Reduced order modelling of optimized transonic compressor rotor blades

The pursuit of efficient, compact and lightweight aircraft engines are important as it leads to a reduction in specific fuel consumption. Compact and lightweight engines result in compressor designs with high transonic rotor speeds and high stage loading in order to maintain the desired pressure ratio. From an aerodynamic perspective, this poses a challenge since it will be more difficult to design a compressor with respect to high efficiency. Finding a faster design for transonic compressor rotor blades with the same efficiency would therefore be beneficial. In this thesis the blade generation method, multiple circular arc (MCA) blades represented on conical surfaces was followed to investigate the feasibility of modelling an already optimized 3D compressor rotor blade. In the MCA method conical surfaces are used to approximate the axi-symmetric streamsurfaces in the blade rows, where the MCA method is based on the piecewise preserving of constant turning rate of camber. The complete MCA blade was formed by stacking individual blade elements, where each blade element is defined by 11 design variables. These variables are local blade angles, thickness and axial distances for the critical points on the blade elements; leading edge, trailing edge, maximum thickness and transition. The MCA blade was constructed with 15 spans, resulting in a total of 165 optimizing variables. By employing variable reduction, the total number of optimizing variables was reduced to 75. A multi-variable and multi-objective optimization was made for these 75 variables with the NSGA-II as the optimizer. For the multi-objective optimization two objective functions were formulated to compare the MCA blade with the already optimized 3D blade profile. The objective functions describe the average deviation in the airfoil shape between the blades and the maximum axial cone deviation, which is to ensure that the blade elements are correctly stacked. A design from the Pareto front, which traded the stacking for lower deviation in the blade shape was evaluated using CFD as a performance tool. From the optimization results, it was shown that the shape of the blade elements match the optimized blade well. However, reaching the same stacking of the two blades was not achieved. Results from CFD simulation indicated that the design point is close to the optimized blade with a deviation in polytropic efficiency of 1 %. Combining the optimization with the CFD results it was concluded that the reference blade can be constructed with the MCA model and reach a operating point which was close to the reference, in regards to operating pressure and corrected mass flow