Page 82 - IJAMD-1-3

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

To mitigate the effects of erosion and thermal stresses, patterns and structural failure in turbine blades. 16,19 These
advanced surface coating theories are often applied to turbine simulations are invaluable for understanding how fluid
blades. These coatings act as a protective barrier between dynamics and material behavior interact under real-world
the blade material and the harsh operational environment. conditions.
Shin and Hamed investigated the effectiveness of TBCs CFD models simulate the complex interactions
77
in reducing erosion and oxidation. TBCs are typically between gas flows and turbine blades, particularly focusing
made of ceramics, which are highly resistant to heat and on how high-velocity flows and solid particles contribute
erosion. However, as erosion wears away these coatings, to erosion. Hamed et al. applied the Navier–Stokes
74
the underlying blade material becomes exposed, leading to equations 32,34,38 to predict airflow patterns around turbine
accelerated degradation. The diffusion theory 69,70 explains blades. By incorporating Eulerian–Lagrangian models, 14,19
how heat and oxygen diffuse through these coatings over
time, contributing to their eventual breakdown. they were able to simulate particle trajectories and the
subsequent impacts on the blade surface. Fluid-structure
This section explains the long-term effects of erosion interaction technique allows for precise identification
on turbine blades. As the blades experience repeated of erosion hotspots, particularly on the leading edges of
cycles of erosion, high temperatures, and mechanical blades where particles tend to accumulate and cause the
loads, the damage accumulates, eventually leading to most damage.
failure. 21,39,45,57,71,72 This understanding is crucial for
developing maintenance strategies that address the While CFD focuses on fluid dynamics, FEA models
cumulative effects of erosion and fatigue. the structural response of the blade material to the forces
generated by particle impacts. Taherkhani et al. used
56
The Paris–Erdogan law for crack growth, combined FEA to simulate how repeated particle collisions create
70
with erosion models, can describe how erosion-induced stress concentrations and lead to material deformation
fatigue leads to material failure. The Paris–Erdogan law 70,71 and fatigue. von Mises stress theory 56,57,60 was employed
is used to model the propagation of cracks in materials to understand how stress is distributed across the blade
subjected to cyclic loading, which is typical in gas turbines surface. By combining CFD and FEA, Branco et al.
58
due to operational cycles. created an integrated model that allowed them to simulate
da
The rate of crack growth per cycle, , is given by not only where particles would hit the blade but also how
Equation III: dN these impacts would lead to long-term structural damage.

da =⋅ K m (III) A growing trend in computational simulations is
C ∆
dN the use of multi-scale modeling, which links microscale
erosion events, such as pitting and surface roughness,
where a is the crack length; N is the number of loading to macroscale turbine blade performance. Zhu et al.
59
cycles; C and m are material-dependent constants; and demonstrated that micro-scale damage caused by erosion
ΔK is the stress intensity factor range (difference between can lead to increased drag, reduced aerodynamic efficiency,
maximum and minimum stress intensities during a and eventually large-scale failure. These models integrate
cycle). The stress intensity factor range, ΔK, is given by CFD data on particle impacts with FEA data on material
78
Equation IV: behavior to create a comprehensive simulation that spans
multiple scales of analysis.
∆K = K max − K min = ∆ ⋅σ π a (IV) Due to the computational cost of running detailed
where Δσ is the stress range (difference between CFD and FEA simulations, researchers are increasingly
maximum and minimum stress in a cycle); and a is the using ML to develop surrogate models that approximate
77
crack length. This equation links erosion-induced cracks, these simulations. Shin and Hamed trained NNs on CFD
which serve as stress concentrators, with the fatigue life of and FEA data to predict erosion patterns and material
the blade. Repeated erosion weakens the blade’s surface, stress in real time. Surrogate models and ML approach
increasing the rate of crack propagation under cyclic significantly reduces the time and computational power
thermal and mechanical loads, leading to failure. required for full-scale simulations, enabling more frequent
and detailed analyses of turbine blade performance under
3.3. Computational modeling and simulation erosive conditions.
theories
The computational modeling and simulation theories
The third pillar focuses on computational methods – section emphasizes the importance of combining fluid
specifically CFD and FEA – to model and predict erosion dynamics and structural analysis to understand how

Volume 1 Issue 3 (2024) 76 doi: 10.36922/ijamd.5188