Numerical Modelling of Heat Treatment and Post Processing Following Additive Manufacturing of Metal Parts

A part produced by laser-based powder bed fusion (LPBF) normally requires postprocessing, for example to improve its residual stress state or material properties. A stress relaxation heat treatment is used to reduce the residual stresses in a part, and the resultant deformation when it is cut from the base plate.

Heating at a higher temperature can cause a change in the microstructure. Finding the optimal processing conditions for these thermal post-processes can require a large amount of experiments. However, a wellcalibrated numerical model can be a useful tool for finding suitable process conditions. For the present work, five different models were developed, which investigate different aspects of the LPBF process chain.

Two of the models focus on macro-scale simulation of a LPBF process chain, consisting of the primary LPBF process, a heat treatment, and a cutting operation to remove the part from the base plate. Both models are implemented in commercial finite element software. The first one is a simplified investigation, capable of showing the effect of the order of the post-processing steps, but it is relatively insensitive to heat treatment temperature and dwell time.

The second model uses a creep equation to imitate the time and temperature dependent evolution of the residual stresses. The model is validated using experimental results, and subsequently used to produce a process map for a stress relaxation heat treatment, which can be used directly to find the optimal material properties.

Two additional studies are presented that aim to demonstrate possible ways to couple models at different length scales. The first of these is a one-way coupling between a singletrack mesoscale model for LPBF, and a microstructural model, which uses CA for a subsequent heat treatment. These two parts are coupled via the grain morphology following from the mesoscale simulation.

The main benefit of this model is that it requires little additional computational cost to couple the two simulations. The second model uses two-way coupling between a macroscale model for the LPBF process and a JMAK-based microstructural model. However, coupling two well-established models together allows investigation of the effect of the intrinsic heat treatment on the residual stress state after the LPBF process.

The two-way simulation shows that inclusion of the intrinsic heat treatment can only be justified for high fidelity simulations, or when the precise composition of the microstructure is important for further processing of the part. Finally, in the present work, a model is developed to simulate the evolution of the β-transus temperature in a part produced in Ti-6Al-4V using LPBF.

The model uses the cellular automata method to imitate microstructural change, and thermal solver to obtain the temperature field during the heat treatment. Additionally, during the second half of the heat treatment, namely while cooling down from the high temperature to room temperature, a correction factor is added to account for the difference between growth direction of the Widmanstätten laths and the orientation of the observation plane for the micrograph.

Comparing the outcome of the microstructural model growth with JMAK kinetics shows that the grains nucleate in a narrow temperature range, resulting in steps in the nucleation profile. During the second part of the heat treatment, the simulated micrographs show that it is possible to obtain a basketweave-type grain morphology without nucleating laths in the centre of the grain.