Multi-objective optimization of bleed slot mechanism for a radial flow compressor based on stall control criteria

In radial flow compressors, the ability to achieve a delayed surge margin represents a significant aspect of the design process. The challenge in optimizing surge delay techniques arises from the inherent difficulty of measuring instability within the confines of steady-state simulations. In this study, our goal is to define the objective functions necessary for optimizing surge delay techniques, thereby eliminating the direct calculation of surge margin. Using multi-objective optimization with a variety of configurations, we produced three bleed slot designs for a radial flow compressor. Subsequently, the performance of the three optimal designs were evaluated through LES and RANS simulations. To ensure a fair comparison, an unsteadiness indicator was developed that includes both the fluctuation and the average of the mass flow rate. The results indicated that reducing the incidence angle while simultaneously maximizing the pressure ratio and efficiency resulted in a bleed slot that was more effective in delaying impeller stalling. Based on the analysis of the mass flow rate across 26 impeller rotations, the optimal bleed slot demonstrated the most pronounced enhancement in impeller stability, with a 31% increase. The channel length and width of the optimized bleed slot were found to be 81% and 6% of the impeller radius, respectively. The optimized slot resulted in a flow recirculation of approximately one-fifth of the machine’s overall flow rate. This yielded an estimated increase of 2.1% in the pressure ratio of the base compressor under near-surge conditions. However, this came at the expense of a 1.3% reduction in efficiency. It is postulated that inflow excitation is the primary cause of airfoil surge delay, which is governed by the bleed flow rate and the dynamics of the bleed cells. The optimization results also confirmed the existence of a direct correlation between the incidence angle and the rate of entropy generation at the leading edge.