Modeling of shrinkage characteristics during investment casting for typical structures of hollow turbine blades

Abstract

To study the coupling mechanism of shrinkage distribution and complex structures in the precision casting process of hollow turbine blades, the blade structure was simplified to a hollow thin-walled structure with resistance and non-resistance. Four different structures of casting and a casting system were designed. Based on the combination of numerical simulation and experimental measurement, the shrinkage distribution and shrinkage model of castings were established. The results show that the simulated and measured shrinkages have the same trend. Then, the structural parameters affecting shrinkage, including wall thickness, outer diameter, and unobstructed structure, were discussed. A mapping model based on a backpropagation (BP) neural network reflecting the relationship between structural parameters and shrinkage rate was constructed. According to the BP neural-network-based mapping model, the average deviations between the predicted and measured values of the transitional and normal sections are 5.8% and 2.4%, respectively, which improves the accuracy compared with existing research, indicating that the shrinkage model has a good performance in predicting shrinkage of the typical structure in hollow thin-walled castings.