The Study Of Aerodynamics Numerical Optimization Design Of Axial Turbomachinery Blades

The thesis constructs an aerodynamic numerical optimization platform for axial turbomachinery fully three-dimensional blades. It includes three parts:parameterization and fitting of blades; simulation of flow field; optimization strategy. Blades are parameterized based on NURBS curve. Because the aerodynamic optimization design is non-linear multi-variable multi-modal and multi-objective, it uses the integrated optimization strategy which includes DOE method and genetic algorithm that can find the global optimal solution. Use this platform to optimize the blades of a NASA transonic single stage turbine, the bended stacking line and the profile of this transonic stator.The main tasks are as follows:●Construct an aerodynamic numerical optimization platform for axial turbomachinery fully three-dimensional blades.●Develop the program to parameterize and fit the blades.●Detailedly investigate the effect of parameterized variables on objectives and choose the target-sensitive variables for optimization, which reduces the search space.●Use two sets weight factor combinations for optimization and analyze the effect of different weight factor combinations on the final solution of optimization.●Choose the Non-dominated Sorting Genetic Algorithm-NSGA-Ⅱfor multi-objective function optimization and analyze the dominated relationship between optimal solution from single objective function optimization and the pareto solutions from NSGA-Ⅱ.The results show that:◆No matter whether the single objective function or multi-objective function optimization strategy it adopts, at the optimized operating point the performance of optimized blades is improved, which illustrates that the aerodynamic numerical optimization platform is reliable.◆Through analyzing the effect of different weight factor combinations on the final optimal solution, we can find that the final optimal solution leans to the objective that has the large weight factor.◆Through analyzing the dominated relationship between optimal solution from single objective function optimization and the pareto solutions from NSGA-Ⅱ, we can find that under the circumstance of the same population number and generation number, the final optimal solution from the single objective function may dominate some solutions in the pareto solutions from NSGA-Ⅱ, but the NSGA-Ⅱcan find much more non-dominated solutions, which can provide more information to decision makers.◆Through comparing the characteristic lines of original blades and optimized blades, the results show that though the performance is improved at the optimized work point, it does not ensure that the performance can also be improved at the non-optimal work points. If we want to improve the performance in the whole working range, we should implement the multi-point optimization.