Erosion Prediction of Gas Turbine Compressor Blades Subjected to Water Washing Process

Abstract

Blade fouling is a relevant problem in turbomachinery applications. It affects both compressors and turbines. In the first case, fouling can be generated by the presence of dust, ashes or brackish air (in offshore applications). In turbines, fouling is mainly generated by residual of combustion process. Blade fouling generally leads to a reduction of the performance due to an increase on profile losses. Here we focus on the fouling due to salt deposition on naval/off-shore applications referring to machines that are part of the fleet of gas turbines manufacturers. In such applications, it is common to introduce on-line washing devices aiming at removing fouling from the early stages of the compressors. The water is sprayed upstream of the first rotor, it impacts on the rotor blades and thus dissolving the deposited salt. However, this procedure possibly leads to blade erosion and/or corrosion. A clear comprehension of the erosion mechanism is the main objective of the present work. To this end, we propose an integrated multi-phase CFD tool. The multi-phase flow is analyzed by adopting a one-way coupling, thus assuming water droplets to be drag by the carrier flow without influencing the main flow. The droplets are dispersed and tracked singularly by adopting a Lagrangian approach. As for the erosion, well-known and widely accepted models are used. The capability of a Lagrangian code, P-Track, developed and validated at the Department of Mechanical & Aerospace Engineering, Sapienza University in Rome, is presented. The code is able to predict the droplets trajectories, as well as to simulate the impact on the solid walls and the erosion mechanism. Simulations were performed using 25 and 100 ?m droplet size. Results, expressed in terms of normalized erosion rate, show the erosion patterns and erosive effect of the two size classes. Erosive capacity is proportional to droplet size, and the most eroded part of the blade is the leading edge, which is in qualitative agreement with measurements.