Page 79 - IJAMD-1-3

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

Table 8. Computational models for erosion prediction
Model Application Strengths Limitations
CFD Predicts particle trajectories Provides detailed insights into Computationally expensive,
and flow patterns high-velocity environments limited by turbulence models
FEA Simulates structural response Models stress distribution and Requires accurate input data
to particle impacts crack propagation from CFD
Eulerian–Lagrangian Tracks particle motion and Combines fluid dynamics and Accuracy depends on mesh
Framework impingement locations particle interactions quality and particle assumptions
ML Predicts erosion patterns Allows rapid predictions, Requires extensive and accurate
based on prior data reduces reliance on training datasets
computational resources
Abbreviations: CFD: Computational fluid dynamics; FEA: Finite element analysis; ML: Machine learning.

trajectories but also the resulting stresses on the blade 2.2.4. Computational challenges and future directions
surface, leading to a more accurate prediction of material Despite the significant advancements in both CFD
fatigue and failure due to erosion.
and FEA, several challenges remain. CFD simulations,
2.2.3. Integrated CFD and FEA approach especially those using LES and DNS, are computationally
expensive and require substantial processing power. FEA
61
The combination of CFD and FEA provides a more models that account for complex material behaviors, such
holistic approach to understanding gas turbine blade as creep and fatigue under high temperatures, also demand
erosion. CFD offers insights into the flow dynamics and considerable computational resources. 52,53,62 Moreover,
particle behavior, while FEA helps analyze the material’s fully coupling CFD and FEA for real-time analysis is
response to erosion-induced stresses. 49,60 This integration still an area of ongoing research due to the difficulty of
is particularly beneficial when investigating how erosion synchronizing fluid dynamics and material responses in a
weakens the structural integrity of blades over time, as well single simulation. 63
as how changes in blade geometry affect erosion patterns.
However, the integration of ML into CFD and FEA
50
Rezamand et al. applied an integrated CFD-FEA
framework to study the long-term effects of particle has the potential to overcome these challenges. Olabi
et al. explored the use of ML-based surrogate models
53
impingement on turbine blades. Their approach involved to approximate CFD and FEA simulations. By training
using CFD to simulate particle flows and impact angles, neural networks (NNs) on existing CFD and FEA data,
followed by FEA to assess how these impacts translated they were able to generate accurate predictions of fluid-
into material degradation. Their study revealed that
blade regions with complex geometries, such as the structure interactions at a fraction of the computational
leading and trailing edges, were the most vulnerable cost. This approach opens the door to real-time erosion
to erosion and subsequent structural failure. By monitoring and adaptive turbine blade designs that
combining CFD and FEA, they were able to propose respond dynamically to changing operational conditions.
design modifications that reduced turbulence and In the future, advancements in parallel computing and
particle impact on critical blade areas, thus extending cloud-based simulations could make high-fidelity CFD
the operational life of the blades. and FEA models more accessible. Han et al. predicted
54
This integrated approach was further advanced by Li that the use of multi-physics simulation platforms – which
et al., who combined multi-scale modeling techniques combine CFD, FEA, thermal analysis, and ML – will become
51
with CFD and FEA to assess how micro-scale erosion increasingly common in turbine design and maintenance.
events (e.g., pitting and surface roughness) accumulate Such platforms could allow for continuous monitoring and
over time to affect macro-scale blade performance. optimization of turbine blades, ensuring higher efficiency
Their research demonstrated that small-scale surface and longer operational life. Table 9 summarizes the key
irregularities caused by erosion can lead to increased drag research contributions that are exclusively highlighted in
and reduced aerodynamic efficiency, which ultimately this article.
affects the overall performance of the turbine. Their 3. Theoretical analysis
integrated model allowed for more accurate life-cycle
predictions for gas turbine blades, highlighting the need The theoretical framework of this research offers a
for more erosion-resistant materials and coatings. structured lens to examine erosion-induced failures in

Volume 1 Issue 3 (2024) 73 doi: 10.36922/ijamd.5188