The multiple reheat thermal history generated by the welding and additive manufacturing (AM) processes is one of the key factors determining the phase transformations and microstructure, and hence the mechanical properties and service performance of welded joints and additive builds. The conventional trial-and-error approach in process development for welding and AM is based on experimental variation of process parameters and extensive characterization and mechanical testing efforts, and therefore is time labor, and materials intensive.
The objective of this work is to demonstrate a computational design of experiment (CDoE) framework for process-microstructure-property optimization in welding and wire-arc direct energy deposition (WA DED).
The CDoE framework utilizes design of experiment (DoE), finite element analysis (FEA), and postprocessing modules. The DoE module generates a matrix of welding or WA DED processes, defined by arc current, voltage, wire feeding and travel speeds, bead sequence, etc. The FEA module utilizes the SysWeldTM software to create and mesh FEA models of the welded joint or additive builds. To allow automatic meshing, the FEA model uses trapezoidal, parallelogram, and/or rectangular bead shapes.
The CDoE framework also includes a heat source calibration procedure. The parameters of the Golak’s double ellipsoid heat source are run through a DoE optimization procedure to identify heat sources that replicate fusion boundary geometries of actual welds or additive builds.
The DoE module runs a matrix of welding or WA DED processes through the FEA moule to predict the spatial distribution of multiple reheat thermal histories experienced in the weldments or additive builds. The post processing module utilizes the predicted thermal histories and predetermined relationships to generate contour maps of microstructure and mechanical properties.
The developed CDoE framework was demonstrated by performing process-microstructure-property optimization for a TBW process in gas tungsten arc weld overlays of Alloy 625 on Grade F22 steel. The DoE matrix included 27 processes defined by three variables, heat input, bead thickness, and heat input ratio between the first and second layer, and three variations levels - low, medium, and high.
The CDoE simulations generated microstructure and hardness maps for the heat affected zone (HAZ) of the Grade F22 steel substrate. The local microstructure was identified by the reheat sequences relative to the AC1 and AC3 temperatures as fresh martensite, tempered martensite, and mixture of fresh and tempered martensite. The local hardness was calculated by utilizing the predicted thermal histories with predetermined tempering response relationships for Grade 22 steel. The tempering efficiency of the 27 TBW processes was evaluated through comparing the HAZ content of fresh and tempered martensite, and the HAZ hardness.
The CDoE simulations were validated by comparing predicted thermal histories and hardness distributions with actually recorded thermal histories and hardness maps of weld overlays.
The developed CDoE process-microstructure-property optimization framework was demonstrated and validated for a temperbead weld overlay of Alloy 625 on Grade 22 steel. Its implementation can reduce the time and cost and improve the efficiency in process development and optimization for welding and additive manufacturing.