不同维度缩放方法在航空发动机总体仿真中的应用

为了研究完全耦合和迭代耦合方法的实现途径及两种方法的差异,针对涡扇发动机,建立了风扇部件的二维仿真模型,发展了基于完全耦合和迭代耦合两种维度缩放方法的发动机多维度仿真模型,对比分析了两种方法在仿真结果、计算速度及可操作性等方面的差异,同时对比了基于部件通用特性图的发动机零维仿真模型与多维度仿真模型的计算结果。结果表明:采用完全耦合方法进行维度缩放时,需要结合部件二维仿真模型对边界条件的要求调整迭代变量,同时需使用发动机零维仿真模型的计算结果作为多维度仿真模型中迭代变量的迭代初值,才能保证模型的收敛性;采用迭代耦合方法进行维度缩放时只涉及部件二维仿真模型与发动机零维仿真模型之间的参数传递控制,更容易实现。在计算均收敛的情况下,基于完全耦合方法与迭代耦合方法的发动机模型的计算结果无明显差异,但后者的计算速度更快。与基准模型相比,基于部件通用特性图的发动机零维仿真模型计算得到的推力最大误差大于8.34%,而多维度仿真模型的推力和耗油率的误差均小于3%,多维度仿真模型可更准确地预估发动机性能。

Application of Different Zooming Strategies in Aero-Engine Simulation

The two-dimensional fan model was built and integrated into the cycle analysis with fully coupled approach and iterative coupled approach to establish the multi-level model for turbofan engine. The differences of simulation result, calculation speed and operability between the two zooming strategies were analyzed. A comparison of the results obtained from the zero-dimensional engine model based on component default map and multi-level was presented. The results indicate that the iterative variables need to be adjusted based on the requirements of the two-dimensional fan model for boundary conditions in order to establish the multi-level model with fully coupled approach, and the result of zero-dimensional engine model is used to initialize the iterative variables to ensure the convergence of the multi-level model. As for iterative coupled approach, the key is the control of the data transfer between the two-dimensional fan model and the zero-dimensional engine model, and it is easier to carry out than fully coupled approach. When the computation converges, there is no difference in the results obtained from the multi-level models using different zooming strategies, but the calculation speed of the multi-level model with iterative coupled approach is faster. Compared with the result of the baseline model, the maximum difference in the thrust solved from the zero-dimensional engine model is bigger than 8.34 percent, and the maximum differences in the thrust and specific fuel consumption of the multi-level model are less than 3 percent what means that the multi-level model can predict the engine performance more accurate.