2024 Annual International Solid Freeform Fabrication Symposium (SFF Symp 2024): LBPF Modeling
Program Organizers: Joseph Beaman, University of Texas at Austin
Monday 1:30 PM
August 12, 2024
Room: 417 AB
Location: Hilton Austin
Session Chair: Guha Manogharan, Pennsylvania State University
1:30 PM
Model Based Control of Fused Powder 3D Printing: Sam Stodder1; 1Stodder Engineering Services
Powder Bed Fusion 3D printing requires precision control of thermal energy deposited onto polymer powder particles in turn to control polymer phase transition at voxel resolution. With HP’s proprietary Multi Jet Fusion process (MJF), such control is delivered through printer parameters that adjust the printer’s thermal system with the application of radiant energy and print agents. Currently, identifying a set of optimal printer parameters requires users to develop process expertise. Embedding a MJF process Digital Twin in a model-based control system holds the promise to significantly simplify and automate the process optimization procedure. I have developed a thermal simulator of the MJF technology for this purpose. This simulator runs near real-time, solving a vectorized 3D multi-physics model using explicit finite difference method accelerated by GPU massive parallelism. It can be used as a “virtual sensor” outputting voxel level thermal and material phase states that are otherwise infeasible to measure experimentally.
1:50 PM
Determining Grain Size and Hardness in Laser Powder Bed Fusion Materials from Simulated Cooling Rates via Thermal Process Simulation: David Anderson1; Kevin Renteria2; Jorge Mireles2; Albert To1; Ryan Wicker2; 1University of Pittsburgh; 2University of Texas at El Paso
Achieving consistent microstructural properties of parts manufactured in laser powder bed fusion additive manufacturing processes remains a significant hurdle for process qualification in sensitive applications such as aerospace and nuclear industries. In this work, a means of predicting grain cell size and local microhardness based on part-scale simulations is presented. First, a stepped brick geometry is designed and utilized to evaluate microhardness and grain cell size at different degrees of residual heat accumulation, represented as a layerwise preheat temperature value. Then layerwise and local scanwise simulation models are built through a GPU-accelerated finite element method (FEM) based process simulation tool capable of efficient part-scale scanwise simulation. From these simulations, the local cooling rate near the melt pool is calculated, and a relationship with experimentally-determined microhardness and grain cell size is determined. The relationship is validated on an unseen geometry.
2:10 PM
Universal Differential Equations for Transient Thermal Modeling of Directed Energy Deposition Additive Manufacturing: Praveen Vulimiri1; Albert To1; 1University of Pittsburgh
Analytical thermal modeling approaches, such as the Green’s function, were some of the first methods to accurately predict the thermal field of a welding or additive manufacturing process. Compared with numerical approaches like finite element modeling, analytical models provide a very fast approximation but simplify the problem by using constant material properties and ignoring nonlinear boundary conditions (i.e. radiation). Using a technique known as universal differential equations, this work approximates the Green’s function with temperature-dependent material properties, complex geometry, and nonlinear boundary conditions using neural networks and integrates the function using ordinary differential equation solvers. The results show better agreement with finite element analysis data for Ti64 directed energy deposition simulations with computational time comparable with the analytical Green’s function.
2:30 PM
CIFEM: Elucidating the Role of Local Thermal Environment on Multi-Track Melt Pool Morphology Variation for Inconel 718 Laser Powder Bed Fusion: Seth Strayer1; Alaaeldin Olleak1; Praveen Vulimiri1; Shawn Hinnebusch1; William Frieden Templeton2; Florian Dugast1; Sneha Narra2; Albert To1; 1University of Pittsburgh; 2Carnegie Mellon University
Despite advancements in finite element (FE) thermal simulation techniques for laser powder bed fusion (L-PBF), these models employ a functional heat source model, which invokes a tedious calibration process and provides inaccurate thermal fields compared to high-fidelity computational fluid dynamics (CFD) simulations. Consequently, the driving force behind multi-track melt pool size variation has remained enigmatic up to this point. In this work, the authors extend CIFEM to multi-track scenarios for Inconel 718 L-PBF to help address these issues. CIFEM's data-driven heat source model is trained to predict the thermal fields from multi-track CFD simulations with different cooling times to establish the role of the local thermal environment. By imposing these fields on the desired FE solution domain, the simulated melt pool sizes are within 10% error regarding experimental measurements up to five consecutive tracks while providing substantially more accurate thermal fields to traditional FE models.
2:50 PM
Predicting Recoater Interference in Laser Powder Bed Fusion Process by Considering Both Global Thermal Deformation and Local Edge Deformation: Wen Dong1; Shawn Hinnebusch1; Albert To1; 1University of Pittsburgh
This work proposes an integrated simulation and experimental framework to predict potential recoater interference for a given part designed for L-PBF fabrication. The largest deformation in the build direction is assumed to occur at the edge of a part and is postulated to be the sum of two contributions: global thermal deformation and local edge deformation. The global thermal deformation, generated by relaxation of the thermal stresses induced by the rapid laser melting and solidification over the entire part, is predicted using the modified inherent strain (MIS) method. A key novelty in this work lies in employing location-dependent inherent strains (ISs) in the MIS method to simulate the global thermal deformation of overhangs, which shows 60% improvement in prediction accuracy compared with that using constant ISs. The validity of the proposed framework for predicting recoater interference is confirmed by experiments on different part geometries with overhangs.
3:10 PM Break
3:40 PM
Analytical Prediction of Texture of Multi-phase Material in Laser Powder Bed Fusion: Wei Huang1; 1Gatech
This paper first developed a physics-based analytical model to predict the texture of multi-phase materials related to the 3D temperature distribution in LPBF, considering boundary conditions, heat input using a point-moving heat source solution, and heat loss due to heat conduction, convection, and radiation. The texture grown on a substrate with random grain orientations was analytically acquired, considering the columnar-to-equiaxed transition (CET). The correlation between texture and process parameters has been effectively established using CET models and the second law of thermodynamics. Ti-6Al-4V was selected. With applied advanced thermal models, the accuracy of the texture prediction is evaluated based on the comparison to experimental data from literature and past model results, and higher accuracy is achieved. This study offers a quick and precise way of analyzing texture prediction in multi-phase mode for metallic materials. It lays the groundwork for future research on texture-affected materials' properties in academic and industrial settings.
4:00 PM
Unveiling Gas–Liquid Metal Reactions in Metal Additive Manufacturing: High-Fidelity Modeling Validated with Experiments: Hou Yi Chia1; Yanming Zhang1; Lu Wang1; Wentao Yan1; 1National University of Singapore
Gases such as oxygen inevitably react with the melt pool during metal additive manufacturing (AM). Excessive uncontrolled oxidation is detrimental, so most machines purge the chamber with inert gases, minimizing but not eliminating such reactions. Alternatively, some users exploit the gas–liquid metal reactivity to derive beneficial properties ("reactive AM"). However, the gas–liquid metal reaction and mechanisms remain unclear. This work clarifies the mass transfer process of oxygen during metal AM through high-fidelity modeling validated with experiments. Counterintuitively, higher temperatures do not necessarily lead to greater oxidation rates during processing. The melt pool has regions of high and low oxygen gains due to temperature-dependent reaction regimes, with concurrent oxygen loss from evaporation of metal suboxides. Thus, the net oxygen flux varies for different materials, and the oxygen content cumulatively changes as multiple tracks are scanned. Overall, we provide guidance to ameliorate or exploit the inevitable gas–liquid interaction in metal AM.
4:20 PM
Exploring the Influence of Geometrical Features on Mechanical Property Variation in LPBF Parts Through Simulation: Junyan He1; Alaa Olleak1; 1Ansys, Inc
Qualifying metal parts produced via Laser Powder Bed Fusion (LPBF) poses challenges due to the expense and time of experimental investigations. Parts qualification is crucial in industries like aerospace and biomedical to ensure high reliability and safety standards. Microstructure produced by LPBF process depends on thermal history. In-situ monitoring can capture unexpected process behavior, but handling the resulting data burden delays decision-making and necessitates further experimental iterations. In this study, simulation-based workflow to estimate cooling rate variability within parts is proposed. Firstly, a part-scale thermal model is developed using Ansys to predict the thermal history at different locations in Inconel 718 cantilevers. Secondly, the mechanical properties variation at different locations are quantified utilizing the EBSD maps from the NIST AMBench2022 dataset along with crystal plasticity models which are performed on microstructure samples to characterize the mechanical properties. Preliminary findings indicate spatial variations in mechanical properties, correlating with predicted thermal histories.
4:40 PM
Effect of Part Thermal History on Microstructural Evolution and Mechanical Properties in Stainless Steel 316L Laser Powder Bed Fusion: Prahalada Rao1; Kaustubh Deshmukh1; Christopher Williams1; Alex Riensche1; Benjamin Bevans1; Harold (Scott) Halliday2; 1Virginia Tech; 2Navajo Tech University
In this study, metal parts resembling tensile dog bones are additively manufactured via laser powder bed fusion, employing varied process parameters. Our primary objective is to elucidate the intricate relationship between process, structure, and properties in Stainless Steel 316L fabricated through this method. We accomplish this by elucidating and quantifying the impact of processing parameters and part-scale thermal history on microstructural evolution and mechanical properties. While previous studies have linked process parameters to flaw formation, microstructural features, and functional properties, a gap persists in understanding the underlying influence of thermal history. Thermal distribution, influenced by processing parameters, material properties, and part geometry, affects microstructure evolution and subsequently influences part properties. Thus, this paper's novelty lies in illuminating the process-thermal history-microstructure-property relationship in laser powder bed fusion.
5:00 PM
Determination of Optimal Beam Shapes in Laser Powder Based Fusion of Metals
-Verification and Validation: Vijaya Holla1; Philipp Kopp2; Jonas Grünewald1; Patrick Praegla1; Christoph Meier1; Katrin Wudy1; Stefan Kollmannsberger2; 1Technical University Munich; 2Bauhaus University Weimar
We present a framework for laser beam shaping in L-PBF/M that uses a heat conduction model to compute a laser intensity profile corresponding to a desired temperature field (i.e. melt pool shape). To solve this inverse problem, we minimize the functional given by the squared difference between the predicted and the desired temperature field integrated over the domain using the adjoint-based optimization method. The melt pool dimensions produced by the computed laser intensity profile are validated using photomicrographs of experiments. However, other quantities such as temperature profiles, temperature gradients, or dynamic effects cannot be validated experimentally. To this end, the computational results are compared to high-fidelity melt pool models (using Smoothed Particle Hydrodynamics).Additionally, we will discuss how the process windows within which the presented models are valid and present new types of laser intensity profiles designed to achieve a wide, shallow, yet stable melt pool in conduction mode.