Gas turbine engine performance can be severely degraded by solid particle ingestion (mineral dust, volcanic ash) during the operation of aero-engines in harsh environments. In the compressor section, erosion and fouling are the main deterioration mechanisms. The accurate prediction of particle fate and compressor damage mechanisms depends on many variables, such as engine operating conditions, airflow phenomena, blade geometry, particle and blade material properties and particle size. During particle-blade interaction, particles can break-up at high impact velocities. Although particle fragmentation significantly affects particle inertial properties, trajectories and impact patterns, this phenomenon is not well-understood. To rapidly assess the influence of various variables on particle impact fate, a generalized reduced-order model was initially developed to predict interaction probability of particles with blade, casing and hub, and the probability of particles escaping the stage without interacting. Particle deposition modes (diffusion, inertia, centrifugal) were found to strongly affect two of the most pertinent variables to post-impact fate: impact velocity and angle. The data-set to develop the probabilistic model was generated by running three-dimensional Computational Fluid Dynamics (CFD) simulations to analyze the particle-laden flow through a well-known single-stage axial compressor test case, the NASA Rotor 37. Lagrangian particle tracking of three different minerals, each with a wide size distribution (0.3-135 \textit{$\mu$}m), was computed on the continuous phase solution at three different blade rotation speeds. The impact and escape relationships were found to be solely a function of a new reduced order parameter termed the `generalized centrifugal Stokes number', $Stk_{cent,gen}$. Predictions of particle impact kinematics were then shown to be possible on the basis of this reduced order parameter alone, thereby enabling fast computation of particle fragmentation and other post-impact behaviour in axial compressors. A physics-based fragmentation approach, which is dependent on material properties and impact conditions, has been derived with the intention of predicting the breakage probability of an ingested mineral dust following interaction with an axial compressor rotor. This was partially achieved through a methodology for predicting which subset of particles are liable to fragment, on the basis of their inertia and strength properties. It was found that despite the higher hardness and fracture toughness of corundum particles than those of quartz, corundum particles had higher breakage probability due to their higher impact strain rate than that of quartz. Fragmentation probability was increased at particle normal impacts. At the highest $Stk_{cent,gen}$, fragmentation probability increased sharply with the increase of $Stk_{cent,gen}$ due to the significant reduction of particle brittle strength and the subsequent increase of contact deformation time. Some limitations for the implementation of the physics-based fragmentation approach were identified. A more comprehensive sensitivity analysis of fracture time and contact deformation time to material properties is further required to improve the physics-based approach. Thus, an empirical fragmentation approach was incorporated in the reduced-order model to show the applicability of the model for predicting particle breakage probability. The empirical approach showed that fragmentation of a typical size distribution of quartz, following impact with the NASA Rotor 37, would not occur to any significant degree. Only a very small subset of particles were found to impact with kinetic energy sufficient enough to effect fragmentation, on the basis of findings from a recent coupon test in the literature. This is thought to be due to the constraints on applicability of the empirical dataset, which were based on only three distinct impact velocities. The present study d
Particle Fragmentation Probability in Axial Compressors
Klaoudatos, D. (Author). 31 Dec 2023
Student thesis: Phd