OPTIMIZATION OF A 3-STAGE BOOSTER PART 1: THE AXISYMMETRIC MULTI-DISCIPLINARY OPTIMIZATION APPROACH TO COMPRESSOR DESIGN

摘要

In this first part of a two part paper, an axisymmetric multi-disciplinary optimization approach for compressors is presented and applied to the design of a three stage booster. The booster has been chosen because its optimization gets little attention in the literature, it has low rotational speed and high curvature, and is also a component with only a few stages to test the capabilities of the approach. Optimization of compressors using a meanline approach have been done in the past, but a mean-line code cannot easily deal with complex curvature effects that are accentuated in a booster. An axisymmetric flow solver with a coupled boundary layer and compressor loss models is used for the aerodynamics, and an axisymmetric disk analysis code is used to generate weight-optimum disks for every rotor. The process is driven by the DAKOTA optimization package available from Sandia Labs. A genetic optimizer is used to create the Pareto front for a multi-objective function that includes efficiency, weight, length and number of airfoils. Following the genetic algorithm, a gradient based algorithm is also used. The design space is specified using physical parameters that completely define the multistage compressor. A booster made of titanium is presented in addition to two design studies. One design study explores using carbon-carbon composites and another design study explores restricting the last stage stator to 10 blades to understand if an integrated strut concept is feasible. Several optimum results along the Pareto front are described, and they are not intuitive. The optimizer has found solutions that have very high reactions in the last stage. The near-wall streamlines at the edge of the boundary layer are used as the resulting flowpath for the design. The benefit of the high stage reaction is to keep the rotor at a high tip radius, and have high turning in the following stator with very low diffusion as it matches to a lower radius high pressure compressor. The optimization process is fast enough to replace a meanline approach and explores a large design space to create a novel design.