
The accuracy of Smoothed Particle Hydrodynamics (SPH) simulations critically depends on the consistency and convergence of discrete schemes for first- and second-order spatial derivatives. Despite decades of development, a unified framework that classifies these derivative schemes by their structural form and consistency order has been lacking. This paper presents a systematic categorization of SPH-like gradient and Laplacian schemes into three first-derivative families– namely, Direct Derivative, Diffuse Derivative, and Non-Singular Diffuse Derivative – and four second-derivative families, including the widely used Mixed form. Within each family, corrections that enforce zeroth-, first-, and second-order consistency are derived, and novel second-order consistent (SOC) schemes are introduced for all families. A truncation-error analysis is also developed, which distinguishes the order of consistency from the actual order of accuracy and shows how the convergence rate depends on both the particle-disorder parameter ϵ and the smoothing-length ratio h/Δx. This analysis reveals an optimal h/Δx value exists for the FOC and SOC schemes. The numerical convergence study employs two test functions, which allow the influence of boundary-induced effects to be isolated. The convergence properties of these schemes are examined in a one-dimensional setting with a quintic Wendland kernel under both regular and irregular particle distributions. Results confirm that consistency order is the primary determinant of convergence: zeroth-order schemes fail to converge, first-order gradient schemes achieve first- to second-order convergence depending on the particle symmetry, and SOC gradient schemes deliver full second-order accuracy. For Laplacians, both single-stage and double-stage approaches are analyzed. Single-stage SOC Laplacians achieve second-order convergence on symmetric arrangements but may reduce to first-order accuracy on irregular particle distributions at small spacings, owing to the increasing dominance of disorder-dependent error terms. The SOC double-stage scheme with a FOC inner gradient and a corrected outer operator provides a path to high-order accuracy in higher dimensions, involving a smaller linear system than the single-stage SOC but with a higher matrix-assembly cost. Although limited to one dimensional numerical results, this classification, truncation-error analysis, and convergence study offer a unified reference that clarifies the relationships among existing methods, introduces new correctors, and facilitates the informed choice and further development of robust derivative approximations for particle-based fluid dynamics simulations.
Failure of extra-thick hard strata (ETHS) is an important source of overburden-failure-related mining-induced seismicity. Using the Dongtan Coal Mine as an engineering case, a heterogeneous grain-based UDEC model containing randomly distributed triangular and polygonal blocks was established. Newly generated block and joint failures were identified as equivalent seismic events. The effects of ETHS thickness, strength, distance from the coal seam, and internal layering on stress evolution, seismic energy, and equivalent seismic-event distribution were investigated. Within the investigated parameter ranges, seismic energy decreased by 59.4%, 51.0%, and 59.0% with increasing ETHS thickness, strength, and distance from the coal seam, respectively. However, the 50 m thickness case exhibited early failure, resulting in a nonmonotonic energy response over the full thickness range. Internal layering primarily influenced the load-bearing configuration, failure morphology, and event distribution, with a nonmonotonic energy response. The normalized relative sensitivities of thickness, distance from the coal seam, strength, and internal layering were 36.50%, 33.06%, 24.47%, and 5.97%, respectively. Thickness and distance together accounted for 69.56% of the normalized relative sensitivity, indicating their greater influence on seismic energy response under the investigated conditions. These findings provide a reference for evaluating ETHS failure and mining-induced seismicity under similar geological and mining conditions.
In this study, discrete element method (DEM) simulations of vibratory peening are combined with a Taguchi L16 screening design, a face-centered central composite design, analysis of variance (ANOVA) and response surface methodology (RSM). The aim is to quantify how vibration amplitude (A = 0.5–2 mm), frequency (f = 20–50 Hz), shot diameter (d = 4–10 mm) and immersion depth (h = 20–50 mm) affect the maximum normal impact velocity of the media on an Almen strip, which is the only response modeled by RSM. The maximum contact force and the impact energy are reported as complementary descriptors of the impact conditions. Amplitude and frequency are the dominant factors, contributing 62.4% and 29.5% of the linear effects, respectively. Their statistically significant interaction shows that high-energy impact regimes are obtained when high amplitude and high frequency act together. Shot diameter and immersion depth act as secondary, dissipative modulators. The reduced quadratic model yields R2 = 91.36%, with R2adj = 89.81% and R2pred = 86.07%. The highest impact velocity (2.184 m/s; composite desirability 0.814) is predicted at A = 2 mm, f = 50 Hz, d = 4 mm and h = 20 mm, i.e. on the boundary of the investigated region. The results provide a simulation-based process map for vibratory peening; experimental validation remains the subject of future work.