Rock is a multiphase material containing porous structures. It is well known that the porous structures of rocks vary at the microscale, while the properties of rocks are consistent at the macroscale. Therefore, predicting the properties of rock by studying the porous structures of rocks is a good way to prevent engineering disasters. Nevertheless, there have been few studies on this issue. The present paper proposes a cooling–solidification annealing reconstruction algorithm, which can be used to reconstruct a numerical rock model with porous structures to predict its properties quickly. In this algorithm, the system is annealed iteratively with decreasing scale and is constrained in the current scale. Moreover, another grey phase is introduced into the ordinary two-phase reconstruction system to form a new three-phase system. Therefore, the novel algorithm is adept at flexibly modelling complex structures. In particular, by using this algorithm, a simple, local and scale-dependent model can generate very complex porous structures, and the uncertain portions at coarser scales can be annealed directionally by constraining the calculation domain. The numerical results of the reconstructed model are also compared with the experimental results to verify the proposed algorithm. The comparison shows that the novel proposed algorithm can quickly generate a rock model with different porous structures to predict its properties quickly.

岩石是一种含有多孔结构的多相材料。众所周知，岩石的多孔结构在微观尺度上是不同的，而在宏观尺度上岩石的性质是一致的。因此，通过研究岩石的多孔结构来预测岩石的性质是预防工程灾害的好方法。然而，关于这个问题的研究很少。本文提出了一种冷却-凝固退火重建算法，该算法可用于重建具有多孔结构的数值岩石模型，以快速预测其性质。在该算法中，系统以递减的尺度进行迭代退火，并被约束在当前尺度上。此外，在普通的两相重构系统中引入另一个灰度相，形成新的三相系统。所以，新算法擅长灵活地对复杂结构进行建模。特别是，通过使用该算法，一个简单的、局部的和尺度相关的模型可以生成非常复杂的多孔结构，并且可以通过约束计算域来定向退火较粗尺度的不确定部分。将重构模型的数值结果与实验结果进行对比，验证了所提出的算法。比较表明，新提出的算法可以快速生成具有不同多孔结构的岩石模型，以快速预测其特性。并且可以通过约束计算域来定向退火较粗尺度的不确定部分。将重构模型的数值结果与实验结果进行对比，验证了所提出的算法。比较表明，新提出的算法可以快速生成具有不同多孔结构的岩石模型，以快速预测其特性。并且可以通过约束计算域来定向退火较粗尺度的不确定部分。将重构模型的数值结果与实验结果进行对比，验证了所提出的算法。比较表明，新提出的算法可以快速生成具有不同多孔结构的岩石模型，以快速预测其特性。