We present a homogenisation framework, applicable to both stretching- and bending-dominated architectures, to derive equivalent continuum models of three-dimensional lattice metamaterials in a fully automatic way and without making any a-priori assumption on the nature of the homogenised medium. Enforcing a standard energetic equivalence between a discrete lattice model and a continuum medium, the constitutive tensors of the homogenised continuum are obtained in a systematic way by invoking only the topological data of the unit cell, i.e. strut directions and distances, cross-sectional and material properties, thus avoiding ad-hoc formulations. In particular we prove that stretching-dominated lattice architectures can be effectively described by a standard Cauchy model and that the resulting elasticity tensor is positive definite provided that the unit cell is internally stable, i.e. if no internal mechanisms exist when all nodes are modelled as spherical hinges. In this way a rigorous link is established between the kinematic properties of the discrete structure and the well-posedness of the equivalent continuum model. The proposed computational framework is validated by recovering, as a special case, the known expression of the constitutive tensor for a lattice based on octet–truss and SC/BCC unit cells. For bending-dominated architectures, the analytical expression of the derived constitutive tensor, which appears to be new, is numerically validated by FE analyses. In particular, we show that the strain-gradient contribution becomes significant with respect to the first-gradient terms only when internal mechanisms, associated with lack of rigidity of the nodal connection, are present in the lattice.
General-purpose Named Entity Recognition (NER) models struggle with domain-specific challenges, particularly in the legal sector, where complex syntax, specialized terminology, and privacy concerns pose significant obstacles.This paper presents a novel data-driven two-stage framework that leverages Large Language Models (LLMs) fine-tuned for legal applications to enhance NER for criminal process documents. Our contributions include: (i) the design of a novel framework for structured entity extraction in legal texts, (ii) the definition of an enhanced ontology adapted to criminal law, and (iii) a comprehensive evaluation of the framework on a real-world dataset. As a side contribution of the paper, we extend the renowned OntoNotes5 ontology by integrating new domain-specific entities tailored to criminal law.To evaluate our approach, we construct and manually annotate a real-world dataset comprising over 100 criminal judgments from four Italian courts, sourced from the Italian Antimafia and Anti-terrorism National Directorate (Direzione Nazionale Antimafia e Antiterrorismo — DNAA). Experimental results demonstrate the effectiveness of fine-tuned LLMs in accurately identifying legal entities. Among the evaluated models, LLaMA-3.2-1B demonstrates the lowest training time, while LLaMA-2.7B achieves the fastest inference time. In terms of predictive performance, LLaMA-2.7B and Vicuna-7B consistently yield the highest accuracy scores across evaluation metrics.
Porous materials such as glass wool are widely used in aircraft fuselage insulation systems for their sound absorption performance. In operational aeronautical environments, their acoustic performance may differ from that measured under nominal laboratory conditions due to several factors such as protective coverings, installation procedures and moisture variations occurring during flight operations and throughout the aircraft service life. Despite their practical relevance, the effects of these non-nominal conditions on sound absorption variability remain insufficiently characterized. This study investigates the frequency-dependent sound absorption coefficient of aeronautical glass wools under controlled non-nominal conditions by combining impedance tube measurements, machine-learning techniques and Shapley additive explanations (SHAP). The investigated factors include material type, protective layers, relative humidity, humidity cycling and both controlled and operator-dependent mounting configurations. Results indicate that material, layer, relative humidity and humidity cycling significantly influence the acoustic response, with moisture-related effects exhibiting a strongly frequency-dependent behavior concentrated within three distinct frequency bands. Installation-related effects are also found to produce identifiable spectral variations, with operator-dependent mounting conditions generally associated with higher absorption levels and shifts of the peak response toward lower frequencies. The results further indicate that more controlled and uniformly distributed contact conditions improve measurement repeatability and reduce installation-induced variability.
Municipal wastewater treatment plants face increasing pressure to improve energy recovery while reducing sludge volumes and operational costs. In this context, low-input process intensification strategies are of growing interest. This study evaluated a continuous side-stream magnetization strategy applied during 44 days of mesophilic anaerobic digestion (37 +/- 2 degrees C) of sewage sludge, where digesting sludge was continuously recirculated through a tubular rare-earth-material magnetic polarizer generating a low-intensity static magnetic field (20 mT). The experimental results revealed a substantial increase in AD performance, with the SMF-exposed reactor achieving a 48.3% increase in specific biomethane production and a 15% higher reduction in volatile solids compared to the control reactor. Additionally, the digestate exposed to the SMF showed an improved dewaterability, evidenced by a 26.8% reduction in capillary suction time compared to the non-magnetized digestate. Microbial community analysis at the end of the AD process indicated that continuous exposure to SMF significantly enhanced the abundance of key methanogens, including Methanosarcina and Methanobacterium, which contributed to the higher biomethane production observed during AD. These findings indicate that continuous SMF exposure during AD can significantly enhance biomethane production and can be regarded as a promising approach to improve energy recovery from sewage sludge.
The production of alternative protein sources such as single-cell protein (SCP) is expected to help mitigate the adverse environmental impacts of current feed and food production methods. This mitigation is especially relevant when SCP production is integrated with a carbon capture and nitrogen recovery system from nitrogen-rich wastes. The present study investigated a novel integrated approach to upcycle both the biogas containing hydrogen and carbon dioxide produced from the dark fermentation (DF) of food waste, and the nitrogen recovered through stripping from the liquid fermentate into protein-rich biomass. SCP was produced using a co-culture of the microalga Chlorella vulgaris and the hydrogen-oxidising bacterium Cupriavidus necator. The fermentate, characterised by an ammonium concentration of around 170.0 ± 36.2 mg N-NH4+/L, was transferred from a continuous 1 L fermenter to a 2 L stripping column, where an NH4+-N removal efficiency of 24 % was achieved. The NH3-laden air flow was supplied to the algal-bacterial culture growing in a 4 L column photobioreactor (PBR). The highest biomass concentration achieved in the PBR was 0.78 ± 0.02 g VSS/L under test conditions, while it reached up to 1.51 ± 0.04 g VSS/L under supply of chemical nitrogen salts and synthetic biogas. The algal-bacterial culture contained 31.6 ± 0.2 % protein/VSS under test conditions and eight of the nine essential amino acids were detected. Overall, the findings of this study highlight the potential of the proposed DF-microbial conversion system for carbon capture and nitrogen upcycling, while identifying ammonia stripping as the main bottleneck for future optimisation.
