DESIGN OPTIMIZATION OF THE WING SHAPE FOR THE RLV BOOSTER STAGE USING EVOLUTIONARY ALGORITHMS AND NAVIER-STOKES COMPUTATIONS ON UNSTRUCTURED GRIDS

Abstract. The wing shape of the booster stage of a two-stage-to-orbit Reusable Launch Vehicle (RLV) is optimized by considering trajectory computations coupled with three 3-D Navier–Stokes flow analyses around the complete booster configuration at three instances of its flight, namely at supersonic, transonic and subsonic flow conditions. The four-objective optimization is carried out using the Evolutionary Algorithms System (EASY 1.3) software. EASY 1.3 makes use of evolutionary algorithms and Artificial Neural Networks (ANN). The role of the latter is to act as surrogate evaluation model for the preliminary approximate evaluation of the population members, in each generation. The role of the so-called Inexact Pre-Evaluation (IPE) phase is to pinpoint the most promising individuals to be then evaluated using the Navier-Stokes solver. Using the IPE technique, optimizations with “expensive ” evaluation tools become affordable.