弗莱贝格工业大学

Reconstructing local stress fields in heterogeneous microstructures under non-linear, history-dependent loading remains a major computational bottleneck in multi-scale simulations. We propose a coupled LSTM-GNN framework that links the temporal and spatial aspects of local stress field reconstruction. A Long Short-Term Memory network encodes macroscopic stress-strain sequences into a compact hidden state that captures the path-dependent constitutive response, while a physics-informed Graph Neural Network reconstructs the spatially-resolved stress field at each time step. We introduce a relative weighting strategy with linear warm-up to balance the data-driven reconstruction loss and a discrete divergence-based equilibrium penalty. This resolves the scale mismatch that prevents fixed-weight formulations from converging in the elasto-plastic regime. The model is trained on 10,000 non-proportional loading paths applied to a periodic plate-with-a-hole microstructure and von Mises elasto-plasticity. The model achieves three orders of magnitude speedup over finite element simulations and generalizes to loading sequences twice the training length, with 3.0% cumulative error. Because the graph relies on mesh connectivity instead of the specific element type, one trained surrogate can be applied directly without retraining to meshes with different element types and to both coarser and finer resolutions, while in all cases reproducing the high-fidelity quad-element FE field used during training. Indeed, the message passing characteristics inherent to GNN and MeshGraphNet architecture render the model mesh-agnostic. Analysis of the LSTM hidden states suggests a low-dimensional structure related to the internal state variables of the constitutive model.

Grain boundary plane distributions are widely used to infer the mechanisms governing grain boundary formation in polycrystalline materials. We show that such interpretations are inherently ambiguous. Using a unified eight-parameter boundary distribution framework, we derive both the grain boundary character distribution (GBCD) and the grain boundary normal distribution (GBND) and identify two limiting cases of boundary network formation. We show that in macroscopically driven networks, e.g. controlled by strain, the crystal-frame GBND is given by a convolution of the specimen GBND with the orientation distribution function (ODF), whereas in crystallographically driven networks, e.g. controlled by anisotropic boundary energy, the specimen GBND is obtained by convolution of the crystal GBND with the ODF. This duality implies that anisotropy in the GBND may arise from macroscopic alignment effects rather than intrinsic crystallographic selection. Conversely, this relationship may be used to identify the dominant formation process in the measured microstructures. Evaluation of a wide variety of simulated microstructures confirms the theoretically predicted relationships between texture, GBND and GBCD. In particular, our examples confirm that neither the GBND nor the GBCD alone is sufficient for identifying grain boundary formation mechanisms.

Stochastic fracture network models are important tools for characterizing heterogeneity and seepage behavior in fractured rock masses in geotechnical engineering. To address the limitations of conventional models-poor generalization, high uncertainty in flow description, and susceptibility to distortion-this paper proposes a novel computational framework: a geometric-topologically constrained stochastic fracture network modeling method that accounts for the synergistic interplay between fracture geometry and topological connectivity. After validating the seepage reliability of the constrained model, uncertainty quantification analysis identifies the respective contributions of various geometric and topological parameters to fracture network connectivity efficiency. Furthermore, a practical modeling approach driven by dominant controlling factors is proposed. Results indicate that fractured rock samples exhibit pronounced dominant-flow characteristics. Fractures create preferential flow channels, accelerating gas migration from free space into the porous medium region. Uncertainty quantification reveals that the average number of connections per branch (CB), the average number of connections per fracture (CL), and the angle deviation of fracture (theta) are the core controlling factors governing system-scale seepage behavior. Fracture aperture (b), the number of X-type nodes (NX), and the number of fracture branches (NB) are identified as key influencing factors. Volumetric fracture intensity (P32) and the number of I-type nodes (NI) indirectly regulate seepage responses through interactions with other topological parameters. Based on these findings, a dominant-factor-driven modeling approach is proposed, which prioritizes accurate characterization of parameters controlling fracture network connectivity and the dominant seepage direction, while reasonably simplifying secondary factors that contribute less to seepage behavior. The fracture network model constructed using this new method achieves a consistency rate of approximately 96.6% with the reference model in terms of seepage behavior, demonstrating its robustness in effectively characterizing the baseline seepage response. These findings provide a practical solution for engineering-scale fractured rock mass modeling, offering a balance between accuracy and computational efficiency.

This study investigates the high–pressure gasification kinetics of a mixed–wood char (MWC) under simulated gasifier atmospheres in the pressurized drop tube reactor KIVAN, in order to generate transferable single–particle kinetics for biomass gasification. Gasification with CO2, H2O, their binary mixture, and a simulated gasifier atmosphere (CO2 + H2O + CO + H2) was examined at 15 bar and 850–950 ℃, covering conditions representative of internal circulation (INCI) gasifiers.Based on axial gas-composition measurements in the pressurized drop tube reactor KIVAN under entrained-flow conditions, conversion curves as a function of particle residence time were derived and evaluated using the random pore model (RPM) combined with Arrhenius and power-law expressions to determine kinetic parameters. For gasification in pure CO2, MWC exhibits an activation energy EA = 173 kJ/mol, while under H2O EA = 144 kJ/mol is obtained, confirming a higher reactivity in steam than in CO2 at identical partial pressures. In the mixed CO2–H2O atmosphere, an intermediate activation energy of EA = 127 kJ/mol is obtained, whereas for the simulated gasifier atmosphere containing CO2, H2O, CO, and H2, a lower value of EA = 95 kJ/mol is observed. For all kinetic descriptions, the structural parameter of the RPM is Ψ = 2.0, and the RPM reproduces the experimental data with correlation coefficients R2 > 95%, while the overall mean model deviations are generally below 5%.Cross–comparison shows that the CO2 and H2O contributions in the mixed atmosphere are strictly additive: the measured kinetics coincide with predictions from the sum of the single–gas rate constants, indicating separate active sites. Although the activation energies for the mixed CO2–H2O atmosphere and the simulated gasifier atmosphere differ, the resulting conversion behavior under the investigated entrained–flow conditions is very similar, and no pronounced inhibition by CO and H2 is observed. Thus, for MWC under the present conditions, carbon conversion is effectively governed by the partial pressures of CO2 and H2O, which simplifies kinetic modelling for design and scale–up of pressurized biomass gasifiers.

Water-in-water (w-w) emulsion electrospinning combines organic solvent-free processing, tunable fiber microstructures, and the potential for the encapsulation and subsequent release of sensitive compounds in a single platform. The key challenge is maintaining stable phase-separated structures throughout the electrospinning process while ensuring spinnability. Tackling this challenge, this study explores the boundaries of spinnability of ww emulsions of poly(ethylene oxide) (PEO) and dextran that are stabilized by cellulose nanocrystals (CNCs) to form w-w Pickering emulsions (PEs). By constructing a phase diagram, the organization and stabilization of PEs at different PEO/dextran ratios (70/30, 50/50, and 30/70), polymer concentrations, and CNC contents (0.1-2 wt%) are systematically investigated. This is followed by the translation of the droplets to fibers, whereby a combined setup of fluorescence and electron microscopy reveals the role of the particle stabilizer in fiber formation, diameter distribution and architecture. Core-shell fibers are obtained from PEO-rich and dextran-rich systems at optimized emulsion compositions (e.g., >= 1 wt% CNC), where the CNCs align with the electric field assisting the fiber core formation, as confirmed by polarized optical microscopy. The optimized PEs are easily electrospun to fiber mats suitable for further characterization by IR spectroscopy, thermogravimetry and calorimetry. Ultimately, this article demonstrates how w-w PEs as fully water-based system can be integrated into a high-end polymer processing technique. The result are multiphasic fibers with high potential for advanced applications ranging from functional foods to biomedicine.