Chen Xu, Giao Vu, Ba Trung Cao, Zhen Liu, Fabian Diewald, Yong Yuan, and Günther Meschke are the authors of the newly published article entitled "Bridging simulation and experiment: A self-supervised domain adaptation framework for concrete damage classification". It has been published by Elsevier in the journal Advanced Engineering Informatics.

Abstract:
Reliable assessment of concrete degradation is critical for ensuring the safety and longevity of engineering structures. This study proposes a self-supervised domain adaptation framework for robust concrete damage classification using coda wave signals. To support this framework, an advanced virtual testing platform is developed, which combines multiscale modeling of concrete degradation with ultrasonic wave propagation simulations. This setup enables the generation of large-scale labeled synthetic data under controlled conditions, reducing the dependency on costly and time-consuming experimental labeling. However, neural networks trained solely on simulated data often suffer from degraded performance when applied to experimental data due to domain shifts. To bridge this domain gap, the proposed framework integrates domain adversarial training, minimum class confusion loss, and the Bootstrap Your Own Latent (BYOL) strategy. These components work jointly to facilitate effective knowledge transfer from the labeled simulation domain to the unlabeled experimental domain, leading to accurate and reliable damage classification in concrete. Extensive experiments demonstrate that the proposed method achieves notable performance gains, reaching an accuracy of 0.7762 and a macro F1 score of 0.7713, outperforming both the plain 1D CNN baseline (accuracy: 0.5867; macro F1: 0.5832) as well as six representative domain adaptation techniques. Moreover, the method exhibits high robustness across independent runs and adds only minimal training overhead (about two additional minutes). These findings underscore the practical potential of the proposed simulation-driven and label-efficient framework for real-world structural health monitoring applications.

Get a free access with this authors' share link (until January 12, 2026) or try this link:

The article "Investigating the sliding behavior of graphene nanoribbons", written by Gourav Yadav, Aningi Mokhalingam, Roger A. Sauer, and Shakti S. Gupta, has been published as open access in "Carbon" by Elsevier.


Abstract:
This work presents a Euler–Bernoulli beam finite element (FE) model to study the interlayer interaction mechanics of graphene nanoribbon (GNR) over a graphene substrate. The FE model is calibrated using molecular dynamics (MD) simulations employing the potential of Kolmogorov and Crespi (2005). This study focuses mainly on the effect of boundary conditions on the sliding behavior, and on the strain transfer between layers when the substrate is subjected to uniform biaxial deformations. The interlayer sliding behavior is found to depend on the presence of critical parameters, namely, the length of the GNR and the applied strain to the substrate. The FE results indicate that the applied strain transferred from the substrate to the GNR varies linearly up to a critical value ∈c beyond which it decreases suddenly. Further, ∈c is found to appear beyond a critical GNR length, Le ≈ 14 nm. Furthermore, a length parameter Ld ≈ 10 nm is computed, beyond which the sliding of GNR is dissipative. Through FE simulations, it is also found that for a GNR length ≥ 17 nm, the edge pulling force saturates. Our results also highlight the importance of the inertia of GNR on its sliding for different boundary conditions. It is also concluded that the maximum strain that can be transferred to GNR lies between 0.59% and 1.15%. The results of the FE approach align with MD simulations within an error of approximately 10% that can be attributed to the choice of material parameters and the simulation setup.

Rodolfo Javier Williams Moises presented his doctoral theses with the title "Computational Modeling and Design of Artificial Soil Freezing in Tunneling" on 6th November 2025 at 2PM.


Well done, Rodolfo! We wish you even more success in the future!

Abstract:
This thesis presents a computational framework for modeling and designing frozen soil structures formed using the artificial ground freezing (AGF) method in tunneling. AGF is a ground improvement technique that uses freeze pipes installed in the soil to form a frozen body over days to months. In tunneling, it provides temporary ground support and watertightness. The framework integrates computational geomechanics modeling with AGF design principles, enabling the optimal design of frozen soil structures in tunneling projects. It includes tools for geomechanical analysis of AGF, simulation of conventional and mechanized tunneling in frozen ground, and optimization of freeze pipe layouts. The backbone of the framework is a thermo-hydro-mechanical (THM) finite element model for soil freezing and thawing, enhanced with constitutive models for pore pressure coefficients, strength, and creep of frozen soils, and is computationally robust for high seepage flow simulations. In a case study under high seepage flow and a fixed number of pipes, the framework uses machine learning to design an optimized pipe layout that outperforms a conventional layout with uniform spacing in reducing freezing time. Overall, the thesis advances computational modeling and data-driven optimization methods for AGF in tunneling, demonstrating potential for integration into the design phases of ground-freezing engineering.

The open access article "Efficient snap-to-contact computations for van der Waals interacting fibers", written by A. Borković, M.H. Gfrerer, R.A. Sauer, B. Marussig, has been published in the Elsevier journal European Journal of Mechanics A/Solids.

Abstract:
We consider van der Waals interactions between in-plane fibers, where the computational model employs the Lennard-Jones potential and the coarse-grained approach. The involved 6D integral over two interacting fibers is split into a 4D analytical pre-integration over cross sections and the remaining 2D numerical integration along the fibers’ axes. Two section-section interaction laws are implemented, refined, and compared. Fibers are modeled using the Bernoulli–Euler beam theory and spatially discretized with isogeometric finite elements. We derive and solve the weak form of both quasi-static and dynamic boundary value problems. Four numerical examples involving highly nonlinear and dynamic snap-to-contact phenomena are scrutinized. We observe that the coarse-graining and pre-integration of interaction potentials enable the efficient modeling of complex phenomena at small length scales.

"Interactive real-time large-scale tunnel alignment design" is the new publication of our institute. It is written by Yaman Zendaki, Ba Trung Cao, Steffen Freitag, and Günther Meschke and published in the journal "Tunnelling and Underground Space Technology" by Elsevier.

Abstract
Alignment design in urban tunneling aims at selecting an optimal tunnel track satisfying design specifications and safety constraints, as well as considering the environmental and socio-economic impact. In terms of structural analysis, traditional alignment design relies on empirical methods, analytical models, and closed-form solutions to assess impacts on the surrounding built environment. Numerical simulations are only partially employed due to the large-scale nature of such projects, which typically require significant computational resources and effort, making them costly and time-consuming for conventional design workflows. In this work, a simulation-based approach is proposed for large-scale tunnel alignment design, aiming at efficiently solving the computational challenges encountered during the design process. An interactive design tool is also developed, which enables the identification of optimal alignments considering tunneling-induced soil–structure interactions as design criteria. The core of the framework is an advanced numerical simulation model based on the Finite Cell Method, which predicts surface settlements and the associated risk of damage to existing buildings for various tunnel alignments. To achieve interactive real-time performance, fast surrogate models are adopted using Proper Orthogonal Decomposition and Radial Basis Functions. The surrogate models demonstrate strong agreement with the numerical results, achieving an average L2 norm error of 4%, while considerably reducing the computation time from 10 to 12 h to just 1 to 2 s. To determine optimal alignments, the Non-dominated Sorting Differential Evolution is used to solve a multi-objective minimization problem. The fitness function balances minimizing construction costs depending on the tunnel depth and length with maximizing the curvature radius, while enforcing structural constraints to prevent building damage. Finally, the entire real-time design optimization framework is integrated into a web-based software application. In the final optimization step, the optimal alignment, predicted surface settlements, and estimated damage risks to existing structures are visualized through an intuitive graphical user interface. The developed application enhances practical tunnel alignment design by providing advanced simulation results instantly to support more informed decision-making.

The article has a 50 days' free access with this personalized share link (until December 16, 2025) or is generally accessible here: