Integration Of Physical Model And Data-driven Approach For Numerical Modeling Of Additively Manufactured Metallic Materials

Additive manufacturing(AM),which fabricates materials through incremental method rather than removal method,offers great design flexibility,low materials waste,and capability for dealing with parts with complicated geometries.Therefore,AM has been considered as a promising candidate to revolutionize the manufacturing industries.Titanium and nickel alloys have been successfully extended to many advanced applications,especially in the aerospace industry and biomedical field,owing to their outstanding combinations of high specific strength,excellent corrosion resistance and fracture toughness.With the development of 3D printing technique,the research of titanium and nickel alloys becomes hotspots.Abundant researchers focus on the theoretical studies and applications of metallic AM.Many experiments have found that a mass of coarse columnar grain structures and textures,cracks,keyholes as well as pores were observed during AM of titanium,nickel and aluminum alloys.However,there is a lack of fundamental knowledge on the formation mechanisms of these microstructures and defects during AM process.Also,the study of quantitative relation between microstructure and mechanical property has never been done.In this paper,we develop a“Tri-prism”element to discretize domain with complex geometry structures,and build a grain growth phase-field model(PFM).By combining with the fast Fourier transform-based elasto-viscoplastic model(EVP-FFT),we propose a computational framework for the systematic investigation of process-structure-property relationship during AM fabrication of metals.Finally,we build a computational platform,integrating the presented physical model,data-driven method and experimental data,to guide the selection of manufacturing parameters during metallic AM fabrication.As a result,the following studies are implemented in this paper:(1)A linear six-node“Tri-Prism”element is presented based on the gradient smoothing technique.The transient temperature fields are also analyzed during Selective Electron Beam Melting(SEBM)of Ti-6Al-4V alloy.It has been found that increasing electron beam currents,i.e.,beam power,and/or decreasing travel speeds contribute to improve the temperature and melt pool dimensions.The maximum temperature remains stable in different building layers during SEBM,while the maximum thermal gradient declines gradually with the increment of building height.(2)A grain growth phase-field model is developed.The formation mechanisms of coarse columnar grain structures and textures are investigated during SEBM fabrication of titanium and nickel alloys.In order to obtain the fine equiaxed grain structures,we further develop two scanning strategies,i.e.,line melting and spot melting,and study the mechanisms of columnar-to-equiaxed transition(CET)of grain structures.It has been found that the development of the predominant grain structures in the thick and thin walls,i.e.,the large vertical columnar<001>_?//N_zgrains and slanted columnar grains with various grain orientations,respectively,can be attributed to the competition and collaboration between the thermal gradient and the crystallographically preferred grain growth.Increasing scanning speed and/or rising preheating temperature are found to reduce the temperature gradient and increase the solidification rate,thus leading to the large undercooling with a high probability of heterogeneous nucleation that contributes to the CET of grain structures.(3)A computational framework,integrating finite-element method(FEM),grain growth PFM and EVP-FFT calculation,is presented here.The proposed model provides the systematic investigation of process-structure-property relationship during AM fabrication of Ti-6Al-4V alloy.It has been found that the equiaxed?-grain structures have the smaller grain size and the shorter slip length,and thus provide much higher yield stress,up to?130 MPa,than that of large columnar?-grain structures.For the columnar?-grain structures,the reduction of elongation in R_x-direction is attributed to a mass of stress concentration along the grain boundaries,which could result in the crack nucleation and propagation as the stress exceeds the ultimate strength of 978 MPa.(4)A computational platform,integrating the presented physical models,data-driven model and experimental data,is developed for the investigation of metallic AM.This model can build the quantitative relations between processing parameters and microstructures,as well as processing parameters and mechanical properties.Thus,it is great significant to guide the manufacturing process.