A Study On Temperature Oscillation And Mesh Generation During Forging Process Simulation

The basis of Finite element method is dispersing the continuous substance into finiteelements, which transform a problem with infinite degrees of freedom into a problem withfinite degrees of freedom, and then simulated results are calculated by numerical methods.Nowadays, FEM plays a very important role on both theoretical study and engineeringapplication with the development of computer science, it becomes a very effective methodin the field of numerical analysisAiming at improving a self-developed software, this paper studies on two keytechnologies:The first one is retraining the oscillation in numerical simulation of temperature field.The simulation of temperature is an important content in the plastic forming FE, becauseits results influence the accuracy during the process of material flow. During thesimulation of temperature field, it happens that the calculated results deviate from the truevalues and oscillate, which affects the calculation or even leads to an error. This paperstudies on the possible causes of temperature oscillation in numerical simulation oftemperature field through theoretical analysis and calculation examples: The continuity oftemperature derivative is used, but not guarantied in FEM. The initial condition of uniformtemperature is against the boundary condition of heat exchange with external environment.Through numerical examples, the effect of both setting a reasonable initial temperaturefield and using the lumped heat capacity matrix in eliminating the temperature oscillat ionand their error sources are discussed. The results show that both setting a reasonable initialtemperature field and using the lumped heat capacity matrix can eliminate the temperatureoscillation, the former one has a better accuracy but more difficult to achieve.The second is mesh generation in FEM. This paper designs a flow chart of tetrahedronmesh generation subroutine, it begins with the reading of geometric information, thensearches the feature points, disperses the edges, the surfaces and the volume, finally outputthe mesh information. This paper studies on the algorithms used during the process above,such as Newton’s method, AFT, Delaunay triangulation etc. Considering that a pentahedron maybe cannot be dissected into tetrahedral using the AFT, and it’s complex torecovery the boundary information using the Delaunay triangulation, this paper combinesthe AFT and the Delaunay triangulation, using the former with the mapping methoddisperse the surfaces and the latter disperse the volume quickly. This method do not onlyavoid the inherent defects in the two algorithms respectively, but also improve the meshquality greatly.