Perpustakaan Digital ITB

Centrifugal compressor is widely used and applied to turbomachinery for industrial and aerospace purposes. Nowadays, compressors are demanded to have high perfor- mance so they will produce fewer emissions or lighter engines with the same output power or thrust. To achieve an excellent performance of the centrifugal compres- sor, there are two important parameters that we need to notice, which are pressure ratio and eciency. However, improving one of these parameters sometimes will reduce the performance of the other parameter. Thus, this study will focus on multi-objective optimization to improve the overall performance of the impeller of the SRV2-O centrifugal compressor by deforming its impeller. Using Bayesian op- timization and NSGA-II assisted by surrogate model, the impeller, which has been parameterized using control point of Bezier curve, will be optimized by two and ten variables (control point) optimization. The optimization processes have successfully obtained several solutions called Pareto front. One solution is chosen as the opti- mized solution, this solution improves the pressure ratio by 1.14% and eciency by 2.67%. Moreover, this design reduces the strength of the shock wave at the leading edge and delays the ow separation closer to the trailing edge. The sensitivity anal- ysis is conducted to gain important knowledge about the physics of the ow and its relation to the input variables. Theta angle control points near the inlet are the most responsible variables for this optimization.