Open Access

Epoxy resins exhibit good mechanical and thermal properties. Most commercially available epoxy resins rely on fossil resources, which is why research is being conducted on the synthesis of bio‐based alternatives. Although many scientists dealt with new bio‐based epoxy resins, there are still only a few publications on new bio‐based amine curing agents. Vanillin is a promising substance for the synthesis of amine curing agents due to its aromatic structure in combination with its potential bio‐origin from lignin. In this article, the synthesis of 3‐amino‐4,5‐dimethoxy‐benzylamine (ADBA) from vanillin and its use as a novel curing agent for epoxy resins is presented. The synthesis comprises four stages: nitration of the aromatic compound vanillin, methylation of the phenolic group, conversion of the aldehyde into an oxime, and hydrogenation to form the diamine. The diamine can be synthesized with a total yield of 51 wt% and a bio‐content of 74 wt%. According to DSC, the new hardener with diglycidyl ether of bisphenol A (DGEBA) shows a gradual curing reaction due to the presence of an aliphatic and an aromatic amine. Glass transition temperatures (Tg) of 135°C were determined via DMA when using ADBA, which is comparable to that of DGEBA cured with IPDA.

Thermoelectric energy harvesters (TEH) serve as autonomous energy sources in Internet of Things applications, particularly for powering wireless sensor nodes. However, they often require complex, computationally expensive multi-physical finite element method (FEM) simulations for precise optimization. Machine learning based surrogate models offer a computationally efficient alternative. This study presents a comprehensive comparative analysis of three prominent surrogate modeling techniques: Polynomial Chaos Expansion (PCE), Gaussian Process Regression (GPR), and Deep Neural Networks (DNN). These models are trained using highfidelity datasets derived from multi-physical FEM simulations, with a specific focus on limited data regimes to emulate realistic computational constraints. The predictive performance and generalization capabilities of PCE, GPR, and DNN are rigorously evaluated against unseen reference data across varying training set sizes. This approach allows for a quantified assessment of how data sparsity impacts model fidelity. Furthermore, the models’ capacity to forecast the influence of individual parameters is directly validated through comparison with empirical data obtained from a simplified experimental setup. The findings reveal distinct operational regimes for each methodology. While GPR demonstrates superior robustness and predictive accuracy in single-parameter optimization, PCE intrinsically offers excellent interpretability and efficiency for smooth parametric dependencies. Conversely, DNNs exhibit strong scalability as dataset sizes increase and deals well with the high-dimensional parameter space. Notably, while a single high-fidelity FEM simulation requires approximately 30 min per parameter set, the trained surrogate models deliver predictions virtually in real-time.

This study investigates the influence of domain definition and interface placement on the steady-state CFD prediction of performance curves of centrifugal fans. The focus is on the separation between rotating and stationary flow domains as well as on the effects of the selected interface treatment (Frozen Rotor vs. Mixing Plane). The results show that a physically consistent definition of the rotating domain—comprising the impeller and the surrounding vaneless space while the inlet nozzle and casing remain stationary — is essential for obtaining realistic total pressure predictions. Extending the rotating region into geometrically stationary components such as the inlet nozzle or volute casing introduces artificial shear work and erroneous inertial effects, resulting in increased losses and an underestimation of the pressure rise. A realistic rotor–outlet interface that spans the full axial height of the impeller exit yields markedly better efficiency predictions than an idealised, radially truncated interface that numerically acts as an extension of the impeller end-walls. For the present low-specific-speed fan, a moderate transition region within the vaneless space, corresponding to approximately 3–16 % above the impeller exit radius R₂, allows sufficient flow relaxation and leads to stable and physically consistent results. For steady-state simulations, Frozen Rotor solutions benefit from an interface distance of about 13 % above R₂, whereas Mixing Plane models provide the most robust and accurate performance predictions for interface positions around 6–7 % above R₂ in combination with a realistic interface geometry. Correct specification of wall motion, use of the SST k–ω turbulence model and appropriate near-wall resolution (y⁺ ≈ 1–3) are additional key factors for numerical consistency. Overall, the study demonstrates that careful domain definition, suitable interface placement, and a consistent numerical setup are essential prerequisites for physical fidelity and comparability of CFD-based performance curve simulations of centrifugal fans.

To estimate the lifetime of polymer-electrolyte membrane fuel cells (PEMFCs) in heavy-duty vehicles (HDVs), a deep understanding of degradation mechanisms under realistic operating conditions is essential. We developed two different accelerated stress tests (AST) for HDVs: (1) The “All Component Aging Test” (ACAT) is based on a synthetic voltage, temperature and humidity profile to equally stress all components of the membrane electrode assembly (MEA). (2) The “Truck Driving Cycle Test” (TDCT) is based on real HDV driving data and applies moderate voltage, temperature, and humidity conditions as well as elevated differential pressure. Additionally, standardized U.S. Department of Energy (DoE) component-specific aging tests were performed for comparison. A commercial HDV MEA was subjected to all aging protocols. As a result, the MEA’s end-of-life (EoL) was reached in the ACAT after 2100 h due to catalyst layer degradation. In contrast, the lifetime determining factor in the TDCT (EoL after 1500 h) was membrane degradation. Interestingly, the DoE tests predicted the aging of the catalyst layer as lifetime-limiting, but did not identify the membrane as critical component. These findings demonstrate that standardized test protocols alone are no adequate substitution of application-specific ASTs to estimate the real-world durability of PEMFC MEA components.

Manufacturing companies have a strong impact on climate change due to their immense degree of greenhouse gas emissions. Hence, it is mandatory that these companies take measures to reduce their greenhouse gas emissions. A promising measure is the reduction of the carbon footprint caused by single processes within manufacturing process networks. Various approaches to reduce the carbon footprint of single manufacturing processes have already been described in literature. However, there is a scientific need for action when it comes to the systematic reduction of the carbon footprint of queueing systems which are typically an integral element of manufacturing process networks. Therefore, this paper presents a novel approach that aims to assess and reduce the carbon footprint of manufacturing queueing systems systematically. First, a general queueing system is defined and its variables that contribute to the carbon footprint in manufacturing are determined. Subsequently, a mathematical model is designed which enables an abstract description of the interactions of different input variables on the carbon footprint of queueing systems in manufacturing. Afterwards, recommendations to reduce the carbon footprint of queueing systems are developed based on the previous findings. Finally, the approach is validated for a specific manufacturing queueing system with a discrete-event simulation study. The results give foundations for further research on dependencies within manufacturing process networks.