Session Sheet - 2026 Annual International Solid Freeform Fabrication Symposium (SFF Symp 2026): Distortions, Microstructures, and Defects

2026 Annual International Solid Freeform Fabrication Symposium (SFF Symp 2026): Distortions, Microstructures, and Defects
Program Organizers: David Leigh, University of Texas at Austin

Monday 1:30 PM
August 3, 2026
Room: Guadalupe
Location: AT&T Center

1:30 PM
Bent Out of Shape – Perspectives on State-of-the-Art Distortion Prediction and Compensation Tools for Laser Powder Bed Fusion: Alex Riensche¹; Abdalla Nassar¹; ¹ASTRO America
     Many software tools advertise the ability to predict and correct component distortion in PBF-LB. However, without extensive and costly trials, end-users lack visibility into the functionality, accuracy, and value of available options. Here we present insights from the ASTRO-InSPIRE challenge on distortion prediction and compensation in PBF-LB wherein commercial solutions were assessed using real-world components with well-defined geometric requirements. Each solution was used to predict component distortion and, where possible, generate distortion-compensated geometry. Distortion prediction and compensation performance was compared to actual parts created with a commercial PBF-LB machine (Colibrium Additive M2 Series 5). Here, we present our evaluations of the solutions, including the accuracy, computational efficiency, usability, and value.

1:50 PM
Deflated Preconditioned Conjugate Gradient for Accelerating Ill-Conditioned Mechanical and Thermomechanical Process Simulations for Laser Powder Bed Fusion: Dhruba Aryal¹; Albert To¹; ¹University of Pittsburgh
    Thermomechanical and mechanical process simulations of large parts produced by Laser Powder Bed Fusion (LPBF) remain computationally prohibitive due to large number of time steps. Approaches such as layer-wise and scan track-wise flash heating as well as inherent strain methods, combined with agglomeration techniques, have been developed to reduce temporal resolution requirements. Even with these approximations, simulations of complex geometries and loosely constrained boundary conditions remain expensive, as the dominant bottleneck shifts to the solution of highly ill-conditioned linear systems. This work proposes to develop the Deflated Preconditioned Conjugate Gradient (DPCG) method within a GPU accelerated matrix-free Finite Element Method (FEM) framework with adaptive mesh refinement (AMR) to address these modes in both thermo-mechanical and inherent strain simulations. The proposed method combined with the Inexact Newton Method reduces iteration counts by over 50x – 100x in the problems considered, with increasing gains observed for finer meshes and greater ill-conditioning.

2:10 PM
Hierarchical Graph Neural Networks with Local Attention for Accelerated Part-Scale Scanwise Thermal Simulation in Laser Powder Bed Fusion: Berkay Bostan¹; Albert To¹; ¹University of Pittsburgh
    Scanwise thermal simulation in laser powder bed fusion (LPBF) is critical for defect prediction and process optimization but remains computationally prohibitive due to extreme spatial (micron-scale laser vs. millimeter–centimeter parts) and temporal (microsecond time steps) disparities. This work presents an accelerated framework for part-scale, scanwise thermal simulation of 3D components. The approach integrates two key modules: (1) a hierarchical graph encoder that captures both local and global geometric influences on thermal behavior and boundary conditions, and (2) a transformer-based decoder with local temporal attention that predicts nodal thermal histories using encoded geometry and time-dependent inputs such as laser trajectories and time-step sizes. The framework accurately handles complex geometries, including overhangs and in-plane variations, across simulations involving millions of nodes and time steps. It achieves approximately two orders of magnitude computational speedup while maintaining low prediction error, enabling efficient and scalable thermal modeling for realistic LPBF applications.

2:30 PM
Extending the Computational Fluid Dynamics Imposed Finite Element Method (CIFEM) to Multi-Layer Geometries for Laser Powder Bed Fusion: Sierra Stevenson¹; Seth Strayer¹; Albert To¹; ¹University of Pittsburgh
    Computational fluid dynamics (CFD) offers highly accurate thermal field predictions for laser powder bed fusion (L-PBF) but is intractable at the part scale. The finite element method (FEM) is much faster but conventional heat source models fail to capture the complex dynamics of the thermal field within L-PBF melt pools. This trade-off was addressed in the development of the CFD-Imposed Finite Element Method (CIFEM), in which a surrogate model trained on CFD simulation data rapidly predicts thermal fields and imposes them in the FEM simulation. This work improves the robustness and accuracy of CIFEM by extending the training dataset from single-layer, constant-length scan tracks to multi-layer and variable-length scan track geometries, allowing the model to better capture layerwise dynamics seen in real builds. A major modification to the calculation method of the local thermal environment, a key component of the CIFEM input vector, is also presented.

2:50 PM
Physics-Informed Neural Operator for Thermal Modeling in Metal Additive Manufacturing: Kaiwen Wu¹; Ajit Panesar¹; ¹Imperial College London
    Metal additive manufacturing involves complex thermal behavior, while conventional numerical methods and existing machine learning approaches often suffer from high computational cost, strong dependence on training data, and limited generalization capability. This study develops a physics-informed neural operator (PINO) framework for transient heat transfer modeling during the manufacturing process. The proposed framework generalizes across varying geometries, process parameters, including laser power and scanning speed, and material properties. By incorporating an initial-condition encoder, the model further enables thermal prediction across multiple deposition layers. Results demonstrate accurate temperature-field prediction with substantially reduced computational cost while maintaining robust performance under unseen manufacturing conditions without retraining. The proposed approach provides a unified and efficient framework for generalized thermal prediction in metal additive manufacturing and supports future process optimization and control.

3:10 PM Break

3:30 PM
Rapid Microstructure Prediction in Laser Powder Bed Fusion Using Bayesian Updating of Eagar–Tsai Model: Kyle Swartz¹; James Hanagan¹; Brent Vela¹; Doğuhan Sarıtürk¹; Raymundo Arroyave¹; ¹Texas A&M University
    Alloy design for laser-based welding and laser powder bed fusion (LPBF) requires rapid prediction of resulting microstructures across large composition–processing design spaces. The Eagar–Tsai (ET) model is a classical analytical thermal model whose low computational cost makes it attractive for high-throughput screening. Thermal gradients, G, and solidification rates, R, can be extracted from ET temperature fields and coupled with classical solidification models, such as Kurz–Giovanola–Trivedi (KGT) theory, to predict microstructural evolution. However, the simplified physics of the ET model, including temperature-invariant thermophysical properties, limits its predictive fidelity relative to finite element method (FEM) thermal simulations. In this work, we combine the speed of ET modeling with the accuracy of FEM simulations by treating ET predictions as Bayesian priors and updating them using FEM-informed thermal data to improve the prediction of as-solidified microstructures within KGT theory.

3:50 PM
Multi-modal Foundational Thermal Model for LPBF Part Reconstruction: Peter Pak¹; Amir Barati Farimani¹; ¹Carnegie Mellon University
     This work explores a method for predicting porosity through the reconstruction of a given part using melt pool temperature fields generated using a surrogate model. For this task a series of melt pool simulations were conducted using FLOW-3D with varying process parameters such as power, velocity, and incident angle. The simulations are utilized as training data for the surrogate model designed to produce melt pool temperature field prediction from select process parameters and top down melt pool images. Using the provided scan path, a reconstruction of the part is built with the predicted melt pool temperature field, highlighting potential areas of unfused powder resulting in porosity. Experimental computed tomography samples provide the ground truth to evaluate the system’s ability to reconstruct porosity mapping for the samepart.

4:10 PM
Spattering in Laser Powder Bed Fusion: Physics-based Simulations, Formation Mechanisms, and Reduction Strategies: Fangzhou Li¹; Haoran Shi¹; Wenda Tan¹; ¹University of Michigan
    Powder spattering in laser powder bed fusion (LPBF) processes leads to multiple quality-critical defects in the additively manufactured parts, and effective strategies for spatter reduction are needed for process optimization. In this work, a multi-physics model that incorporates computational fluid dynamics and computational powder dynamics is developed to simulate the keyhole evolution, vapor plume dynamics, and the powder spattering during LPBF. The model is validated against in-situ observation of the powder motion in LPBF, and leveraged to quantitatively understand the formation mechanisms of spatters. The spatial distributions of the spattering particles were found from the simulations, which offer a new perspective to quantify the spattering behavior of the particle. Inspired by this new perspective, different beam shapes have been selected to reduce spatters in LPBF, and their effectiveness has been investigated.

4:30 PM
Quantifying Process Uncertainty in Laser Powder Bed Fusion: A Modeling Approach for Surface Topography Prediction: Kubra Sekmen¹; Pinar Acar¹; Bart Raeymaekers¹; ¹Virginia Tech
    In Powder Bed Fusion – Laser Beam (PBF-LB), process parameters such as laser power and scan speed are treated as constants, overlooking their inherent variability in real manufacturing environments. This study presents a modeling framework that incorporates process uncertainty into surface topography parameters prediction. Surface measurements of IN718 specimens were used to characterize variability in these properties. In the absence of direct process parameter measurements, multiple distribution types were assumed to represent uncertainty in power and scan speed. These uncertainties were embedded in a non-intrusive polynomial chaos expansion model relating process parameters to surface topography parameters. We demonstrated that uncertainty-aware models offer robust prediction with minimal accuracy loss, despite variations in input distributions. Our approach provides an improved understanding of process–surface relationships under uncertainty. It offers a pathway toward predictive control of PBF-LB surface quality and can be extended to different materials, additional process parameters, and broader manufacturing conditions.

4:50 PM
Multiscale Computational Study of Laser Powder Bed Fusion Additive Manufacturing of Refractory TiTaNb Medium Entropy Alloy: Md. Solayman¹; Md Tusher Ahmed¹; AMM Nazmul Ahsan¹; ¹University of Texas Rio Grande Valley
    Refractory medium entropy alloys (RMEA) are promising candidates for ultra-high temperature applications. The laser powder bed fusion (LPBF) additive manufacturing of TiTaNb RMEA from its elemental powders remains largely unexplored. This study investigates whether a second laser pass, representative of re-scanning in LPBF, changes the microstructure of TiTaNb alloy formed during the first pass. CALPHAD equilibrium calculations, Scheil-Gulliver solidification modeling, and molecular dynamics simulations are combined to analyze TiTaNb during two consecutive laser passes. CALPHAD predicts BCC and HCP coexistence at room temperature, whereas Scheil modeling indicates rapid solidification into single-phase BCC. Molecular dynamics tracks thermal history, cooling rate, grain structure, and dislocation evolution after each pass. The second pass strongly remelts the first-pass structure, but the final microstructure changes only slightly, indicating stable BCC formation under repeated thermal cycling in LPBF fabricated TiTaNb.