02 Fakultät Bau- und Umweltingenieurwissenschaften
Permanent URI for this collectionhttps://elib.uni-stuttgart.de/handle/11682/3
Browse
5 results
Search Results
Item Open Access Double U‐net : improved multiscale modeling via fully convolutional neural networks(2023) Lißner, Julian; Fritzen, FelixIn multiscale modeling, the response of the macroscopic material is computed by considering the behavior of the microscale at each material point. To keep the computational overhead low when simulating such high performance materials, an efficient, but also very accurate prediction of the microscopic behavior is of utmost importance. Artificial neural networks are well known for their fast and efficient evaluation. We deploy fully convolutional neural networks, with one advantage being that, compared to neural networks directly predicting the homogenized response, any quantity of interest can be recovered from the solution, for example, peak stresses relevant for material failure. We propose a novel model layout, which outperforms state‐of‐the‐art models with fewer model parameters. This is achieved through a staggered optimization scheme ensuring an accurate low‐frequency prediction. The prediction is further improved by superimposing an efficient to evaluate U‐net, which captures the remaining high‐level features.Item Open Access Material modeling for parametric, anisotropic finite strain hyperelasticity based on machine learning with application in optimization of metamaterials(2021) Fernández, Mauricio; Fritzen, Felix; Weeger, OliverMechanical metamaterials such as open‐ and closed‐cell lattice structures, foams, composites, and so forth can often be parametrized in terms of their microstructural properties, for example, relative densities, aspect ratios, material, shape, or topological parameters. To model the effective constitutive behavior and facilitate efficient multiscale simulation, design, and optimization of such parametric metamaterials in the finite deformation regime, a machine learning‐based constitutive model is presented in this work. The approach is demonstrated in application to elastic beam lattices with cubic anisotropy, which exhibit highly nonlinear effective behaviors due to microstructural instabilities and topology variations. Based on microstructure simulations, the relevant material and topology parameters of selected cubic lattice cells are determined and training data with homogenized stress‐deformation responses is generated for varying parameters. Then, a parametric, hyperelastic, anisotropic constitutive model is formulated as an artificial neural network, extending a recent work of the author extending a recent work of the author, Comput Mech., 2021;67(2):653‐677. The machine learning model is calibrated with the simulation data of the parametric unit cell. The authors offer public access to the simulation data through the GitHub repository https://github.com/CPShub/sim‐data. For the calibration of the model, a dedicated sample weighting strategy is developed to equally consider compliant and stiff cells and deformation scenarios in the objective function. It is demonstrated that this machine learning model is able to represent and predict the effective constitutive behavior of parametric lattices well across several orders of magnitude. Furthermore, the usability of the approach is showcased by two examples for material and topology optimization of the parametric lattice cell.Item Open Access Many‐scale finite strain computational homogenization via Concentric Interpolation(2020) Kunc, Oliver; Fritzen, FelixA method for efficient computational homogenization of hyperelastic materials under finite strains is proposed. Multiple spatial scales are homogenized in a recursive procedure: starting on the smallest scale, few high fidelity FE computations are performed. The resulting fields of deformation gradient fluctuations are processed by a snapshot POD resulting in a reduced basis (RB) model. By means of the computationally efficient RB model, a large set of samples of the homogenized material response is created. This data set serves as the support for the Concentric Interpolation (CI) scheme, interpolating the effective stress and stiffness. Then, the same procedure is invoked on the next larger scale with this CI surrogating the homogenized material law. A three‐scale homogenization process is completed within few hours on a standard workstation. The resulting model is evaluated within minutes on a laptop computer in order to generate fourth‐scale results. Open source code is provided.Item Open Access Residual stresses in Cu matrix composite surface deposits after laser melt injection(2023) Zhang, Xingxing; Kornmeier, Joana R.; Hofmann, Michael; Langebeck, Anika; Alameddin, Shadi; Alessio, Renan Pereira; Fritzen, Felix; Bunn, Jeffrey R.; Cabeza, SandraTungsten carbide particles reinforced metal matrix composite (MMC) coatings can significantly improve surface wear resistance owing to their increased surface hardness. However, the presence of macro‐ and micro‐residual stresses in MMC coatings can have detrimental effects, such as reducing service life. In this study, neutron diffraction was used to determine the residual stresses in spherical fused tungsten carbide (sFTC) reinforced Cu matrix composite surface deposits after laser melt injection. We also developed a thermo‐mechanical coupled finite element model to predict residual stresses. Our findings reveal that sFTC/Cu composite deposits produced with a preheating temperature of 400°C have low residual stresses, with a maximum tensile residual stress of 98 MPa in the Cu matrix on the top surface. In contrast, the sFTC/bronze (CuAl10Ni5Fe4) composite deposit exhibits very high residual stresses, with a maximum tensile residual stress in the Cu matrix on the top surface reaching 651 MPa. These results provide a better understanding of the magnitudes and distributions of residual stresses in sFTC‐reinforced Cu matrix composite surface deposits manufactured via laser melt injection.Item Open Access Reduced order homogenization of thermoelastic materials with strong temperature dependence and comparison to a machine-learned model(2023) Sharba, Shadi; Herb, Julius; Fritzen, FelixIn this work, an approach for strongly temperature-dependent thermoelastic homogenization is presented. It is based on computational homogenization paired with reduced order models (ROMs) that allow for full temperature dependence of material parameters in all phases. In order to keep the model accurate and computationally efficient at the same time, we suggest the use of different ROMs at few discrete temperatures. Then, for intermediate temperatures, we derive an energy optimal basis emerging from the available ones. The resulting reduced homogenization problem can be solved in real time. Unlike classical homogenization where only the effective behavior, i.e., the effective stiffness and the effective thermal expansion, of the microscopic reference volume element are of interest, our ROM delivers also accurate full-field reconstructions of all mechanical fields within the microstructure. We show that the proposed method referred to as optimal field interpolation is computationally as efficient as simplistic linear interpolation. However, our method yields an accuracy that matches direct numerical simulation in many cases, i.e., very accurate real-time predictions are achieved. Additionally, we propose a greedy sampling procedure yielding a minimal number of direct numerical simulations as inputs (two to six discrete temperatures are used over a range of around 1000 K). Further, we pick up a black box machine-learned model as an alternative route and show its limitations in view of the limited amount of training data. Using our new method to generate an abundance of data, we demonstrate that a highly accurate tabular interpolator can be gained easily.