Results in Engineering (Dec 2024)
Effect of the liquid droplet and flow pattern on centrifugal compressor blade fatigue in an industrial olefin unit via CFD
Abstract
Blade breaking is one of the main issues with industrial process compressors. In this regard, the goal of this study is to look into the potential for droplet formation to see how it can affect the impeller blade breaking (an alloy of 7175 aluminum) of an Olefin plant under the operating conditions (18300–34500 rpm, 286–307 K, 720–969 kPa). At first, the Peng-Robinson (PR) and Soave-Redlich-Kwong (SRK) equations of states were used in simulating the turboexpander process and conducting a thermodynamic analysis to investigate the probability of liquid droplet production in the turboexpander. Then, using the Euler-Lagrange approach and the Realizable k-Ɛ turbulence models, the two-phase computational fluid dynamic (CFD) modeling of liquid droplet motion was carried out. Process simulation reveals that the compressor is operated far from two-phase conditions, indicating a negligible probability of droplet formation. The results also show that the area of the blade that had already broken down experiences the highest pressure of the gas stream for each of the three examined rotor speeds and that the gradients of the pressure, stress, and temperature of the gas in the high-pressure (HP) compressor are significantly higher than those in the low-pressure (LP) compressor. Additionally, two-phase modeling of liquid droplets reveals that the presence of condensate has a negligible effect on increasing pressure, shear stress, and other factors affecting the blades. Therefore, under the steady-state conditions, the impact of liquid droplet condensation on the blades can be ignored, and by raising the mass percentage of liquid droplets, the shear stress, pressure, and coefficient of friction all slightly increase. Finally, it can be concluded that, under the operating conditions considered in this work, the droplet production is not the main cause of blade fatigue.
