Efficient modeling of liquid splashing via graph neural networks with adaptive filter and aggregator fusion DOI

Jinyao Nan,

Pingfa Feng, Jie Xu

et al.

International Journal of Numerical Methods for Heat &amp Fluid Flow, Journal Year: 2024, Volume and Issue: 34(6), P. 2513 - 2538

Published: June 24, 2024

Purpose The purpose of this study is to advance the computational modeling liquid splashing dynamics, while balancing simulation accuracy and efficiency, a duality often compromised in high-fidelity fluid dynamics simulations. Design/methodology/approach This introduces efficient graph neural network simulator (FEGNS), an innovative framework that integrates adaptive filtering layer aggregator fusion strategy within architecture. FEGNS designed directly learn from extensive splash data sets, capturing intricate intrinsically complex interactions. Findings achieves remarkable 30.3% improvement over traditional methods, coupled with 51.6% enhancement speed. It exhibits robust generalization capabilities across diverse materials, enabling realistic simulations droplet effects. Comparative analyses empirical validations demonstrate FEGNS’s superior performance against existing benchmark models. Originality/value originality lies its layer, which independently adjusts weights per node, novel enriches network’s expressive power by combining multiple aggregation functions. To facilitate further research practical deployment, model has been made accessible on GitHub ( https://github.com/nanjinyao/FEGNS/tree/main ).

Language: Английский

Large-Scale Hydrological Models and Transboundary River Basins DOI Open Access
Charalampos Skoulikaris

Water, Journal Year: 2024, Volume and Issue: 16(6), P. 878 - 878

Published: March 19, 2024

Large-scale hydrological modeling is an emerging approach in river hydrology, especially regions with limited available data. This research focuses on evaluating the performance of two well-known large-scale models, namely E-HYPE and LISFLOOD, for five transboundary rivers Greece. For this purpose, discharge time series at rivers’ outlets from both models are compared observed datasets wherever possible. The comparison conducted using well-established statistical measures, namely, coefficient determination, Percent Bias, Nash–Sutcliffe Efficiency, Root-Mean-Square Error, Kling–Gupta Efficiency. Subsequently, models’ bias corrected through scaling factor, linear regression, delta change, quantile mapping methods, respectively. outputs then re-evaluated against observations same measures. results demonstrate that neither consistently outperformed other, as one model performed better some basins while other excelled remaining cases. bias-correction process identifies regression most suitable methods case study basins. Additionally, assesses influence upstream waters water budget. highlights significance presents a methodological their applicability any basin global scale, underscores usefulness cooperative management international waters.

Language: Английский

Citations

2
Predicting drought stress under climate change in the Southern Central Highlands of Vietnam DOI
Phong Nguyen Thanh,

Thinh Le Van,

Xuan Ai Tien Thi

et al.

Environmental Monitoring and Assessment, Journal Year: 2024, Volume and Issue: 196(7)

Published: June 20, 2024

Language: Английский

Citations

2
Environmental flow assessment in transboundary rivers: Challenges and opportunities using big data - A Greek case study DOI Creative Commons

Eleni Karamanoli,

Charalampos Skoulikaris

E3S Web of Conferences, Journal Year: 2024, Volume and Issue: 585, P. 03005 - 03005

Published: Jan. 1, 2024

Environmental or ecological flow refers to the minimum needed sustain river- based ecosystems and their services. Evaluating environmental flows is of paramount importance, particularly in light rapid changes induced by climate change, anthropogenic pressures, continued damming river courses. The research aims evaluate four transboundary rivers Greece, namely Axios, Strymonas, Nestos Evros Rivers, using Tennant Tessman hydrological methods, assess compatibility with Greek national legislation. rivers’ runoff determined large-scale models applied at European scale, which simulate thousands basins simultaneously, generating extensive big data datasets. results demonstrate that legislation underestimates compared those derived from both methods. Furthermore, values method for winter months generally exhibit lower magnitudes obtained method, whereas during summer months, there appears be a convergence methodologies. proposed methodology can any within Union serve as significant roadmap further advancements assessment flow.

Language: Английский

Citations

1
A novel framework for uncertainty quantification of rainfall–runoff models based on a Bayesian approach focused on transboundary river basins DOI
Truyen Nguyen, Duc Hai Nguyen, Hyun‐Han Kwon

et al.

Journal of Hydrology Regional Studies, Journal Year: 2024, Volume and Issue: 57, P. 102095 - 102095

Published: Dec. 5, 2024

Language: Английский

Citations

1
Efficient modeling of liquid splashing via graph neural networks with adaptive filter and aggregator fusion DOI

Jinyao Nan,

Pingfa Feng, Jie Xu

et al.

International Journal of Numerical Methods for Heat &amp Fluid Flow, Journal Year: 2024, Volume and Issue: 34(6), P. 2513 - 2538

Published: June 24, 2024

Purpose The purpose of this study is to advance the computational modeling liquid splashing dynamics, while balancing simulation accuracy and efficiency, a duality often compromised in high-fidelity fluid dynamics simulations. Design/methodology/approach This introduces efficient graph neural network simulator (FEGNS), an innovative framework that integrates adaptive filtering layer aggregator fusion strategy within architecture. FEGNS designed directly learn from extensive splash data sets, capturing intricate intrinsically complex interactions. Findings achieves remarkable 30.3% improvement over traditional methods, coupled with 51.6% enhancement speed. It exhibits robust generalization capabilities across diverse materials, enabling realistic simulations droplet effects. Comparative analyses empirical validations demonstrate FEGNS’s superior performance against existing benchmark models. Originality/value originality lies its layer, which independently adjusts weights per node, novel enriches network’s expressive power by combining multiple aggregation functions. To facilitate further research practical deployment, model has been made accessible on GitHub ( https://github.com/nanjinyao/FEGNS/tree/main ).

Language: Английский

Citations

0