Effects of water-to-air mass ratio on long-term washing efficiency and erosion risk in an axial compressor under online washing conditions

One of the main reasons for gas turbines’ performance losses is the deposition of dirt on the compressor blades. Dirt deposit has to be periodically removed to keep the engine performance as high as possible. This is the reason motivating the presence of online water washing systems in most compressor gas turbines. Such systems aim at cleaning the compressor blades to recover efficiency; thus, the larger the water flow, the better it is assumed the compressor is cleaned (fixing all the other conditions). In the present work, we simulate the long-term behavior of a real axial compressor, from the inlet to the first-stage rotor, subject to online water washing with different water flowrates. The frozen rotor approach is adopted to solve the flow field in the rotor region. Simulations are performed by using the unsteady k-ε realizable model coupled with a Lagrangian tracking of the injected liquid phase. Water droplet erosion is handled by using a semi-empirical model developed by the authors. In each simulation, 504,000 parcels have been tracked, providing statistically reliable predictions. To simulate the long-term evolution of the washing process, a discrete mesh morphing technique coupled with the use of specific scale factors is adopted. Each of the tested configurations is composed of three successive erosive steps up to the blade compressor end-of-life. By varying the water-to-air mass fraction (WAMF*), six different injection configurations are assessed in terms of long-time average washing efficiency and erosion risk. The results predicted show the dependence of the considered washing indices on water mass flowrate and set the stage for the development of a washing optimization tool, which can help the design and management processes. In scenarios where washing indices are given minimal importance and the objective is to reduce the risk of erosion, the optimal injection configuration was shown to correspond to a WAMF* value of 0.250. Conversely, when washing efficiency is prioritized, the optimal injection configuration has been shown to correspond to the case where WAMF* = 0.750.