Turbomachinery performance degradation due to erosion effect

• Main Author: Ghenaiet, Adel
• Other Authors: Elder, R. L.
• Language: en
• Published: Cranfield University 2009
• Online Access: http://hdl.handle.net/1826/3542

Erosion of gas turbines operating in sandy or dusty environments can result in serious damage to the engine components, particularly the compressor unit. This phenomenon is a result of the ingestion of the sand particles into the engine and their consequent abrasive impacts on the blade surfaces. In order to understand the mechanism of sand ingestion and the resulting performance degradation, a general methodology has been developed for predicting the trajectories of particles, the erosion rates and blade profile changes, with predictive capabilities for performance degradations within more general configurations of turbomachines. This methodology was applied to an axial fan with upstream guide vanes (contra whirl) and was supported by experimental results. The numerical models for calculating the particle trajectory are based on the Lagrangian tracking technique and the eddy lifetime concept. The turbulence effect is assumed to prevail as long as the particle eddy interaction time is less than the eddy lifetime, and the displacement of a particle relative to the eddy is less than the eddy length. The flow field was solved separately using the Navier-Stokes finite volume flow solver " TASCflow " commercially available from ASC. The governing equations of the particle motion are solved using the Runge-Kutta Fehlberg technique. The tracking of particles and their locations is based on a finite element interpolation method. The developed Fortran code for predicting particle trajectory and erosion due to particle impact accounts for different types of boundary conditions and handles different frames of reference. The fragmentation of particles after rebound was also implemented. The number of particles seeded upstream of the IGV blades can be determined either by a user defined concentration profile or by a measured concentration profile. Also, particles can be seeded separately in a group at a release position. In the present study, the concentration profile and the initial particle velocity and angle of particle spread were determined from a laser transit anemometer. Two types of particles were used, a narrow size bandwidth (150-300micron) quartz particle and MIL-E5007E quartz particle, both of which have a normal distribution. The global rate of erosion, the reduced mass of blades and the changes of the blade geometry were predicted and compared with experimental results at different concentration levels. The baseline axial fan characteristics were measured at different mass flow conditions at a constant speed of rotation. To assess the effects of erosion, the characteristic measurement was repeated after each step of sand ingestion. The predicted aerodynamic performance; adiabatic efficiency, pressure rise coefficient and stall margin before and after erosion degradation were also determined from a developed Fortran program, which is basically a mean line method that uses advanced correlations for aerodynamic losses. Prediction of the particle trajectories show that high numbers of impacts (and maximum erosion) occurred near leading edge and tip region, which were also borne out by locally injected sand tests. The global rate of erosion and the consequent changes of the blade geometry were also predicted and compared with experimental results. The erosion pattern at high concentration of MIL-E5007E sand particles depicts net loss of material over the leading edge and the tip corner. The tip clearance increased markedly with a rounding of blade leading edge, which is the main cause of the decrease in efficiency, pressure rise, and surge margin. A parametric study with turbulence and fragmentation effects show that both parameters can influence the erosion rate and blade geometry deterioration. The results of the aerodynamic performance simulation using mean line method, which includes an erosion fault model, show good agreement with experimental results.