Length scale of nonlocal gradient regularisation inferred from DEM virtual experiments

Géotechnique Letters, Год журнала: 2025, Номер 15(3), С. 1 - 6

Опубликована: Апрель 9, 2025

Nonlocal gradient continua are a reliable way to regularise numerical models of boundary value problems affected by strain localisation. However, the determination additional parameters such formulations, including those controlling length scale potential shear bands, is not straightforward, in that it interplay among stress–strain relations and band morphology. To address this challenge, study proposes parameter strategy for nonlocal strain-softening Mohr–Coulomb model based on discrete-element (DEM) virtual biaxial tests dense granular material. The involves initial calibration conventional inclination, followed subsequent parameter, l. It shown unique combination these exists ensures an effective replication both Finally, effects analysed varying sample size W l ratio. simulated mechanical response patterns align well with DEM data obtained changing ratio mean particle d 50 . These results indicate direct relation between

Язык: Английский

Machine Learning Assisted Cross-Scale Hopper Design for Flowing Biomass Granular Materials

ACS Sustainable Chemistry & Engineering, Год журнала: 2025, Номер 13(16), С. 5838 - 5851

Опубликована: Апрель 16, 2025

The promise of biomass-derived biofuels is often overshadowed by intricate material handling challenges such as hopper clogging and screw feeder jamming. These issues stem from the knowledge gap among particle-scale properties (e.g., particle size), bulk-scale attributes relative density), macro-scale equipment design inclination), flow performance probability clogging). This work combines physical experiments, validated numerical simulations, data augmentation to develop a machine learning-based for flowing granular woody biomass materials. behavior simulated against tests utilizing developed smoothed hydrodynamics (SPH) solver modified hypoplastic model. A comprehensive evaluation performance, including rate, stability, pattern, conducted on an extensive set encompassing various sizes, moisture contents, densities, operating conditions. feed-forward neural network trained optimized with this correlate cross-scale metrics. results reveal promising predictive accuracy seen unseen sets. Further how input affect predicted metrics carried out. indicate that opening width primarily dictates throughput, while density, wall friction, inclination angle, collectively impact stability. Additionally, patterns are predominantly governed angle. Moreover, potential found be exclusively characterized index dedicated combination high dense packing, smooth low angles, small widths substantially elevates risk unstable flows clogging. study serves potent tool milled materials in hoppers all stakeholders biorefineries manufacturing.

Язык: Английский

A high-performance discrete-element framework for simulating flow and jamming of moisture bearing biomass feedstocks

Powder Technology, Год журнала: 2024, Номер unknown, С. 120548 - 120548

Опубликована: Дек. 1, 2024

Язык: Английский