Searcharxiv⌕ Search

arXiv subjects

Antonia Larese

Publications and source records attributed to Antonia Larese.

2 recordsLinked to original sources

Graph Neural Network based non-local models: Toward application to RANS turbulence modeling

Most machine-learning closures for the Reynolds-averaged Navier-Stokes (RANS) equations are local, predicting the closure quantities at each point solely from local flow features, despite the non-local nature of turbulent interactions. We propose a non-local closure framework based on Graph Convolutional Networks (GCNs) that exploits the mesh connectivity. Each finite-volume cell becomes a graph node, and cells sharing a face are connected by an edge. Through message passing, each layer aggregates information from neighboring nodes, so that after $k$ layers the prediction at a cell depends on its $k$-hop neighborhood, which naturally accounts for non-local effects. To make training feasible when each simulation yields a single large graph, we also introduce a breadth-first-search $k$-hop subgraph sampling strategy, which acts as data augmentation and cuts the cost per processed graph during training. We compare GCNs with feed-forward networks (FNNs) of equal size on two periodic-hills benchmarks: a laminar advection-diffusion transport problem that mimics the non-local character of RANS closures, and the turbulent eddy-viscosity prediction of the Closure Challenge dataset at $Re = 5600$. In the laminar case, FNNs are accurate only within the training regime and with sufficiently informative inputs, whereas GCNs generalize better to unseen geometries, including out-of-distribution flows, and partly compensate for missing input features. In the turbulent proof-of-concept case, the graph-based models achieve modest but consistent improvements compared to FNNs, supporting GCNs as a promising route to non-local RANS closures.

physics.flu-dyn↗

Free surface flow through rigid porous media -- An overview and comparison of formulations

In many applications free surface flow through rigid porous media has to be modeled. Examples refer to coastal engineering applications as well as geotechnical or biomedical applications. Albeit the frequent applications, slight inconsistencies in the formulation of the governing equations can be found in the literature. The main goal of this paper is to identify these differences and provide a quantitative assessment of different approaches. Following a review of the different formulations, simulation results obtained from three alternative formulations are compared with experimental and numerical data. Results obtained by 2D and 3D test cases indicate that the predictive differences returned by the different formulations remain small for most applications, in particular for small porous Reynolds number ReP < 5000. Thus it seems justified to select a formulation that supports an efficient algorithm and coding structure.

physics.flu-dyn↗