Optimization Of Numerical Simulation Of Microstructure’s Algorithm Based On Phase-field Method

Phase-field method of solidification microstructure simulation is an effective way.The model of the main idea is to introduce one or more of the change order parameter,and instead of the traditional sharp interface by using a diffuse-interface model. Theclassical theory in the sharp interface coupling diffusion equation needs to track theinterface at the same time, then resulting in defects of large amount of calculation.Phase-field model (PFM) can avoid tracking simulation of solid-liquid interface underthe condition of complex microstructure. This method results the widespread attentionof workers. With the coupling of phase-field, solute concentration field, temperaturefield, flow field and external field, the solidification simulation is closer to the realistic.The PFM simulation method is applied more and more nowadays. Simulation ofSolidification process in grain and smaller dimension simulation of solidificationmicrostructure evolution and realization of prediction of casting solidificationmicrostructure and mechanical properties, which served as a research focus of the recentforming simulation. But the calculating quantity of phase-field method has been a veryimportant bottleneck, which limits the further development. Now it is a problem to besolved that how to use Phase-Field Simulation of microstructure faster and moreefficiently.In this article, the phase-field model coupling the field and solute field isestablished by basing on the pure diffusion of binary alloy. combining governingequation with finite difference method for discrete processing, and apply this quad-treegrid technology to study on self-adaptive Mesh Model Based on Phase-FieldSimulation for microstructure.The study of this article includes:Firstly, this article introduces the quad-tree grid technology, including quad-treestructure characteristics, storage, grid rules of coding, data structures of grid, meshgeneration, and the traverse of grid, furthermore, it uses quad-tree grid technology tobuild on Phase-field adaptive model of the net. This article also skillfully uses thesearch value of this variable layer to handle the traverse of grid and classification ofdifferent cell of the grid, building rules of the grid, ensuring that the difference betweenthe value of every grid and its must not exceed1, which ultimately makes sure that the data is convergent，and simulation graphics more realistic.Secondly, the discretization of Phase field control equation and solute field controlequation is solved by quad-tree grid technology, when discreting, calculating ofderivative discussion three cases was included, meanwhile, different floors of the gridcell neighbor, and the grid cell divided is calculated in cubic polynomial interpolation,which can reduce the interpolation error in the calculation process.Finally, the results of the data, which has been processed with quad-tree gridfinite difference method,need to get data point that can generate Finite Element meshnodes to display in the graphics through the conversion of the value of the nodecoordinates and physical field. Simulation results compared to the uniform gridsimulation results with the optimization of size of root element grid,the value of thelayer of grids and temperature values, The results indicate that the method which basedon quadtree structure of adaptive grid method not only simulate the dendrite growthvery well, but also greatly improve the computational efficiency, reduce the loss ofcomputing resources.