An Evolutionary Strategies Method to Optimize Turbine and Compressor Blades
Type of DegreeMaster's Thesis
MetadataShow full item record
The main goal of this study is to improve a two-dimensional optimization scheme for turbine and compressor blades, specifically by improving the overall efficiency of each blade row with respect to objective and penalty functions. The study will utilize the NASA Energy Efficient Engine high pressure turbine stage 1 and NASA Compressor Rotor 37. Optimization of compressor and turbine blades utilizing a range of advanced learning techniques such as Genetic Algorithms, Evolutionary Strategies, and Neural Networks, have been the subject of numerous studies. This research uses Evolutionary Strategies since it has applicable characteristics which promote speed, reliability, and simplicity of implementation as compared to Genetic Algorithms or Neural Networks. To drive the analysis to an optimal solution, objective and penalty functions will be utilized and discussed as they are used to evaluate each offspring and ensure the optimum solution has the desired flow characteristics. While optimization is critically important for improving the efficiency of the compressor and turbine; of equal or greater importance is the modeling approach used for predicting performance. One element of modeling discussed in this work is the previous use of Bezier Curves and its inability to generate the entire blade, specifically at the leading and trailing edge. An improved method for modeling the blades utilizing “Class” and “Shape” Functions will be implemented as they are useful for controlling the curvature of the leading and trailing edges and still provide the same continuity as Bezier Curves.