Genetic optimization of longitudinal ribs on comprehensive heat transfer performance in hollow rotor within axial through flow

Abstract

The gas foil bearing (GFB) is widely applied to in high-speed, low-power and high-precision oil-free turbo-machinery system, and it faces a thermal management issue of aerodynamic heat caused by viscous dissipation in ultra-high speed conditions. The longitudinal ribs are applied to enhance the convective heat transfer of axial through flow, and the effects of rib parameters, including rib height (e), rib width (θ) and rib number along circumferential distribution (n), on the heat transfer and flow losses are numerically studied by CFD calculations. The radial basis function neural network (RBFNN) coupling with genetic algorithm is applied to optimize the comprehensive heat transfer performance of the longitudinal ribs. The comprehensive performance factor (PEC) is considered as the unique optimization goal for the single-objective optimization method, while the heat transfer coefficient (Nu¯/Nu0) and relative pressure loss (f/f0) are both set as the optimization objectives for the multi-objective optimization method. It shows that the Opt-4 optimized structure searched by single-objective optimization method has a best comprehensive heat transfer performance, with a relatively low and uniform distribution of rotor temperature and a moderate pressure loss. The multi-objective optimization method is more suited to the two optimization targets difficult to unify or with additional restrictions.