Date of Award
2025-08-01
Degree Name
Doctor of Philosophy
Department
Computational Science
Advisor(s)
Natasha S. Sharma
Second Advisor
Zhengtao Gan
Abstract
Laser Powder Bed Fusion (L-PBF) is a well-established additive manufacturing technique for fabricating intricate metal components with exceptional precision. A significant challenge in L-PBF is the formation of complex microstructures that influence final material properties. We propose a physics-guided, machine learning-aided approach to optimize scan paths for desired microstructure outcomes, such as equiaxed grains. We employed a phase-field method (PFM) to model the evolution of the crystalline grain structure. To reduce computational costs, we trained a surrogate machine learning model, a 3D U-Net convolutional neural network, using single-track phase-field simulations with varying laser powers to predict crystalline grain orientations based on initial microstructure and thermal history. We investigated three scanning strategies across various hatch spacings within a square domain, achieving a speed-up of three orders of magnitude using the surrogate model. To reduce trial and error in designing laser scan toolpaths, we use deep reinforcement learning (DRL) to generate optimized scan paths for target microstructure. The results of three cases demonstrate the effectiveness of the DRL approach. We integrate the surrogate 3D U-Net model into our DRL environment to accelerate the reinforcement-learning training process. The reward function minimizes both aspect ratio and grain volume of the predicted microstructure from the agent's scan path. This presents the first DRL framework that directly minimizes phase-field predicted grain metrics during scan path design. The reinforcement learning algorithm, benchmarked against conventional zigzag approach for smaller and larger domains, demonstrates machine learning methods' potential to enhance microstructure control and computational efficiency in L-PBF optimization.
Language
en
Provenance
Received from ProQuest
Copyright Date
2025-08
File Size
145 p.
File Format
application/pdf
Rights Holder
Augustine Twumasi
Recommended Citation
Twumasi, Augustine, "Laser Scan Path Design For Controlled Microstructure In Additive Manufacturing With Integrated Reduced-Order Phase-Field Modeling And Deep Reinforcement Learning" (2025). Open Access Theses & Dissertations. 4488.
https://scholarworks.utep.edu/open_etd/4488