Page 89 - IJAMD-1-3

International Journal of AI for
Materials and Design
Review of gas turbine blade failures by erosion

Figure 1. Flow diagram of the entire CFD-FEA process
Abbreviations: CFD: Computational fluid dynamics; FEA: Finite element analysis.

models that approximate the results of full-scale simulations. cyclic loading. In the context of gas turbine blades, repeated
These models significantly reduce computational time particle impacts create localized stress concentrators, such
while maintaining accuracy. 33,34,37 By training NNs on as microcracks and pits, which propagate over time due
large datasets generated from CFD and FEA simulations, to the cyclic nature of operational stresses. This crack
surrogate models can predict erosion patterns and growth eventually compromises the structural integrity
structural responses in near real-time. Poursaeidi et al. of the blade, leading to failure. The Paris Law provides
61
used CNNs to predict particle impact locations and erosion a mathematical framework to predict this progressive
rates based on input parameters such as flow velocity, damage as presented in Equation III.
particle size, and blade geometry. These models can predict In turbine blades, the repeated impacts from high-
erosion hot spots without requiring the full CFD solution, velocity particles act as cyclic loads that initiate and
drastically speeding up the design optimization process. In propagate cracks. These impacts create small pits and
our framework, ML-based surrogate models can be used to scratches on the blade surface, which serve as nucleation
complement CFD and FEA simulations, providing a fast and sites for cracks. As the turbine operates, thermal and
efficient tool for predicting erosion damage and structural mechanical stresses combine with these impacts to
degradation in real time. 55,56 This capability is critical for cyclically load the blade material, causing the cracks
adaptive design and real-time monitoring of turbine blades, to grow incrementally with each cycle. The Paris Law
improving the overall reliability of gas turbine systems. allows engineers to estimate the rate of crack growth by
The Paris Law is a widely used model in material fatigue correlating it with the stress intensity range (ΔK), which
theory to describe the rate at which a crack grows under depends on the material properties and loading conditions.

Volume 1 Issue 3 (2024) 83 doi: 10.36922/ijamd.5188