Rapid Solution And Pre-and Post-processing Techniques For 3D Simulation Of Injection Molding

Plastic injection molding CAE plays an important role in process optimization,production cost reduction and product cycle reduction in the field of plastic processing.The 2.5D model based on the Hele-Shaw hypothesis cannot accurately simulate the parts with uneven thickness and thick-walled parts because it ignores some variables in the thickness direction.With the increasing complexity and precision of injection molded parts,the research and development of three-dimensional simulation technology has become an inevitable trend.Because of the complex shape and structure of injection molding parts,in order to better calculate the complex solution domain,unstructured grids are used in three-dimensional simulation.Compared with two-dimensional and half-model,the size of grids increases by more than one order of magnitude.At the same time,because the physical variables change dramatically during the injection molding process,unstructured grids must be partitioned very complex and meticulous to ensure the efficiency and stability of the simulation solution.For example,the surface layer is composed of several layers of fine hexahedrons,while the inner layer is divided into larger tetrahedrons.Complex unstructured grids greatly increase the difficulty of pre-and post-processing in three-dimensional simulation.Aiming at the key technology of three-dimensional simulation of injection molding process,the following work has been done in this paper:Complex unstructured grids are difficult to describe well because of the large scale of data,many cell types and complex topology.This brings great challenges to the realization of data access,graphics rendering and boundary condition setting in the pre-processing process of complex unstructured grids.In order to achieve efficient pre-processing module,this paper defines the general data structure and file reading interface of complex unstructured grids,realizes fast rendering of complex unstructured grids by extracting surface grids,and studies the acceleration of rendering of three-dimensional solid grids by using programmable rendering pipeline technology.On this basis,the function of setting boundary conditions for complex unstructured grids is finally realized.In post-processing,complex unstructured grids have a huge amount of data in the postprocessing field,and the internal structure of the data field is complex.The inefficiency of data processing algorithms or rendering process will cause serious performance problems.In order to render complex unstructured grids efficiently,the external rendering mode of VTK is used in this paper.On the basis of its efficient and stable data processing algorithm,the efficient rendering techniques for scalar field of complex unstructured grids such as color nephogram,isosurface and isochromatic band,three-dimensional symbolic graph of vector field and curve graph are studied,and the visualization functions such as cutting and sectioning are further realized.On this basis,the development of post-processing module for unstructured grids is completed.In the three-dimensional filling simulation module,the velocity-pressure coupled algorithm will produce a large-scale ill-conditioned block linear equation system,which is difficult to solve by algebraic multi-grid method.In this paper,the block characteristics of the equation system and the difficulties in solving them are analyzed in detail.A Schur complement preconditioned FGMRES algorithm is proposed.The velocity-pressure coupled equation is separated into velocity and pressure subsystems,and the subsystems are solved by block Jacobi and algebraic multigrid methods respectively,thus accelerating the convergence speed of iterative solution.Finally,the feasibility of the three-dimensional simulation system based on unstructured grid is verified by an application example,and the simulation accuracy of the threedimensional simulation system is verified by short shot experiment.In the final equation test,the Schur complement preconditioned FGMRES algorithm spends 47.4% less time than other algorithms such as BGS-FGMRES,which reflects the efficiency of the algorithm.The algorithm also maintains good stability and scalability in different scale grids.The parallel efficiency test further verifies the good parallel performance of the algorithm.