Shrinkage greatly influences the mechanical and fatigue properties of compacted graphite iron and it is necessary in order to study the causes of shrinkage in compacted graphite iron and to predict it effectively. In this paper, a kind of cylindrical necking test sample was designed to evaluate the shrinkage in compacted graphite iron, and a method to calculate the size of shrinkage was proposed. By observing the microstructure around the shrinkage zone, it is concluded that concentrated shrinkage mainly appears in the solidification region where the dendritic gap is closed, and the isolated shrinkage mainly occurs in the final solidification region, and the supersaturated carbon elements are gathered on the surface of the shrinkage. The cause of shrinkage in compacted graphite iron is caused by its solidification method, where the austenite dendrites and the eutectic clusters are generated close to the melt zone during the solidification process, leading to the inability to feed the shrinkage. Based on the thermodynamic analysis, the equations between the volume change of each phase, solid phase rate, and time during solidification of compacted graphite iron were established to theoretically explain the formation mechanism of the shrinkage. Taking nine parameters such as the chemical elements and characteristic values of thermal analysis as the input nods, a four-layer BP neural network model for predicting the size of shrinkage in compacted graphite iron was constructed, and the R-squared of the model reached 97%, which indicates it could be used to predict the shrinkage tendency.

The hot deformation behavior of 301 L stainless steel is investigated at 1000–1250 °C and 0.1–50 s−1. Deformation characteristics are studied through compressive stress–strain curves and constitutive equations. According to the dynamic material model and a series of diffusion annealing experiments, the hot processing map of 301 L stainless steel is established to determine the suitable processing domain. The results show that the stable domain is located in 1085–1240 °C and high strain rate range of 4.5–50 s−1, and the strain rate is the main factor affecting the machinability. Based on the hot compression tests, δ‐ferrite gradually forms a network structure with the increase of temperature; however, it can be inhibited at high strain rate. Therefore, it is necessary to reduce the temperature and increase the strain rate in the actual processing of the machinable area. It is found that diffusion annealing can reduce the content of δ‐ferrite from 12% to 0.67%, and the dissolution of δ‐ferrite is controlled by the diffusion of Cr and Ni, especially noticeable at the interface between γ and δ phase. Finally, the optimum diffusion annealing process parameter is determined at 1300 °C and 10 min.