合金铸件中微观缩孔的动态形成过程与枝晶凝固过程直接相关。为了模拟Al-4.7%Cu (质量分数)合金在凝固过程中发生的枝晶收缩现象，本工作建立了一个元胞自动机-格子Boltzmann方法(CA-LBM)耦合模型。该模型利用CA模拟枝晶生长过程中缩孔的形成，并通过LBM研究缩孔在液相中的扩散过程。首先用真空孔洞在液体中扩散均匀化的数值模拟验证了所建立的CA-LBM数值模型的正确性，然后对比计算了考虑和不考虑枝晶收缩的单枝晶凝固过程，最后计算了考虑收缩的多枝晶凝固收缩过程。对单枝晶凝固过程的计算结果表明，不考虑收缩的单枝晶在凝固结束后内部不存在孔洞，而考虑收缩的单枝晶在凝固结束后出现了均匀的微小缩孔；孔洞的出现对单枝晶的生长形貌有显著的影响，即会促进单枝晶二次分枝的出现。对多枝晶收缩模拟的结果也表明，微小孔洞出现在最后凝固的枝晶间交界处。缩孔总量的计算值与理论值的误差很小，证明所建立的数值模型是有效可靠的。

The dynamic formation process of microshrinkage pole in alloy castings is directly related to the dendrite solidification process. To simulate dendrite shrinkage in the Al-4.7%Cu (mass fraction) alloy during solidification, we proposed a coupling model combining cellular automata (CA) and the lattice Boltzmann method (LBM), referred to as the CA-LBM model. In this model, CA was used to simulate the formation of shrinkage pores during dendrite growth, whereas LBM was applied to study the diffusion process of shrinkage pores in the liquid phase (fully liquid conditions). First, the accuracy of the proposed CA-LBM numerical model was verified through the numerical simulation of the diffusion homogenization of a vacuum cavity in liquid. Then, the solidification process of a single dendrite—with and without dendrite shrinkage were compared, followed by the calculations of the multi-dendrite solidification contraction process with and without dendrite shrinkage. Simulations of the single-dendrite solidification process indicated that no internal pores were formed in the single dendrites when shrinkage was not considered. However, when shrinkage was considered, uniform microshrinkage pores appeared in the single dendrites. Moreover, the presence of shrinkage pores notably influenced dendrite morphology by promoting secondary branching. The results of the multidendrite contraction simulation also showed that micro-shrinkage pores tended to form at the junctions of the last solidified dendrites. A comparison between the calculated number of shrinkage poles and the theoretical value showed a small error, indicating the effectiveness and reliability of the proposed numerical model.