2023 Annual International Solid Freeform Fabrication Symposium (SFF Symp 2023): Modeling: Uncertainty and Thermomechanical
Program Organizers: Joseph Beaman, University of Texas at Austin
Monday 1:30 PM
August 14, 2023
Room: 416 AB
Location: Hilton Austin
Session Chair: Guha Manogharan, Pennsylvania State University
1:30 PM Cancelled
Uncertainty Quantification in Laser Powder Bed Fusion from Mesoscale to Part Scale: Daniel Moser1; Nicole Aragon1; Helen Cleaves1; Michael Heiden1; Jeffrey Horner1; Kyle Johnson1; Mario Martinez1; Aashique Rezwan1; Theron Rodgers1; David Saiz1; Michael Stender1; 1Sandia National Laboratories
Laser powder bed fusion (LPBF) is an inherently multiscale process with detailed physics at the powder particle scale impacting part scale behavior. Quantifying uncertainties in LPBF process models is an important step in building model credibility so results can be used to support decision-making, particularly for qualification. However, applying uncertainty quantification to the multiscale physics models needed to simulate the LPBF process is challenging as bridging length scales introduces model form uncertainty that is difficult to represent. The work investigates techniques for quantifying uncertainties between powder scale and part scale LPBF models in order to make credible predictions with uncertainty of the outcomes of the LPBF process, particularly part distortion and microstructural features. Predictions are compared with experimental results to assess the performance of the developed techniques.This work was supported by the LDRD program at SNL, managed and operated by NTESS under DOE NNSA contract DE-NA0003525.
1:50 PM
Evaluating Residual Heat-driven Melt Pool Variation through GPU-based Thermal Process Simulation: David Anderson1; Haolin Zhang1; Shawn Hinnebusch1; Xiayun Zhao1; Albert To1; Florian Dugast1; Alaa Olleak1; 1University of Pittsburgh
Despite well-studied process parameters, melt pool variation due to residual heat accumulation has remained a potential source of porosity and spattering defects. However, at the part-scale, traditional finite element methods require extremely fine meshes relative to the part scale to accurately resolve melt pool dimensions, resulting in long processing times and high memory requirements. To resolve this, a matrix-free finite element model has been developed, implementing adaptive remeshing and GPU-based parallelization to improve processing times. The model has been calibrated against optical microscopy data to improve the implementation of a Goldak heat source model, and compared against in-situ monitoring data to validate simulated results. This model enables investigation into melt pool variations at the part scale due to residual heating effects from previous layers, part geometry, and hatching patterns, to better understand and predict transitions from conduction to keyhole regimes.
2:10 PM
Modeling Morphological Development of Ti-6Al-4V for Cyclic Thermal Histories in Laser Powder-bed Fusion: Evan Adcock1; Anthony Rollett1; 1Carnegie Mellon University
Laser powder-bed fusion (LPBF) introduces rapid temperature changes and cyclic thermal histories impacted by printing parameters and part geometry. This research aim is to develop a model that can efficiently compute predicted morphological outcomes of Ti-6Al-4V from LPBF temperature histories. Baseline microstructure results have been analyzed in an “inverted pyramid” part that was developed to generate variability in thermal history throughout the build. The geometry induced elevated preheat temperatures in subsequent layers which result in different cooling rates and thermal cycling. More data for validation will be gathered in future builds from process monitoring with a controlled stage preheat. Simultaneously a thermal simulation has been performed to provide thermal history input to the microstructure model. Comparison of the thermal histories with observed α morphology suggests the model can be simplified to predict microstructure based on the cooling rate after the last peak above the temperature of α dissolution.
2:30 PM
Residual Stress Estimation in a Complex Additively Manufactured Component with an Internal State Variable Material Model: David Failla1; Matthew Dantin2; Chuyen Nguyen1; Matthew Priddy1; 1Mississippi State University; 2Naval Surface Warfare Center
Internal state variable models are well suited to predict the effects of an evolving microstructure as a result of the additive manufacturing (AM) process in components with complex features. As AM becomes more utilized, accurate methods for predicting residual strains grow in need. To this end, the evolving microstructural model of inelasticity (EMMI) is adapted to modeling these residual strains due to its ability to capture the evolution of rate- and temperature-dependent hardening and softening as a result of the rapid thermal cycling present in AM processes. The current effort contrasts the efficacy of using EMMI with an elastic-perfectly plastic material model to predict the residual strains for an IN718 component produced via laser powder bed fusion. Both constitutive models are used within a thermo-mechanical finite element framework and are validated by published neutron diffraction measurements to demonstrate the need for high-fidelity models to predict residual strains in complex components.
2:50 PM
Multi-load Support Optimization for Minimizing Part Deformation in LPBF: Subodh Subedi1; Dan Thoma1; Krishnan Suresh1; 1University of Wisconsin Madison
Support structures act as primary conduits for heat flow in laser powder bed fusion (LPBF). Truss-type supports have proven to be a good choice for LPBF because of no metal powder entrapment and ease of removal. In a typical build process, these supports experience time-varying thermal and structural loads. We propose a multi-load strategy for optimizing truss-type support structures for minimizing part deformation. Equivalent static loads (ESLs) are computed using the inherent strains induced at each layer of deposition. These ESLs are applied as multi-loads to optimize the cross-section area by minimizing the compliance of the truss-type supports. The results demonstrate a reduction in part and support deformation when the optimized supports are used instead of un-optimized ones.
3:10 PM Break
3:40 PM
Process Modeling of Multi-Material Laser Powder Bed Fusion: Jacklyn Griffis1; Kazi Shahed1; Chinedum Okwudire2; Guha Manogharan1; 1Pennsylvania State University; 2University of Michigan
Thermomechanical simulation of the laser powder bed fusion process has been a valuable tool to help researchers and practitioners across the AM production cycle. For instance: Design for AM (DfAM), material development, process mapping, prediction, and support generations, among others. In this study, multi-material laser powder bed fusion (MM-LPBF), specifically of 904LSS and CuSn10 are examined through process simulation and non-destructive techniques to determine the impact of component orientation on defect mitigation. It is determined that material orientation along the build direction is a large contributor in as-build defects. Introductory MM-LPBF simulation is established to better understand the capabilities of current LPBF simulation tools in accurately predicting and mitigating the new challenges of MM-LPBF simulation.
4:00 PM
Generation and Simulation of Layer Wait Time to Prevent Overheating: Zack Francis1; 1Ansys
Control over thermal conditions during additive manufacturing is key to producing reliable parts consistently. Overheating during the build process leads to keyholing and increased variability and uncertainty in downstream outcomes involving part quality. One of the simplest methods to gain control over the thermal conditions through a part is to implement additional wait time between layers to allow the part to cool to a consistent temperature. This can be accomplished through “ghost parts” or other means of manipulating the scan time. Simulation can be a powerful tool by enabling users to determine the appropriate cooling time to reach a target temperature as well as simulate the effects of such a wait time to ensure that parts are reaching acceptable temperatures for the process parameters. This work will demonstrate how simulation can be used to simulate thermal conditions during a build as well as generate solutions to prevent overheating.
4:20 PM
Towards Large-scale Grain Growth Modeling in Powder Bed Fusion: Michael Paleos1; Albert To1; 1University of Pittsburgh
Grain growth models in the context of additive manufacturing are dealing with both the inherent complexity of the process and the computational expense of thermal process simulations. Cellular automata models have been successful in approximating the true physics of melt pool solidification, but they are typically confined to relatively small spatial domains. Building on recent advances in powder bed fusion process and microstructure modeling, we propose an integration framework based on several computational schemes that can lead to accurate simulations on unprecedented scales. For that purpose, we leverage and properly combine both the recently developed matrix-free FEM-based PAMSIM process simulator and the open-source ExaCA software. Our work centers around efficiently capturing information about several thermal signatures that would then guide grain growth in a decoupled manner. This framework would enable the computational study of microstructure (and property) heterogeneity and of the effect of unconventional scanning strategies.