arXiv subjects

Pietro Prestininzi

Publications and source records attributed to Pietro Prestininzi.

4 recordsLinked to original sources

arXiv2026

Vortex dynamics and air entrainment in dam break wave impacting on vertical walls: A multiphase lattice Boltzmann study

Air entrainment often plays a crucial role in determining impact loads exerted by free-surface wave flows interacting with structures, yet its modelling is often oversimplified in numerical approaches. In this study a two-phase numerical model, based on the Lattice Boltzmann Method coupled to a conservative Allen--Cahn interface-capturing equation is employed to perform direct numerical simulations of dam-break waves propagating over a dry bed and impacting on vertical walls. Access to high-resolution simulations enables a detailed assessment of how accurately resolving both air--water and solid--water interfaces affects local and overall dynamics, as well as quantities of extreme engineering interest. Indeed, the magnitudes and locations of the pressure peaks are strongly affected by wave front deflection and local aeration induced by a small corner vortex. Additionally, comparisons between no-slip and free-slip implementations suggest that the large air cavity formation, commonly observed as trapped inside the reflected jet falling back onto the incoming flow, may be the result of modeling assumptions rather than intrinsic flow physics, again highlighting the key role of near-wall shear in jet breakup dynamics.

physics.flu-dyn↗

arXiv2024

Can physical information aid the generalization ability of Neural Networks for hydraulic modeling?

Application of Neural Networks to river hydraulics is fledgling, despite the field suffering from data scarcity, a challenge for machine learning techniques. Consequently, many purely data-driven Neural Networks proved to lack predictive capabilities. In this work, we propose to mitigate such problem by introducing physical information into the training phase. The idea is borrowed from Physics-Informed Neural Networks which have been recently proposed in other contexts. Physics-Informed Neural Networks embed physical information in the form of the residual of the Partial Differential Equations (PDEs) governing the phenomenon and, as such, are conceived as neural solvers, i.e. an alternative to traditional numerical solvers. Such approach is seldom suitable for environmental hydraulics, where epistemic uncertainties are large, and computing residuals of PDEs exhibits difficulties similar to those faced by classical numerical methods. Instead, we envisaged the employment of Neural Networks as neural operators, featuring physical constraints formulated without resorting to PDEs. The proposed novel methodology shares similarities with data augmentation and regularization. We show that incorporating such soft physical information can improve predictive capabilities.

cs.LG↗

arXiv2021

Deformation and breakup dynamics of droplets within a tapered channel

In this paper we numerically investigate the breakup dynamics of droplets in an emulsion flowing in a tapered microchannel with a narrow constriction. The mesoscale approach for multicomponent fluids with near contact interactions is shown to capture the deformation and breakup dynamics of droplets interacting within the constriction, in agreement with experimental evidences. In addition, it permits to investigate in detail the hydrodynamic phenomena occurring during the breakup stages. Finally, a suitable deformation parameter is introduced and analyzed to characterize the state of deformation of the system by inspecting pairs of interacting droplets flowing in the narrow channel.

physics.flu-dyn↗

arXiv2016

Mapping Reactive Flow Patterns in Monolithic Nanoporous Catalysts

The development of high-efficiency porous catalyst membranes critically depends on our understanding of where the majority of the chemical conversions occur within the porous structure. This requires mapping of chemical reactions and mass transport inside the complex nano-scale architecture of porous catalyst membranes which is a multiscale problem in both the temporal and spatial domain. To address this problem, we developed a multi-scale mass transport computational framework based on the Lattice Boltzmann Method (LBM) that allows us to account for catalytic reactions at the gas-solid interface by introducing a new boundary condition. In good agreement with experiments, the simulations reveal that most catalytic reactions occur near the gas-flow facing side of the catalyst membrane if chemical reactions are fast compared to mass transport within the porous catalyst membrane.

physics.flu-dyn↗