The greatest challenge when using deep convolutional neural networks (DCNNs) for automatic segmentation of microstructural X-ray computed tomography (XCT) data is the acquisition of sufficient and relevant data to train the working network. Traditionally, these have been attained by manually annotating a few slices for 2D DCNNs. However, complex multiphase microstructures would presumably be better segmented with 3D networks. However, manual segmentation labeling for 3D problems is prohibitive. In this work, we introduce a method for generating synthetic XCT data for a challenging six-phase Al–Si alloy composite reinforced with ceramic fibers and particles. Moreover, we propose certain data augmentations (brightness, contrast, noise, and blur), a special in-house designed deep convolutional neural network (Triple UNet), and a multi-view forwarding strategy to promote generalized learning from synthetic data and therefore achieve successful segmentations. We obtain an overall Dice score of 0.77. Lastly, we prove the detrimental effects of artifacts in the XCT data on achieving accurate segmentations when synthetic data are employed for training the DCNNs. The methods presented in this work are applicable to other materials and imaging techniques as well. Successful segmentation coupled with neural networks trained with synthetic data will accelerate scientific output.

Cast near eutectic Al-Si alloys with addition of transition elements such as Cu, Fe, and Ni are commonly used materials in the aerospace and automotive industries. [1,2] The microstructure of these alloys is characterized by a 3D interconnected network formed by eutectic and primary Si and several Ni-, Fe-, and Cu-rich aluminides embedded in the Al matrix. [3-7] Under prolonged service time at high temperature (up to around 300-350 C), the aluminum matrix is overaged, what deteriorates its strength and creep properties. To improve the strength and creep resistance of these Al-Si alloys, additional ceramic reinforcements such as short fibers and particles can be used. [8-10] It has been shown that the micromechanical behavior of such composites strongly depends on the orientation of the fibers, the spatial distribution of the particles,

No abstract

The quality of components made by laser beam melting (LBM) additive manufacturing is naturally influenced by the quality of the powder bed. A packing density <1 and porosity inside the powder particles lead to intrinsic voids in the powder bed. Since the packing density is determined by the particle size and shape distribution, the determination of these properties is of significant interest to assess the printing process. In this work, the size and shape distribution, the amount of the particle’s intrinsic porosity, as well as the packing density of micrometric powder used for LBM, have been investigated by means of synchrotron X-ray computed tomography (CT). Two different powder batches were investigated: Ti–6Al–4V produced by plasma atomization and stainless steel 316L produced by gas atomization. Plasma atomization particles were observed to be more spherical in terms of the mean anisotropy compared to particles produced by gas atomization. The two kinds of particles were comparable in size according to the equivalent diameter. The packing density was lower (i.e., the powder bed contained more voids in between particles) for the Ti–6Al–4V particles. The comparison of the tomographic results with laser diffraction, as another particle size measurement technique, proved to be in agreement.

No abstract