当前位置: X-MOL 学术Comp. Mater. Sci. › 论文详情
Our official English website, www.x-mol.net, welcomes your feedback! (Note: you will need to create a separate account there.)
Simulation of eutectic growth using phase field method and LBM based on OpenCL
Computational Materials Science ( IF 3.1 ) Pub Date : 2020-04-01 , DOI: 10.1016/j.commatsci.2020.109523
Chang-sheng Zhu , Zhong-yan Deng , Li Feng , Yu-jie Li

Abstract During the solidification of alloys, numerical simulation of forced convection eutectic growth is essential for accurate prediction and control of solidified microstructure. The phase field lattice Boltzmann model (PF-LBM) is established by coupling the phase field model of KKSM model with the LBM model of computational fluid. In order to solve the problems of PF-LBM model, which has a large amount of computation, the simulation region is small and can not be solved in a reasonable time. The parallel computing architecture based on OpenCL can give full play to the powerful computing power of NVIDIA GPU and accelerate the implementation of the algorithm to improve the efficiency of numerical simulation. The effect of forced convection on the eutectic growth of CBr4-C2Cl6 eutectic alloy is systematically studied. The results showed that liquid flow altered the eutectic evolution by affecting the distributions of the solute ahead of the eutectic solid/liquid interface. Compared with serial CPU code, a single GPU can achieve a maximum acceleration ratio of 136.X. Solved the problem of 3-D numerical simulation of forced convection eutectic growth due to computational limitations.

中文翻译:

基于OpenCL的相场法和LBM模拟共晶生长

摘要 在合金凝固过程中,强制对流共晶生长的数值模拟对于准确预测和控制凝固组织至关重要。相场晶格玻尔兹曼模型(PF-LBM)是将KKSM模型的相场模型与计算流体的LBM模型耦合建立起来的。为了解决PF-LBM模型计算量大,模拟区域小,无法在合理时间内求解的问题。基于OpenCL的并行计算架构可以充分发挥NVIDIA GPU强大的计算能力,加速算法的实现,提高数值模拟的效率。系统研究了强制对流对CBr4-C2Cl6共晶合金共晶生长的影响。结果表明,液体流动通过影响共晶固/液界面前溶质的分布改变了共晶演化。与串行CPU代码相比,单个GPU最高可实现136.X的加速比。解决了由于计算限制导致的强制对流共晶生长的 3-D 数值模拟问题。
更新日期:2020-04-01
down
wechat
bug