Experimental study on the origin of lobe-cleft structures in a sand storm

Abstract

Since the famous work by Kolmogorov on incompressible turbulence, the structure-function theory has been a key foundation of modern turbulence study. Due to the simplicity of Burgers turbulence, structure functions are calculated to arbitrary orders, which provides numerous implications for other compressible turbulent systems. We present the derivation of exact forcing-scale resolving expressions for high-order structure functions of the burgers turbulence. Compared with the previous theories where the structure functions are calculated in the inertial range based on the statistics of shocks, our expressions link high-order structure functions in different orders without extra information on the flow structure and are valid beyond the inertial range, therefore they are easily checked by numerical simulations.

An Adaptive Physics-Informed Neural Network with Two-Stage Learning Strategy to Solve Partial Differential Equations

“…Once the network is trained, the solution is obtained in a faster way compared to previous numerical methods. At present, PINN has achieved remarkable results in the field of computational science and engineering, including computational and solid mechanics [1,10,12,35], fluid mechanics [4,13,17,41,53], high frequency partial differential equations [6], ordinary differential equations [32], fault detection [38], state-space modeling [2], biomedical science [5,8,20,51], thermodynamics [55], and design of metamaterials [7,23] among others.…”

“…Wang et al [47] proposed a learning rate annealing algorithm, which uses gradient statistics to balance the interaction between different terms in the composite loss function during model training. Xiang et al [53] proposed an adaptive loss function method, which automatically assigns loss weights by updating noise parameters based on maximum likelihood. Zobeiry et al [55] proposed an adaptive normalization scheme to reduce the loss term at the same time.…”

Hydrodynamics and Sediment Transport of Downslope Turbidity Current Through Rigid Vegetation

“…Extensive studies have investigated the sedimentary and hydrodynamic characteristics of turbidity currents that flow through bare beds using low‐cost and efficient physical laboratory experiments (Ali et al., 2022; Gray et al., 2005; Oehy et al., 2010; Pohl et al., 2020). Some comprehensive conclusions are widely accepted: (a) the size of sediment particles is the primary factor controlling the hydrodynamics process of turbidity currents and also their deposition‐erosion patterns (Ali et al., 2022; Nomura et al., 2021; Oehy et al., 2010); (b) the balance of material exchange between the currents and the bed layer allows dividing the propagation regimes of turbidity currents into self‐deceleration (deposition surpasses erosion), self‐suspension (deposition and erosion are equivalent), and self‐acceleration (erosion outweighs deposition) (Dorrell et al., 2019; Hu et al., 2015; Wells & Dorrell, 2021); (c) turbidity current with fine particles more easily accomplishes self‐acceleration and enters the long‐distance transport state (Soler et al., 2020; Zhang et al., 2021); (d) the moderate extension of fine sediment in turbidity currents dramatically promotes their particle‐carrying capacity (Eggenhuisen et al., 2019; Gray et al., 2005); and (e) the transport distance of coarse particles in turbidity currents is proportional to the percentage of fine particles in the composition (Soler et al., 2021). Numerical simulation is another common method to solve turbidity‐current issues, mainly including Large Eddy Simulation (Kneller et al., 2016; Salinas et al., 2019) and Direct Numerical Simulation (Breard et al., 2019; Ouillon et al., 2019), which are generally coupled with the Discrete Element Method (Xie et al., 2022; Zhu et al., 2022), to reproduce the transport feature of particles within turbidity currents.…”