科研之友
微信
新浪微博
Facebook
分享链接
摘要
Designers of aerospace turbopumps have to decide which specifications should be given to the machine geometry so that it performs the desired tasks in the best possible way. The aim of this work is to develop an useful tool helping the designer to obtain the best pump performance without using computationally expensive codes, e.g. those based on Computational Fluid Dynamics (CFD). Therefore, a meanline pump model, based on theoretical equations and empirical correlations is implemented to provide a fast means of modeling pumps for cryogenic rocket engines; anyway it can predict the performance of pumps when they operate within design as well as off-design operative conditions. Moreover, the pump model can simulate axial inducer, mixed-flow and centrifugal pumps and also multistage pumps in series which are very common in aerospace applications. The pump model is tested and calibrated against experimental data using an appropriate optimization genetic algorithm which searches the best parameters set enabling the maximum superposition between the simulated and the experimentally measured operative points. Next, a Multi-Objective Evolutionary Algorithm (MOEA) using a combination of three population evolution methods searches the best turbopump geometrical configurations which, for a given head curve, improve the pump efficiency as much as possible. Such multi-objective genetic optimizer is based on the ordination of nominated, solutions into a non-dominated set of solutions, which in turn are based on the dominance concept. The overall procedure allows finding a great number of optimal and feasible constructive configurations for the considered turbopumps.
- 出版日期2010
