Optimasi Desain Trailing Edge Splitter untuk Mengurangi Turbulensi dan Pressure Drop pada Exhaust Silencer Turbin Gas Menggunakan Metode CFD dan Bayesian Optimization

Gas turbines are widely used in modern power generation due to their high efficiency and flexibility. However, using exhaust silencers in exhaust gas systems can increase back pressure, reducing overall turbine performance. The aerodynamic behavior of the splitter inside the exhaust silencer, especially at the trailing edge, significantly affects pressure drop and turbulence intensity. This study aims to optimize the trailing edge geometry of a GE Frame 6FA exhaust silencer splitter using a combination of computational fluid dynamics (CFD) and Bayesian optimization (BO). CFD simulations were conducted in ANSYS Fluent using a poly-hexcore mesh, while the optimization employed a Gaussian process (GP) surrogate model with an expected improvement (EI) acquisition function. Seven trailing edge angles (5°–20°) were evaluated to observe the aerodynamic trade-off between pressure drop and turbulent kinetic energy (TKE). The optimal configuration was found at 13.9°, yielding a TKE of 0.43 J/kg and a pressure drop of 18.38 Pa, well below the industrial allowable pressure drop limit of 34 Pa. The results confirm that the BO-GP framework effectively reduces computational cost while achieving a more aerodynamically balanced silencer design.