A 3D solidification model was developed to predict the solidification grain structure at the mesoscopic level and the induced chemical segregation at the casting scale. The model introduces an innovative new coupling scheme between two numerical modeling approaches: a Finite Volume (FV), to solve the flow, heat and mass transfer, and a Discrete Particle, (DP), to mimic grain structure development. The coupled FV-DP model was built using the OpenFoam open source solver library. A series of validation benchmarks were used to demonstrate the model capability to predict the grain structure and to deal with mass exchange induced by the grain envelopes growth. The model has been applied to predict the temperature, grain structure and final segregation map of an Inconel 718 casting. The developed coupling scheme is shown to be a more accurate tool for quantitative prediction of the segregation in a superalloy casting when compared to single FV modeling approach. Introduction A typical vacuum induction melted (VIM) superalloy cast ingot manifests a complex solidification grain structure fabric constituting equiaxed and columnar grains morphology, columnar to equiaxed transition (CET), and variable dendritic arms spacing. This fabric is the result of grain envelopes interdependence with the heat, mass, and bulk flow [1, 2] and therefore macrosegregation [3]. Nastac et al conducted an experimental and numerical study to correlate the macrosegregation in the as-cast state to the cooling rate for an Inconel 718 ingot [4]. However, in their numerical analysis the grain structure was not solved and therefore its effect was neglected. This is the case of both standalone Finite Volume and Finite Element (FE) models which offer limited quantitative predictions of the chemical components distribution. Numerically modeling the physics at the mesoscopic level of solidification grain structure envelope growth, morphology and incorporating it in solving the macro-segregation at the casting scale is a challenge [5]. One particular approach is often used which is based on the coupling of FE and Cellular Automaton (CA) models [6]. These CAFE models have been improved over the years since first developed two decades ago [7]. Their key feature, which is the ability to couple the mushy zone growth to the FE balance equations, has been extended to 3D [8]. Yet, the basic idea and scheme remains the same; two discrete meshes of the fluid domain are created, one used for FE model and a second regular cubic cells based lattice for the CA at the micro scale. Furthermore, a two-way interpolation and mapping of the different key problem physical quantities (temperature, solute composition, etc.) is done to convey the 8th International Symposium on Superalloy 718 and Derivatives Edited by: Eric Ott, Anthony Banik, Xingbo Liu, Ian Dempster, Karl Heck, Joel Andersson, Jon Groh, Tim Gabb, Randy Helmink, and Agnieszka Wusatowska-Sarnek TMS (The Minerals, Metals & Materials Society), 2014
Read more