Reliability-based design optimization of a centrifugal compressor considering manufacturing uncertainties

Abstract

A multiobjective, reliability-based design optimization (RBDO) is performed considering manufacturing uncertainties in order to facilitate the performance and reliability of a centrifugal compressor. During the RBDO procedure, Monte Carlo Simulation (MCS) is utilized to calculate failure probability. Latin hypercube design (LHD) is adopted to sample and form simulation experiment scheme. Besides, response surface model (RSM) is constructed to reduce the computational cost of RBDO. Results show that in deterministic optimization (DO), the reliability of mass flow rate is only 0.7843 while in RBDO the reliability of mass flow rate reaches 0.9976. Meanwhile, the value of efficiency and total pressure ratio descend 0.00197 and 0.00822 respectively, but both of them still satisfy the reliability requirements. Accordingly, a more reliable design scheme of the compressor is obtained.