Percorrer por data de Publicação, começado por "2026-03"
A mostrar 1 - 4 de 4
Resultados por página
Opções de ordenação
- Higher education settings menus have low compliance with the Mediterranean Diet and high carbon and water footprint: a case study from Portugal, Croatia and TurkeyPublication . Neto, B.; Dikmen, D.; Ferreira, L.; Viegas, Cláudia; Filipec, S.; Drobac, L.; Šatalić, Z.; Rocha, AdaThis study focuses on evaluating the compliance of menus from Higher Education Institutions (HEI) with the Mediterranean Diet (MD) and calculates their respective carbon and water footprints. From September 2023 to June 2024, menus from 52 HEIs across Portugal, Croatia, and Turkey were analysed using a Mediterranean Diet Compliance Index (MeDCIn). Also, the footprints of 300 meals from 30 different menus were calculated. Overall results show a low compliance with the MD (mean score 2.7 ± 3.4). Turkish menus scored the highest values (5.2 ± 1.7) while Portuguese menus scored the lowest (1.10 ± 3.7) (MeDCIn varies between −20.5 and 27). The limited availability of dishes with eggs, wholegrains, olive oil, nuts, seeds, and seasonal products was a key factor contributing to the low compliance observed, as well as insufficient variety in Mediterranean dishes, vegetables, pulses, seafood, and lean meat. The average water footprint was 1785.41 ± 909.3 m3/ton, with Turkish menus having the highest consumption (2271.90 ± 1016.11 m3/ton) and Portuguese menus the lowest (1485.46 ± 767.28 m3/ton). The average carbon footprint was 1.9 kg CO2-eq, with Turkish menus again scoring the highest (2.91 ± 2.13 kg CO2-eq) and Portuguese menus the lowest (1.42 ± 1.26 kg CO2-eq). The findings reveal a complex relationship between MD compliance and environmental footprints, with moderate positive correlations observed. These results provide valuable insights to develop targeted interventions to improve menu options in HEI cafeterias and reduce their environmental impact.
- Optimisation of hydraulic lime mortars incorporating an oil-refinery catalyst by-product for sustainable building rehabilitationPublication . Costa, Carla; Nunes, SandraAbstract This investigation employs a Central Composite Design-based Design of Experiments (DoE) methodology to develop hydraulic lime mortars incorporating equilibrium catalyst (ECat), a by-product generated at the fluid catalytic cracking unit in oil refineries. The derived mathematical models describe the quantitative effects of key mixing variables, specifically ECat content, water-to-binder ratio and water repellent dosage, as well as their cross-interactions, on mortar properties, namely workability, compressive strength, ultrasound propagation velocity and dynamic modulus of elasticity. Numerical optimisation techniques enabled the identification of optimal lime mortar compositions that maximise eco-efficiency while ensuring compliance with both regulatory and technological requirements for diverse masonry applications, including the rehabilitation of ancient buildings. Results confirm the by-product upcyclability of ECat, with feasible incorporation levels up to 56.6 % by mass, yielding mortars with significant potential for reducing the environmental impact of the built environment while advancing the circular economy and fostering technological innovation in the construction sector.
- Fourier-enhanced sequence-to-sequence latent graph neural networks for multi-node spatiotemporal forecasting in a hydroelectric reservoirPublication . Seman, Laio Oriel; Stefenon, Stefano Frizzo; Yow, Kin-Choong; Coelho, Leandro dos Santos; Mariani, Viviana CoccoThis paper presents a Fourier-enhanced dynamic sequence-to-sequence latent graph neural network (Seq2SeqLatentGNN), a deep learning architecture for multi-node spatiotemporal forecasting in hydroelectric reservoir systems. The model integrates three key components: (i) a custom Fourier layer that analyzes global temporal patterns through frequency-domain transformations, (ii) a latent correlation graph convolutional network that infers relational structures between monitoring stations without requiring predefined adjacency matrices, and (iii) an attention-based sequence-to-sequence model that processes temporal dependencies while enabling multi-step forecasting. The architecture simultaneously learns graph structure and forecasting tasks, adapting to changing spatial relationships between reservoir nodes. The proposed architecture was evaluated using a comprehensive dataset derived from 19 interconnected hydroelectric reservoirs located in southern Brazil. The dataset encompasses multiple years of high-resolution (hourly) measurements, including reservoir water levels, inflow and outflow rates, precipitation records, and energy production metrics. Experimental results demonstrate that Seq2SeqLatentGNN achieves superior performance compared to conventional statistical models and contemporary machine learning methods, as measured by standard error metrics. Analysis of the learned latent correlations reveals meaningful spatial dependencies that align with hydrological principles. The model exhibits consistent performance across varying temporal patterns, adapts to regime transitions, and captures both periodic and nonstationary dynamics. The proposed architecture contributes to spatiotemporal forecasting by combining spectral processing, dynamic graph learning, and sequence modeling in a unified framework applicable to systems with evolving connectivity patterns.
- Fourier-enhanced sequence-to-sequence latent graph neural networks for multi-node spatiotemporal forecasting in a hydroelectric reservoirPublication . Seman, Laio Oriel; Stefenon, Stefano Frizzo; Yow, Kin-Choong; Coelho, Leandro dos Santos; Mariani, Viviana CoccoAbstract This paper presents a Fourier-enhanced dynamic sequence-to-sequence latent graph neural network (Seq2SeqLatentGNN), a deep learning architecture for multi-node spatiotemporal forecasting in hydroelectric reservoir systems. The model integrates three key components: (i) a custom Fourier layer that analyzes global temporal patterns through frequency-domain transformations, (ii) a latent correlation graph convolutional network that infers relational structures between monitoring stations without requiring predefined adjacency matrices, and (iii) an attention-based sequence-to-sequence model that processes temporal dependencies while enabling multi-step forecasting. The architecture simultaneously learns graph structure and forecasting tasks, adapting to changing spatial relationships between reservoir nodes. The proposed architecture was evaluated using a comprehensive dataset derived from 19 interconnected hydroelectric reservoirs located in southern Brazil. The dataset encompasses multiple years of high-resolution (hourly) measurements, including reservoir water levels, inflow and outflow rates, precipitation records, and energy production metrics. Experimental results demonstrate that Seq2SeqLatentGNN achieves superior performance compared to conventional statistical models and contemporary machine learning methods, as measured by standard error metrics. Analysis of the learned latent correlations reveals meaningful spatial dependencies that align with hydrological principles. The model exhibits consistent performance across varying temporal patterns, adapts to regime transitions, and captures both periodic and nonstationary dynamics. The proposed architecture contributes to spatiotemporal forecasting by combining spectral processing, dynamic graph learning, and sequence modeling in a unified framework applicable to systems with evolving connectivity patterns.