Optimization of Smooth Straight-Through Labyrinth Seal Based on XGBoost and Improved Genetic Algorithm

Abstract

Improving the sealing performance of labyrinth seal is beneficial in reducing the loss of downstream components as well as improving the aero-engines' efficiency. Meanwhile, the optimization of labyrinth seal is more cost-effective than optimizing other components of aero-engines. Based on extreme gradient boosting (XGBoost) and improved genetic algorithm (GA), an automatic optimizer for smooth straight-through labyrinth seal is proposed. According to the ordering of feature importance, the ten selected geometric parameters of smooth straight-through labyrinth seals with two gaps (0.1 mm, 0.25 mm) are optimized when the axial length of labyrinth seal is limited. The optimization results show that the leakage rate of the optimized labyrinth seal with two gaps is 19.67% and 23.80% lower than the reference labyrinth seal, respectively. It is found that the decrease of fin height, fin angle, fillet radius, and fin width, and the increase of fin pitch are beneficial to improve the sealing performance of smooth straight-through labyrinth seal. The analysis of flow field reveals the reasons for the improvement of sealing performance due to the change of geometric parameters. However, it must be noted that the optimization effect decreases as the pressure ratio increases, primarily as the result of the increase in carry-over coefficient. This work provides a new approach to optimize smooth straight-through labyrinth seal.