in turbine cooling optimization. Energies, 17(13), 3177. https://doi.org/10.3390/en17133177
13. Liu, Y., Zhang, M., & Guo, W. (2022). Early-stage fault diagnosis of gas turbines using dynamic neural networks. Energy, 240, 122652. https://doi.org/10.1016/j.energy.2021.122652
14. Mane, S., Pawar, S., & Minde, S. (2023). Gas turbine and its applications. International Journal of Advanced Engineering Research and Science.
15. Melina, M., Sukono, H., Napitupulu, H., & Mohamed, N. (2023). A conceptual model of investment-risk prediction in the stock market using extreme value theory with machine learning: A semisystematic literature review. Risks, 11(60). https://doi.org/10.3390/risks11030060
16. Plevris, V., & Papazafeiropoulos, G. (2024). AI in structural health monitoring for infrastructure maintenance and safety. Infrastructures, 9(12), 225. https://doi.org/10.3390/infrastructures9120225
17. Shivaditya, M. V., Alves, J., Bugiotti, F., & Magoules, F. (2022). Graph neural network-based surrogate models for finite element analysis. arXiv Preprint, arXiv:2211.09373. https://arxiv.org/abs/2211.09373
18. Wang, H., Li, S., & Zhang, X. (2022). Ensemble learning for temperature distribution prediction of turbine blisk. Computational Materials Science, 202, 111040. https://doi.org/10.1016/j.commatsci.2021.111040
19. Zhang, X., de Sturler, E., & Paulino, G. H. (2016). Stochastic sampling for structural topology optimization with many load cases: Density-based and ground structure approaches. arXiv Preprint, arXiv:1609.03099. https://arxiv.org/abs/1609.03099
20. Zhao, Y., Guo, W., Li, J., & Wang, Z. (2021). Artificial intelligence in condition assessment and fault detection for gas turbines: A review. Journal of Turbomachinery, 143(10), 101001. https://doi.org/10.1115/1.4053086
21. Zhou, L., Zhang, J., & Liu, Q. (2019). AI-driven topology optimization of compressor rotor blisk for improved aeromechanical performance. Structural and Multidisciplinary Optimization, 59(6), 2171–2185. https://doi.org/10.1007/s00158-019-02348-3
22. Zhu, X., Zhang, W., Li, Y., & Li, L. (2022). Deep learning-based prediction of low cycle fatigue life for turbine blades in gas turbines. Journal of Engineering for Gas Turbines and Power, 144(3), 031001. https://doi.org/10.1115/1.4051232