Application of the ball-spine algorithm to design an axial-flow comprssor blade

In this paper, a novel inverse design algorithm, called a ball spine algorithm, is developed to design the blade to blade ow passage of axial- ow compressors. The algorithm considers the blade surfaces as a set of virtual balls that move freely along the specied directions, called ʹspinesʹ. To start the solution, an initial guess for the blade geometry is required. Then, the blade-to-blade ow eld is solved using an inhouse inviscid solver. Comparing the Computed Pressure Distribution (CPD) with the Target Pressure Distribution (TPD) over the blade surfaces gives guidelines in dierential movements for the balls to obtain a modied geometry. For the modied geometry, new grids are automatically generated by an algebraic-elliptic grid-generator. The sequence is repeated until the target pressure is met. An error parameter,
PD, indicating the dierence between CPD and TPD along the suction and pressure surfaces, is computed, while the blade geometry evolves toward the target geometry. For the initial guess,
PD is obtained as 13700 Pa, while, after 100 generations,
PD is reduced to 222 Pa.