Exploring Spanwise Loss Patterns: Numerical Analysis and Modeling of Linear Cascades with Tip Clearance Based on Leakage Flow Intensity

Leakage and secondary flow losses significantly affect the efficiency of highly loaded compressors. Previous research on the impact of tip clearance on loss patterns has primarily focused on half-span analysis, neglecting the full-span loss redistribution caused by passage blockage effects. This paper analyzes and models the full-span loss patterns in a linear cascade of rotor blades, taking into account tip clearance and passage blockage. A database of over 200 numerical cases was constructed using an orthogonal design scheme. A novel indicator, i.e., leakage flow intensity (LFI), was developed to classify loss patterns based on different flow characteristics. Further, a new surrogate loss model, incorporating LFI and based on random forest regression, was established and validated. Results show that the loss decreases at the tip while increasing at the hub due to tip clearance, which reduces blockage at the tip but enhances blockage at the hub. Consequently, local incidence decreases at the tip and increases at the hub. Validation demonstrates strong consistency between the machine learning model and simulation data. Additionally, the LFI selector enhances the model's predictive capability and physical interpretability by selecting training samples more relevant to the prediction target.