Response Surface-Based Transition Duct Shape Optimization

Abstract

Demands on improved efficiency and reduced noise levels cause a strive toward very high by-pass ratio (BPR) turbo-fan-engines resulting in large fans and small high-pressure-ratio cores. The trend of increasing the radial offset between low- and high-pressure systems has made the design of intermediate transition ducts an area of growing importance. Shape optimization techniques for turbo-machinery applications have become a powerful aero-design tool and thanks to the rapid development of computer technology, computational fluid dynamics (CFD) may be used for optimization purposes. Surrogate model-based optimization has in recent years become a good alternative to the gradient-based search algorithms. One well-known surrogate model-based approach is response surface methodology (RSM), which has been used in the present work. RSM used together with design of experiments (DOE) can be a very efficient and robust method for CFD-based optimization. Optimization of two different transition ducts has been performed and evaluated; one 2D axi-symmetric turbine duct and one 3D compressor duct. The optimization objective was to minimize the total pressure loss. For the turbine duct both a high-Re and a low-Re turbulence closure were evaluated. The influence of design space size on the turbine duct optimization was investigated and for the 3D compressor duct the effects of adding an outflow constraint were examined. It has been found that end-wall optimization has the potential to reduce duct losses significantly. An early rapid diffusion and a stream-wise curvature shift toward the outlet seem to be important mechanisms for reducing transition duct losses.