119092 research outputs found
Sort by
Neural networks meet hyperelasticity: a monotonic approach
No full textWe propose and apply a novel parametrized physics-augmented neural network (PANN) constitutive model to experimental data of rubber-like materials whose behavior depends on manufacturing parameters. For this, we conduct experimental investigations on a 3D printed digital material at different mix ratios and consider several datasets from literature, including Ecoflex at different Shore hardness, a photocured 3D printing material at different grayscale values, and a EPDM rubber synthesized with different amounts of curatives. We introduce a parametrized hyperelastic PANN model which can represent material behavior at different manufacturing parameters. The proposed model fulfills common mechanical conditions of hyperelasticity. In addition, the hyperelastic potential of the proposed model is monotonic in isotropic isochoric strain invariants of the right Cauchy–Green tensor. In incompressible hyperelasticity, this is a relaxed version of the ellipticity (or rank-one convexity) condition. Using this relaxed ellipticity condition, the monotonic PANN model provides more flexibility than comparable approaches from literature that are elliptic by construction by formulating the PANN model to be both monotonic and convex. The monotonic PANN yields excellent results for a variety of different materials with largely varying qualitative and quantitative stress behavior. Although calibrated on uniaxial tensile data only, it leads to a stable numerical behavior of 3D finite element simulations. The findings of our work suggest that monotonicity could be a promising alternative to more constrained PANN models that include both convexity and monotonicity, in particular, when considering highly nonlinear and parametrized materials. This paper has three key novelties: (1) We propose a novel parametrized hyperelastic PANN model that is monotonic in both strain invariants and additional parameters. (2) We apply parametrized hyperelastic PANN models to experimental data of rubber-like materials whose behavior depends on manufacturing parameters. (3) With these highly nonlinear datasets, we benchmark the monotonic PANN model against existing PANN model formulations from literature. Furthermore, we compare the performance of different PANN models in terms of material stability and performance in finite element simulations
Robust optimal design of hybrid renewable energy systems for constant green hydrogen supply
No full textThermoelectric brick with solid–solid phase change materials for building-integrated energy harvesting: Design, simulation, and global evaluation
No full textThe buildings sector accounts for over 30% of global energy consumption and 20% of energy-related CO emissions, yet substantial thermal energy generated within buildings is wasted. This study aims to recycle this wasted thermal energy by designing and globally evaluating a novel thermoelectric brick, which converts wasted building thermal energy into electricity. The key innovation lies in integrating encapsulated thermoelectric generators with solid–solid phase change materials (SS-PCMs) as thermal buffers, eliminating leakage risks associated with conventional liquid–solid PCMs while maintaining structural integrity. A multiphysics model incorporating p- and n-type thermoelectric materials, conductive adhesive, copper, ceramics, and concrete was developed and validated against analytical and experimental solutions. The thermoelectric brick performance was evaluated across six representative global cities (Beijing, Doha, Helsinki, Las Vegas, Paris, and Sydney) spanning tropical, temperate, and cold climates. Results demonstrate strong seasonal and geographical sensitivity, with peak voltages reaching 210 mV and power outputs up to 2 W/m in cold climates during winter. Daily energy generation ranged from 0.29 to 9.28 Wh/m in facade system, sufficient to power low-consumption household devices. The system achieved thermoelectric efficiencies of 0.08–0.1% and exhibited stable voltage generation at night, demonstrating temporal complementarity with solar photovoltaic systems. These findings position thermoelectric bricks as a promising solution for sustainable, building-integrated energy harvesting with potential for hybrid renewable energy systems
Machine learning guided design and ablation behavior of ZrC-TaC-SiC ternary coatings
No full textUltra-high-temperature ceramic coatings are critical for thermal protection systems in hypersonic vehicles. However, their compositional optimization is constrained by the inefficiency of traditional trial-and-error method. In this study, a machine learning-driven design strategy is proposed for the ZrC-TaC-SiC ternary system (denoted ZnTmSt, n, m, t = 0-100) to enhance ablation resistance. A random forest regression model was constructed using 170 literature data to predict surface ablation temperature, mass ablation rate, and linear ablation rate across the full compositional space of the ZrC-TaC-SiC ternary system. Four optimized formulations (Z60T30S10, Z60T10S30, Z80T15S5, Z80T5S15), each exhibiting stable mass and linear ablation rates in the order of 10-4 g/s and 10-4 mm/s, respectively, were selected through the model and fabricated via supersonic atmospheric plasma spraying. Oxyacetylene ablation test results confirmed the reliability of the model, with prediction errors for linear ablation rate within the order of 10-5 mm/s. Microstructural characterization revealed that high-TaC content (15-30 wt%) led to excessive formation of Zr6Ta2O17 during ablation, resulting in reduced thermal stability of the oxide scale and discontinuities in structure. Moreover, non-uniform sintering caused an increase in porosity. Interestingly, the low-TaC (5 wt%) Z80T5S15 sample achieved a dense and continuous oxide layer by a proper Zr/Ta content ratio in oxide phase, exhibiting excellent ablation resistance. This study establishes a machine-learning-assisted strategy combined with experimental validation, which can provide a new paradigm for intelligent design and performance prediction of ultra-high-temperature ceramic coatings
Single‐source‐precursor synthesis of novel BCN and HfBCN nanocomposites for energy conversion application
No full textMulticomponent M BCN ceramics, where M represents a transition metal, are garnering increased interest due to their multifunctionality and ultrahigh‐temperature stability. Currently, research on M BCN primarily focuses on its preparation and application in the form of thin films. In this article, we prepared monolithic BCN and HfBCN nanocomposites using a single‐source precursor approach and explored their potential application as thermoelectric materials for energy conversion. Cubic HfB x C y N 1– x – y nanocrystallites ranging from 10 to 200 nm were formed in situ within amorphous HfBCN materials after sintering at 1600°C. Furthermore, the presence of turbostratic BC x N 1– x phase within the matrix, significantly surpassing the percolation threshold, contributes to both BCN and HfBCN nanocomposites, exhibiting exceptionally high electrical conductivity exceeding 100 S cm –1 just at room temperature. A power factor of approximately 40 μW m –1 K –2 was achieved in HfBCN‐based nanocomposites, setting a record high among polymer‐derived ceramics
Reversibly redox-active iron oxide structures in FeNC catalysts identified by microscopy and spectroelectrochemical EPR and Mössbauer methods
No full textIdentifying active sites in FeNC catalysts for oxygen reduction reactions (ORR) and active site changes during preparation, storage, and electrochemical cycling are key challenges in the quest for improved catalysts. In this work, high-resolution transmission electron microscopy (TEM) is combined with 57Fe Mössbauer and electron paramagnetic resonance (EPR) spectroscopies to investigate iron centers in high-performance FeNC catalysts with regard to their structure, coordination, and oxidation and spin states. Reversible and irreversible changes during storage, the preparation of FeNC electrodes, and their use in electrochemical cells are investigated by complementary spectroelectrochemical Mössbauer and EPR methods. Microscopy of the as-prepared FeNC materials reveals iron to be evenly distributed in isolated sites or a few atoms containing sites. Mössbauer and EPR identify weakly and strongly magnetically coupled high-spin Fe(III) in rhombically distorted octahedral coordination or superparamagnetic clusters, high-spin Fe(II) sixfold coordinated in iron oxides, and intermediate-spin Fe(II) in square planar coordination. Upon oxygen exposure, a notable oxidation state change from Fe(II) to Fe(III) is observed, the iron is less evenly distributed, and larger iron oxide nanoparticles are formed. It is noted that for this catalyst, before and after oxygen exposure, most of the iron is bound in iron oxide structures. Under the applied potential, Fe(III) is partially reduced to Fe(II) in clustered and isolated or weakly coupled sites. This change is mostly reversible, suggesting structural retention of the majority of the catalyst
High pressure derived SiHfBN ceramics: toward amorphous ceramics with exceptional hardness and thermal stability
No full textThe reinforcement of dense Si3N4-based ceramics with transition metal nitrides (e.g., HfN, TiN, ZrN) has attracted great attention owing to their potential to enhance mechanical properties and high temperature stability. In the present work, fully dense amorphous SiHfBN ceramics and polycrystalline HfN/α,β-Si3N4 ceramic composites were prepared by high-pressure and high-temperature (HPHT) technique using a polymer-derived amorphous SiHfBN precursor as raw material. The densification and crystallization behavior of the SiHfBN amorphous samples were studied under 5 GPa within a temperature range from 1000 °C to 1800 °C. The amorphous SiHfBN ceramics exhibit Vickers’ hardness and fracture toughness comparable to those of HfN/α,β-Si3N4 ceramic composites, reaching up to 17.37 GPa and 4.79 MPa·m1/2, respectively. Notably, the amorphous SiHfBN ceramics show improved oxidation resistance compared with that of the HfN/α,β-Si3N4 ceramic composites, with a mass loss of less than 2 wt % at 1500 °C in air. This work serves as a valuable reference for advancing the development of amorphous ceramics with outstanding mechanical properties and thermal stability
Reward Modeling for Scientific Writing Evaluation
No full textScientific writing is an expert-domain task that demands deep domain knowledge, task-specific requirements and reasoning capabilities that leverage the domain knowledge to satisfy the task specifications. While scientific text generation has been widely studied, its evaluation remains a challenging and open problem. It is critical to develop models that can be reliably deployed for evaluating diverse open-ended scientific writing tasks while adhering to their distinct requirements. However, existing LLM-based judges and reward models are primarily optimized for general-purpose benchmarks with fixed scoring rubrics and evaluation criteria. Consequently, they often fail to reason over sparse knowledge of scientific domains when interpreting task-dependent and multi-faceted criteria. Moreover, fine-tuning for each individual task is costly and impractical for low-resource settings. To bridge these gaps, we propose cost-efficient, open-source reward models tailored for scientific writing evaluation. We introduce a two-stage training framework that initially optimizes scientific evaluation preferences and then refines reasoning capabilities. Our multi-aspect evaluation design and joint training across diverse tasks enable fine-grained assessment and robustness to dynamic criteria and scoring rubrics. Experimental analysis shows that our training regime strongly improves LLM-based scientific writing evaluation. Our models generalize effectively across tasks and to previously unseen scientific writing evaluation settings, allowing a single trained evaluator to be reused without task-specific retraining
SciCoQA: Quality Assurance for Scientific Paper--Code Alignment
No full textWe present SciCoQA, a dataset for detecting discrepancies between scientific publications and their codebases to ensure faithful implementations. We construct SciCoQA from GitHub issues and reproducibility papers, and to scale our dataset, we propose a synthetic data generation method for constructing paper-code discrepancies. We analyze the paper-code discrepancies in detail and propose discrepancy types and categories to better understand the occurring mismatches. In total, our dataset consists of 611 paper-code discrepancies (81 real, 530 synthetic), spanning diverse computational science disciplines, including AI, Physics, Quantitative Biology, and others. Our evaluation of 21 LLMs highlights the difficulty of SciCoQA, particularly for instances involving omitted paper details, long-context inputs, and data outside the models' pre-training corpus. The best performing model in our evaluation, GPT-5, can only detect 45.7\% of real-world paper-code discrepancies