TY - CONF A1 - Evsevleev, Sergei A1 - Paciornik, Sidnei A1 - Gollwitzer, Christian A1 - Bruno, Giovanni T1 - Application of deep learning to the segmentation of synchrotron X-ray tomography data of multiphase metal matrix composites N2 - The 3D microstructure of an Al alloy matrix composite with two ceramic reinforcements was investigated by synchrotron X-ray tomography. A deep learning algorithm was used for the segmentation of four different phases. We show that convolutional networks with the U-Net architecture are able to solve complex segmentation tasks with small amount of training data. T2 - International Conference on Tomography of Materials & Structures CY - Cairns, Australia DA - 22.07.2019 KW - Synchrotron X-ray tomography KW - Deep learning KW - Segmentation KW - Metal matrix composite PY - 2019 AN - OPUS4-48606 LA - eng AD - Bundesanstalt fuer Materialforschung und -pruefung (BAM), Berlin, Germany ER - TY - CONF A1 - Evsevleev, Sergei A1 - Mishurova, Tatiana A1 - Sevostianov, Igor A1 - Garcés, Gerardo A1 - Requena, Guillermo A1 - Cabeza, Sandra A1 - Bruno, Giovanni T1 - Load transfer and damage evolution in multiphase Al alloy matrix composites under uniaxial compression N2 - Aluminum alloys are extensively used in the automotive industry. Particularly, squeeze casting production of Al-Si alloys is employed in the conception of metal matrix composites (MMC) for combustion engines. Such materials are of a high interest since they allow combining improved mechanical properties and reduced weight and hence improve efficiency. We investigate two types of MMCs, which can be potentially used for production of combustion engine pistons: 1) a near-eutectic cast AlSi12CuMgNi alloy reinforced with 15%vol. Al2O3 random planar oriented short fibers and 2) the same alloy reinforced with 7%vol. Al2O3 random planar oriented short fibers + 15%vol. SiC particles. Complex 3D microstructure of the samples in as-cast condition, consisting of four and five phases (Al matrix, eutectic Si, intermetallics, Al2O3 fibers and SiC particles) was investigated by synchrotron computed tomography (CT). Advanced methods based on machine learning were applied for segmentation of all phases. This allowed extracting quantitative information such as volume fraction, shape and interconnectivity of every phase. In-situ compression tests during neutron diffraction experiments were used to track the load transfer among phases, while CT on pre-strained samples to monitor and quantify damage. A micromechanical model was developed to simulate the evolution of the internal stress of each phase during uniaxial compression. We showed that in composites with the Al2O3 fiber mat plane perpendicular to the load axis the Al-alloy matrix presents a large hydrostatic stress component, i.e. undergoes compression also in the direction transverse to the external load. This feature holds with and without the addition of SiC ceramic reinforcement and is absent in the case when the Al2O3 fiber mat plane is parallel to the load axis. We show, that the intermetallics play a decisive role at very high loads, when all other reinforcement phases suffer (extensive) damage. The addition of SiC particles does alleviate the load on the Al2O3 fibers, on the eutectic Si, and on the intermetallic phases in both cases of parallel and orthogonal (to the load axis) Al2O3 fiber orientation. Apart from the beneficial addition of ceramic reinforcement, the presence of intermetallic phase and eutectic silicon, forming an interconnected network even at high loads, when single particles break, confers peculiar properties to these multi-phase composites. T2 - The 3rd International Conference on Light Materials CY - Manchester, UK DA - 05.11.2019 KW - Aluminum alloy KW - Metal Matrix Composite KW - Computed tomography KW - Neutron diffraction KW - Stress analysis KW - Load partition PY - 2019 AN - OPUS4-49640 LA - eng AD - Bundesanstalt fuer Materialforschung und -pruefung (BAM), Berlin, Germany ER - TY - CONF A1 - Evsevleev, Sergei A1 - Mishurova, Tatiana A1 - Cabeza, Sandra A1 - Garcés, Gerardo A1 - Requena, Guillermo A1 - Sevostianov, Igor A1 - Bruno, Giovanni T1 - Stress partitioning and damage evolution in near-eutectic cast AlSi12CuMgNi alloy N2 - Near eutectic cast Al–Si alloys are the basis of all Al-alloys used for pistons due to their high fluidity and relatively high strength-to-weight ratio. Their microstructure is characterized by the presence of eutectic-Si embedded in an age hardenable -Al matrix. During solution heat treatment, the eutectic Si network undergoes a process of gradual disintegration, reducing the load bearing capability and the strength of the alloy. The improvement of the high temperature strength is achieved by addition of transition elements such as Cu, Mg or Ni through the formation of stiff intermetallic (IM) phases. In present study a near-eutectic AlSi12CuMgNi alloy produced by squeeze casting was investigated in as-cast condition. The in-situ neutron diffraction compression tests revealed the stress partitioning mechanism between phases of the alloy. Large stresses were found in IMs, showing their significant role as a reinforcement elements. After the failure of IM and Si phases the load is transferred back to the Al matrix, inducing further plastic deformation in it, yet without creating microcracking. The neutron diffraction experiments were coupled with assessment of internal damage after ex-situ compression tests by synchrotron radiation computed tomography. The quantitative analysis of CT data revealed that local tensile stresses in IM and Si phases induce microcracks with preferential orientation, parallel to the axis of applied load. Altogether, present experimental data was used as an input for the developed micromechanical model based on Maxwell homogenization scheme. In contrast to the neutron diffraction experiment, which can only determine stress differences between the axial and radial sample directions, the model allows predicting the principal stresses in each phase of the alloy. T2 - European Conference on Residual Stresses - ECRS10 CY - Leuven, Belgium DA - 11.09.2018 KW - Aluminum alloy KW - Intermetallics KW - Computed tomography KW - Neutron diffraction KW - Stress analysis KW - Load partition PY - 2018 AN - OPUS4-46054 LA - eng AD - Bundesanstalt fuer Materialforschung und -pruefung (BAM), Berlin, Germany ER -