This study proposes a novel framework called ‘Micro3Diff’ for 2D-to-3D reconstruction of microstructures using diffusion-based generative models (DGMs). Specifically, this approach solely requires pre-trained DGMs for the generation of 2D samples, and dimensionality expansion (2D-to-3D) takes place only during the generation process (i.e., reverse diffusion process). The proposed framework incorporates a concept referred to as ‘multi-plane denoising diffusion’, which transforms noisy samples (i.e., latent variables) from different planes into the data structure while maintaining spatial connectivity in 3D space. 

A novel defect detection framework for the laser powder bed fusion (L-PBF) process with a three-dimensional convolutional neural network (3D-CNN) and in-situ monitoring of light intensities in three-dimensional space was developed. The proposed model performs both the classification of the defect (lack-of-fusion or keyhole induced porosity) and the prediction of local volume fraction simultaneously. The efficient pre-processing of photodiode signals and the sampling method with micro-CT data were proposed to aid the defect detection using the 3D-CNN model.

A novel hybrid heat source model is developed considering the different absorption mechanisms for porous and dense state materials, and an effective absorptivity is adapted to the proposed model to analyze the melting mode transition. The proposed model can predict the melt pool characteristics including the melt pool dimensions and the melting modes in the selective laser melting (SLM) process. The problem is formulated using the heat transfer equation considering the phase transition and the degree of consolidation based on the phase-field approach.

In this study, an optimization methodology for designing a thermally conductive lattice support structure is proposed considering the layerwise heating and cooling in the PBF process. A part-scale numerical model with homogenized properties of lattice material is constructed using the temperature-thread multiscale modeling approach to simulate the transient temperature field in PBF process. The transient analysis model is then integrated into the lattice structure topology optimization (LSTO) with the equivalent static loads method (ESLM) based sensitivity analysis to derive the optimal density profile of the support.

This paper presents a framework for reconstructing anisotropic microstructures solely based on two-dimensional (2D) micrographs using conditional diffusion-based generative models (DGMs). The proposed framework involves the spatial connection of multiple 2D conditional DGMs, each trained to generate 2D microstructure samples for three orthogonal planes. The connected multiple reverse diffusion processes then enable effective modeling of a Markov chain for transforming noise into a 3D microstructure sample. Furthermore, a modified harmonized sampling enhances the sample quality while preserving the spatial connection between the slices of anisotropic microstructure samples in 3D space. 

• An integrated microstructure design framework utilizing the diffusion-based generative model (DGM) is proposed.

• The problem of microstructure reconstruction is addressed using the DGM with forward/reverse Markovian diffusion processes.

• The conditional formulation of DGM is introduced to enable the inverse design of multifunctional composites.