Fact-checked by Grok 2 weeks ago

SPH

Smoothed particle hydrodynamics (SPH) is a meshless, Lagrangian computational method for simulating the dynamics of continuum media, such as fluids and deformable solids, by discretizing the domain into a set of particles that carry and interact with physical properties like mass, momentum, and energy.^[1] This particle-based approach approximates field variables through a kernel interpolation, enabling the modeling of complex phenomena like free-surface flows, multiphase interactions, and large deformations without the constraints of traditional grid-based methods.^[2] Developed independently in 1977, SPH was first proposed by Leon B. Lucy in the context of testing astrophysical fission hypotheses through numerical gasdynamic simulations.^[3] In the same year, Robert A. Gingold and Joseph J. Monaghan introduced the method for hydrodynamic modeling of non-spherical stars, establishing its foundational theory for arbitrary-dimensional spaces and polytropic stellar models.^[4] Originally tailored for astrophysical problems involving self-gravitating fluids, SPH's mesh-free nature quickly proved advantageous for handling irregular geometries and extreme deformations.^[5] Over the decades, SPH has evolved into a versatile tool across multiple disciplines, with key applications in engineering fields like ballistics simulations, where it excels in capturing high-speed impacts and material fragmentation.^[1] In ocean engineering, it models fluid-structure interactions, such as wave impacts on coastal structures, leveraging its ability to track free surfaces and interfaces accurately.^[6] Biomedical applications include simulating blood flow, tissue deformation, and surgical procedures, benefiting from the method's capacity to handle nearly incompressible materials.^[7] Recent advances have focused on improving numerical stability, convergence, and coupling with other techniques like finite element methods for hybrid simulations.^[8] Despite its strengths in robustness and adaptability, SPH faces challenges such as tensile instability and artificial viscosity, which can affect accuracy in certain regimes, prompting ongoing research into enhanced formulations like Godunov SPH and Riemann-solver-based schemes.^[9] Today, SPH remains a cornerstone in computational mechanics, supported by open-source codes and commercial software for high-fidelity predictions in both research and industry.^[10]

History and Development

Origins in Astrophysics

Smoothed-particle hydrodynamics (SPH) originated in the field of astrophysics during the late 1970s, developed as an innovative numerical method to simulate complex fluid behaviors in stellar and cosmological environments. The method was introduced independently in 1977 by Robert A. Gingold and Joseph J. Monaghan to model stellar dynamics and the fluid-like properties of matter under extreme conditions, such as those found in stars and gaseous nebulae, and by Leon B. Lucy for testing astrophysical fission hypotheses through numerical gasdynamic simulations.^[4]^[3] Their approach was motivated by the limitations of traditional grid-based Eulerian methods, which struggled to resolve the vast density contrasts and dynamic ranges in astrophysical phenomena, including the explosive outflows during supernova events. By representing fluids as a collection of particles with smoothed properties, SPH offered a Lagrangian framework that naturally adapted to irregular geometries and large deformations without fixed computational grids. This foundational work was detailed in their seminal paper published in the Monthly Notices of the Royal Astronomical Society. The core innovation of Gingold and Monaghan's method lay in its ability to handle the nonlinear and compressible flows prevalent in astrophysical simulations, where traditional techniques often failed due to numerical instabilities from shock waves and variable densities spanning many orders of magnitude. For instance, in modeling supernova explosions, grid-based codes could not efficiently track the rapid expansion of ejected material into the interstellar medium without excessive computational cost or artificial diffusion. SPH addressed these issues by interpolating physical quantities over a smoothing length, enabling robust simulations of self-gravitating fluids and providing insights into processes like stellar collapse and accretion disks. Their 1977 formulation emphasized the particle-based discretization of continuum equations, marking a shift toward meshless methods in computational astrophysics. Over time, the astrophysical roots of SPH paved the way for its adaptation to broader fluid dynamics problems beyond stellar contexts.

Evolution and Key Milestones

Following its initial development for astrophysical simulations, smoothed particle hydrodynamics (SPH) underwent significant expansion in the 1980s under the leadership of Joseph J. Monaghan, who adapted the method to general fluid dynamics problems beyond astrophysics. Monaghan's work emphasized Lagrangian formulations that conserved linear and angular momentum, enabling broader applications in hydrodynamics. A pivotal contribution was the 1983 formulation of artificial viscosity by Monaghan and Gingold, which introduced a viscous term to capture shocks and prevent particle interpenetration in high-Mach-number flows, using parameters α=1 and β=2 for signal velocity-based dissipation. This Monaghan formulation became a standard for handling compressive waves and has been refined in subsequent decades for stability. A landmark standardization occurred in Monaghan's 1985 review, which consolidated SPH principles for hydrodynamic simulations, deriving consistent equations for density, momentum, and energy while addressing numerical artifacts like tensile instabilities.^[11] This paper established SPH as a versatile meshless method for continuum mechanics, influencing implementations in both compressible and incompressible regimes. In the 1990s, advancements focused on improving boundary handling and initial multi-phase capabilities, addressing limitations in free-surface and interface problems. Key developments included repulsive boundary forces and dummy particle methods to enforce no-slip conditions without grid dependency, as proposed by Monaghan in 1994 for stable incompressible flows treated as weakly compressible. For multi-phase flows, early formulations by Johnson et al. in 1996 enabled simulations of material interfaces with distinct equations of state, using interface reconstruction to mitigate artificial mixing. These enhancements broadened SPH's utility for engineering applications like impact dynamics. The 2000s saw integration with adaptive resolution techniques and hybrid coupling to other numerical methods, enhancing efficiency for large-scale problems. Springel and Hernquist's 2002 hybrid multiphase model introduced variable smoothing lengths proportional to local density, allowing adaptive particle resolution in cosmological hydrodynamics while maintaining second-order accuracy. Coupling SPH to finite elements emerged as a high-impact approach for fluid-structure interactions; for instance, a 2004 algorithm by Rabczuk and Belytschko enabled seamless transitions between particle-based fluids and mesh-based solids, conserving momentum at interfaces for problems like blast loading. Recent milestones through 2025 have leveraged computational advances, including GPU acceleration and machine learning for kernel optimization. GPU-accelerated frameworks, such as the 2024 mixed-precision SPH implementation using FP16 for neighbor searches, achieved up to 1.5x speedup from mixed precision and overall GPU speedups up to 1000x for simulations up to 1 million particles.^[12] In parallel, physics-informed machine learning has enhanced kernel functions; a 2023 hierarchy of reduced Lagrangian models used neural networks to optimize SPH smoothing for turbulent flows, improving subgrid-scale predictions without explicit closures.^[13] These innovations have enabled real-time, high-fidelity simulations in fields like coastal engineering and additive manufacturing.

Core Principles

Particle Representation and Smoothing

In smoothed particle hydrodynamics (SPH), the continuous distribution of matter in a fluid is approximated by a finite set of discrete particles, each representing a small portion of the continuum and carrying essential physical properties such as mass m, position \mathbf{r}, velocity \mathbf{v}, and density \rho. These particles also store additional thermodynamic quantities like internal energy or pressure as needed for the simulation. Unlike grid-based methods, particles lack predefined connectivity or volume, serving instead as Lagrangian tracers that move with the local flow velocity to naturally follow the material deformation.^[4] The approximation of continuum fields in SPH relies on a smoothing process that interpolates particle properties to estimate values at arbitrary points in space. This is accomplished using a kernel function W(|\mathbf{r} - \mathbf{r}_b|, h), where \mathbf{r} is the evaluation position, \mathbf{r}_b denotes the position of particle b, and h is the smoothing length that sets the spatial extent of the kernel's influence. The kernel provides a smooth, localized weighting that diminishes rapidly beyond a few times h, effectively averaging contributions from neighboring particles within this compact support. Early formulations employed kernels like the Gaussian W(R, h) = \frac{1}{h \sqrt{\pi}} \exp\left( -\frac{R^2}{h^2} \right) in one dimension, while the cubic spline kernel became widely adopted for its compact support and accuracy.^[4]^[14] In one dimension, it is given by

W(R, h) = \frac{2}{3h} \begin{cases} 1 - \frac{3}{2} q^2 + \frac{3}{4} q^3 & 0 \leq q < 1 \\ \frac{1}{4} (2 - q)^3 & 1 \leq q < 2 \\ 0 & q \geq 2 \end{cases}

with q = R/h.^[14] The kernel function must satisfy a normalization condition to ensure the method's consistency for reproducing uniform fields:

\int W(\mathbf{r}, h) \, d\mathbf{r} = 1,

which holds over the entire domain and is verified for specific kernel choices like the Gaussian and cubic spline in their respective dimensions. This property guarantees that, in the continuum limit, SPH exactly interpolates constant quantities, providing a foundation for accurate discretization of differential equations.^[4] For simulations requiring variable spatial resolution, the smoothing length h is made adaptive by assigning a unique value h_a to each particle a, typically initialized and evolved based on local conditions such as h_a = \sigma (m_a / \rho_a)^{1/3} in three dimensions, where \sigma \approx 1 ensures the kernel encompasses a sufficient number of neighbors (around 20–50). The evolution of h_a follows the local density change via \frac{dh_a}{dt} = -\frac{1}{3} h_a \frac{1}{\rho_a} \frac{d\rho_a}{dt}, allowing automatic refinement in dense regions and coarsening in sparse ones to optimize computational resources without manual intervention.

Lagrangian Framework

Smoothed particle hydrodynamics (SPH) operates within a Lagrangian framework, where the computational domain is discretized into particles that advect with the local fluid velocity, thereby naturally tracking material properties and flow evolution without reliance on a fixed spatial grid. This approach, originally developed for astrophysical simulations, replaces the continuum fluid with a set of discrete particles, each carrying properties such as mass, position, velocity, and density, which evolve according to the underlying equations of motion.^[4] In this particle-following paradigm, the position of each particle \mathbf{r}_i updates via \frac{d\mathbf{r}_i}{dt} = \mathbf{v}_i, where \mathbf{v}_i is the particle velocity, ensuring that particles move precisely with the material flow and inherently capture Lagrangian coordinates. This eliminates the need for explicit tracking of convective transport, as the motion of particles automatically accounts for advection terms present in Eulerian formulations. Consequently, SPH excels in simulations involving substantial distortions, such as those in high-strain astrophysical events or free-surface flows, where the method seamlessly accommodates large deformations and topology alterations without grid distortion or remeshing.^[9] A key aspect of the Lagrangian framework in SPH is the estimation of density, which is computed locally for each particle through a summation over neighboring particles weighted by a smoothing kernel function W. The density \rho_i at particle i is given by

\rho_i = \sum_j m_j W(|\mathbf{r}_i - \mathbf{r}_j|, h_i),

where m_j is the mass of particle j, \mathbf{r}_i and \mathbf{r}_j are the positions, and h_i is the smoothing length associated with particle i. This formulation preserves mass conservation since particle masses remain constant, and the kernel W provides the interpolation basis for the density field.^[4] Compared to Eulerian methods, which fix the computational grid and require explicit discretization of convective fluxes—often leading to challenges like numerical diffusion or Courant-Friedrichs-Lewy (CFL) condition limitations in highly dynamic flows—the Lagrangian structure of SPH simplifies the treatment of material transport and interface tracking. This inherent Galilean invariance and adaptability make SPH particularly advantageous for problems with evolving geometries or multi-phase interactions, though it may introduce tensile instabilities in certain scenarios.^[9]

Mathematical Formulation

Kernel Approximation

In smoothed particle hydrodynamics (SPH), the kernel approximation provides the foundational method for interpolating continuous field quantities, such as density or velocity, across a set of discrete particles representing the continuum. This approach begins with an integral representation of a function A(\mathbf{r}) smoothed over a kernel function W:

A(\mathbf{r}) \approx \int A(\mathbf{r}') W(|\mathbf{r} - \mathbf{r}'|, h) \, d\mathbf{r}',

where h is the smoothing length that controls the resolution of the approximation. In the discrete particle formulation, this integral is replaced by a summation over neighboring particles b, weighted by their volumes m_b / \rho_b:

\langle A(\mathbf{r}_a) \rangle = \sum_b \frac{m_b}{\rho_b} A_b W_{ab}(h),

with W_{ab}(h) = W(|\mathbf{r}_a - \mathbf{r}_b|, h), \rho_b the density at particle b, and m_b its mass; this form ensures conservation properties when applied to governing equations. The kernel approximation was introduced in the seminal SPH formulations to handle irregular particle distributions without a fixed mesh. The kernel function W must satisfy several key properties to ensure accurate and stable approximations. It is normalized such that \int W(\mathbf{r}, h) \, d\mathbf{r} = 1, allowing the approximation to reproduce constant functions exactly. As h \to 0, W approaches the Dirac delta function \delta(\mathbf{r}), recovering the exact value in the continuum limit. Kernels are typically positive (W \geq 0) to preserve positivity of quantities like density, compactly supported (vanishing beyond a finite multiple of h, e.g., $2h) for computational efficiency by limiting interactions to nearby particles, and monotonically decreasing with distance to reflect local smoothing. These properties minimize truncation and discretization errors while enabling second-order accuracy for linear fields under uniform particle distributions. Common kernel choices balance smoothness, compactness, and computational cost; early SPH implementations favored the Gaussian kernel due to its analytic simplicity:

W(r, h) = \frac{1}{(\sqrt{\pi} h)^3} \exp\left(-\frac{r^2}{h^2}\right),

which has infinite support but is truncated at $3h in practice to approximate compactness, though this can lead to pair-wise inconsistencies. The cubic spline kernel, widely adopted for its compact support and C^1 continuity, is defined in three dimensions as:

W(r, h) = \frac{1}{\pi h^3} \begin{cases} 1 - \frac{3}{2}q^2 + \frac{3}{4}q^3 & 0 \leq q \leq 1, \\ \frac{1}{4}(2 - q)^3 & 1 < q \leq 2, \\ 0 & q > 2, \end{cases}

where q = r/h; it reproduces polynomials up to second degree exactly but can introduce tensile instabilities. Wendland functions, such as the C^2-Wendland kernel, offer higher-order smoothness and better boundary behavior, defined as compactly supported radial basis functions like W(q, h) = \alpha (1 - q/2)^4 (2q + 1) for $0 \leq q \leq 2 (with normalization \alpha = 21/(16 \pi h^3)), improving convergence in modern applications.^[15] For deriving gradients and other derivatives essential to SPH equations, the kernel approximation extends to:

\langle \nabla A(\mathbf{r}_a) \rangle \approx \sum_b \frac{m_b}{\rho_b} (A_b - A_a) \nabla_a W_{ab}(h),

where \nabla_a W_{ab} = - \nabla_b W_{ab} ensures antisymmetry, enhancing conservation of momentum; this differenced form reduces bias compared to direct summation \sum_b (m_b / \rho_b) A_b \nabla_a W_{ab}. The kernel derivative inherits properties like odd parity to maintain physical consistency.

Discretization of Governing Equations

In smoothed particle hydrodynamics (SPH), the fundamental conservation laws of mass, momentum, and energy are discretized by approximating the spatial derivatives in the governing partial differential equations using the kernel interpolation scheme. This approach replaces continuum field variables with sums over neighboring particles, weighted by the smoothing kernel, to derive ordinary differential equations for the evolution of particle properties such as density, velocity, and internal energy. The resulting formulations ensure numerical consistency and, through symmetric pairwise interactions, promote the conservation of physical quantities when integrated over all particles. The continuity equation, which governs mass conservation, is discretized in SPH as

\frac{d \rho_i}{dt} = \sum_j m_j (\mathbf{v}_i - \mathbf{v}_j) \cdot \nabla_i W_{ij},

where \rho_i is the density of particle i, m_j is the mass of particle j, \mathbf{v}_i and \mathbf{v}_j are the velocities, and W_{ij} is the kernel function evaluated at the separation between particles i and j. This form arises from applying the kernel approximation to the material derivative of density and leverages the antisymmetric nature of the relative velocity and kernel gradient terms across particle pairs to maintain total mass conservation exactly. The momentum equation, describing the conservation of linear momentum, takes the symmetric form

\frac{d \mathbf{v}_i}{dt} = -\sum_j m_j \left( \frac{P_i}{\rho_i^2} + \frac{P_j}{\rho_j^2} + \Pi_{ij} \right) \nabla_i W_{ij},

with P_i and P_j denoting the pressures, and \Pi_{ij} representing the artificial viscosity term that stabilizes the scheme against interparticle penetration. The symmetric averaging of pressure gradients ensures that the force exerted by particle j on i is equal and opposite to that on j from i, thereby conserving total linear and angular momentum in isolated systems. For energy conservation, the evolution of the specific internal energy u_i is given by

\frac{d u_i}{dt} = \frac{1}{2} \sum_j m_j \left( \frac{P_i}{\rho_i^2} + \frac{P_j}{\rho_j^2} + \Pi_{ij} \right) (\mathbf{v}_i - \mathbf{v}_j) \cdot \nabla_i W_{ij}.

This expression, derived from the work done by pressure and viscous forces, incorporates a factor of $1/2 to avoid double-counting pairwise contributions and maintains consistency with the momentum equation. The symmetric structure guarantees that total energy is conserved, as gains and losses balance across interacting particles, provided the artificial viscosity is formulated appropriately. Antisymmetric components in the velocity differences further support angular momentum preservation in these discretizations.

Numerical Methods

Time Integration Schemes

In Smoothed Particle Hydrodynamics (SPH), time integration schemes are essential for advancing the positions, velocities, and other properties of particles while maintaining numerical stability and conserving physical quantities such as energy and momentum. These schemes discretize the ordinary differential equations governing particle motion, typically derived from the Lagrangian form of the Navier-Stokes equations. Explicit methods dominate due to their simplicity and efficiency in many applications, but implicit approaches are employed when stiffness arises, such as in highly viscous flows. Variable time stepping enhances adaptability to local flow conditions, ensuring computational efficiency without compromising accuracy. The leapfrog-Verlet integrator is a widely adopted explicit scheme in SPH simulations, prized for its symplectic nature that preserves energy over long times in conservative systems. This second-order method updates positions and velocities in a staggered manner, avoiding the evaluation of forces at the same time level as positions, which reduces errors in Hamiltonian systems. The update equations are given by:

\mathbf{r}_b^{n+1} = \mathbf{r}_b^n + \Delta t \, \mathbf{v}_b^{n+1/2},

\mathbf{v}_b^{n+3/2} = \mathbf{v}_b^{n+1/2} + \Delta t \, \mathbf{a}_b^{n+1},

where \mathbf{r}_b is the position, \mathbf{v}_b the velocity, \mathbf{a}_b the acceleration, and \Delta t the time step for particle b at step n. This formulation ensures time-reversibility and good long-term stability, making it suitable for astrophysical and free-surface flow simulations. Time step selection in explicit SPH is constrained by the Courant-Friedrichs-Lewy (CFL) condition to prevent instability from information propagating faster than the numerical scheme can resolve. The condition typically limits \Delta t \propto h / c_s, where h is the smoothing length and c_s the sound speed, ensuring the time step is a fraction (often 0.1–0.25) of the acoustic crossing time over the kernel support. Additional constraints from viscous diffusion or particle forces may further restrict \Delta t, such as \Delta t \propto h^2 / \nu for viscosity \nu. This adaptive limit maintains stability in compressible flows with varying densities and speeds.^[16] For stiff problems like high-viscosity flows, implicit schemes offer larger time steps by solving coupled equations that account for feedback between velocities and forces. The backward Euler method, a first-order implicit integrator, is commonly used in viscous SPH to handle the stiffness from diffusion terms, formulating the velocity update as a linear system solved via iterative methods like conjugate gradients. For instance, the viscous acceleration contribution is discretized as \Delta \mathbf{v}_i = \Delta t \sum_j m_j (\frac{\mathbf{s}_i}{\rho_i^2} + \frac{\mathbf{s}_j}{\rho_j^2}) \nabla W_{ij}, leading to a sparse matrix equation for all particles. This approach allows time steps orders of magnitude larger than explicit limits, improving efficiency in simulations of thick fluids or coupled boundary interactions.^[17] Variable time stepping refines efficiency by assigning individual \Delta t_b to each particle based on local conditions, such as the CFL number or maximum acceleration. In hierarchical schemes, particles synchronize at common levels, with \Delta t_b = \min(\eta h_b / c_{s,b}, \sqrt{h_b / |\mathbf{a}_b|}), where \eta \approx 0.25 and \mathbf{a}_b is acceleration; this reduces computational cost in multi-scale problems like shocks or clustered particles. Such methods are standard in production codes for balancing accuracy and speed.^[16]

Boundary Handling Techniques

In Smoothed Particle Hydrodynamics (SPH), boundary handling is essential due to the absence of a fixed mesh, which complicates the enforcement of conditions at domain edges or solid interfaces. Traditional approaches rely on auxiliary particles or force fields to approximate boundary effects without altering the core Lagrangian framework. These methods ensure conservation properties while maintaining computational efficiency, particularly for complex geometries where grid-based techniques falter.^[18] Ghost particles, also known as dummy particles, are a widely adopted technique for imposing boundary conditions in SPH simulations. These virtual particles are placed symmetrically across the boundary relative to fluid particles, mirroring their properties to extend the kernel support and prevent truncation errors near walls. For no-slip conditions, ghost particles are assigned velocities opposite to the adjacent fluid particles, enforcing zero relative velocity at the boundary, while free-slip conditions set tangential velocities to match and normal components to zero. This method, introduced in early SPH implementations, has been refined for arbitrary geometries by dynamically generating ghost particles at each time step based on boundary normals.^[19]^[20] Repulsive boundary forces provide an alternative or complementary approach to prevent fluid particles from penetrating solid boundaries, often modeled as fixed layers of boundary particles. A common formulation applies a Lennard-Jones-like repulsive force on fluid particles within a cutoff distance of the boundary, given by

\mathbf{F} = -k d \mathbf{n},

where k is a stiffness coefficient, d is the signed distance to the boundary surface, and \mathbf{n} is the outward unit normal vector. This force acts only when d < 0, pushing particles away to simulate impenetrability without explicit velocity enforcement. Such forces are particularly effective for curved or moving boundaries, as they integrate seamlessly with the particle dynamics and reduce numerical diffusion compared to purely kinematic methods.^[21]^[22] For open boundaries involving inflow and outflow, SPH employs particle injection and deletion to maintain mass conservation and flow continuity. In inflow regions, new particles are continuously introduced with prescribed velocity and density profiles, positioned along the boundary to match the incoming flux, while outflow regions delete particles that exit beyond a buffer zone to avoid artificial reflections. These techniques often incorporate damping or extrapolation in buffer layers to stabilize the simulation, ensuring smooth transitions without disrupting the overall particle distribution. A key challenge addressed is the accurate specification of pressure or velocity at inlets, typically achieved through semi-analytical Riemann invariants.^[23]^[24] In the 2020s, Riemann-based boundary conditions have emerged as advanced methods to enhance accuracy in SPH, particularly for compressible and multiphase flows. These approaches solve local Riemann problems between fluid and boundary states to compute stable interface fluxes, incorporating ghost-like extrapolations while minimizing dissipation. For instance, semi-analytical Riemann solvers enforce unsteady open boundaries by propagating characteristic waves, improving convergence for high-speed inflows or shocks. Recent implementations, such as those coupling Riemann SPH with dynamic particle shifting, demonstrate reduced boundary-induced errors in complex domains, with validation against analytical solutions compared to traditional ghost methods.^[25]^[26]^[27]

Modeling Specific Physics

Hydrodynamics and Compressibility

Smoothed Particle Hydrodynamics (SPH) formulations for hydrodynamics primarily solve the Navier-Stokes equations in a Lagrangian framework, discretizing the fluid as a set of particles that carry mass, momentum, and other properties, with interactions mediated through kernel functions. Basic fluid flow is captured by evolving the continuity equation for density and the momentum equation for velocity, where pressure gradients drive acceleration. Compressibility handling is central, as SPH must balance computational efficiency with physical accuracy in enforcing the divergence-free velocity condition for liquids or allowing controlled density variations for gases. Incompressible SPH (ISPH) enforces the incompressibility constraint \nabla \cdot \mathbf{v} = 0 through projection methods, which correct the velocity field to eliminate divergence after an intermediate advection step. This involves solving a Poisson equation for pressure, \nabla^2 P = \frac{\rho_0}{\Delta t} \nabla \cdot \mathbf{v}^*, where \mathbf{v}^* is the uncorrected velocity, \rho_0 is the reference density, and \Delta t is the time step; the pressure gradient then adjusts velocities to satisfy incompressibility.^[28] Pioneered by Cummins and Rudman, this approach avoids explicit equations of state by treating pressure as a Lagrange multiplier, enabling stable simulations of low-Mach-number flows like sloshing or dam-breaking without spurious oscillations from artificial compressibility.^[28] Weakly compressible SPH (WCSPH) approximates incompressibility by permitting small density fluctuations, using an artificial equation of state to link pressure to density deviations: P = c^2 (\rho - \rho_0), where c is an artificial sound speed chosen such that the Mach number remains low (typically Ma < 0.1) to mimic incompressible behavior while avoiding the cost of solving a Poisson equation iteratively. Introduced by Morris et al.,^[29] this stiff EOS generates large pressure responses to minor density changes, facilitating explicit time integration but requiring small time steps proportional to h/c, where h is the smoothing length. WCSPH is widely used for free-surface flows due to its simplicity and natural handling of variable densities. Density evolution in SPH hydrodynamics can employ either direct summation, \rho_i = \sum_j m_j W(\mathbf{r}_i - \mathbf{r}_j, h), which approximates the density field via kernel interpolation and ensures exact mass conservation locally, or evolution via the continuity equation, \frac{d\rho_i}{dt} = -\rho_i \sum_j m_j (\mathbf{v}_i - \mathbf{v}_j) \cdot \nabla W_{ij}/\rho_j, which integrates density changes from velocity divergences and better preserves total mass globally but can introduce errors near free surfaces. The summation approach, original to Gingold and Monaghan, suffers from particle clustering or depletion at boundaries, leading to inaccurate pressures, while the continuity method, favored in later formulations, mitigates these but requires careful initialization. A key challenge in WCSPH is tensile instability, where negative pressures in stretching regions cause particle clumping and unphysical fragmentation, arising from the monotonic kernel and linear EOS under tension.^[30] Monaghan addressed this by introducing artificial stresses that shift the EOS to produce repulsive forces in tensile regimes, effectively stabilizing simulations of coherent structures like jets without altering compressive behavior.^[30] This modification has become standard in WCSPH implementations for improved robustness in high-strain flows. For shock-capturing in compressible hydrodynamic regimes, Godunov SPH employs Riemann solvers to compute fluxes across particle interfaces, treating each pair as a one-dimensional Riemann problem solved exactly or approximately to determine time-averaged states that drive momentum and energy updates. Developed by Inutsuka,^[31] this reformulation enhances post-shock accuracy and reduces oscillations compared to standard SPH differencing, particularly for strong discontinuities in astrophysical contexts like supernova remnants, by incorporating signal speeds from the solver into the time step and dissipation terms.

Viscosity and Additional Effects

In smoothed particle hydrodynamics (SPH), viscosity is typically incorporated through artificial viscosity terms designed to handle shocks and prevent particle interpenetration while providing numerical dissipation analogous to physical viscous effects. The standard artificial viscosity formulation adds a pressure-like term \Pi_{ij} to the momentum equation between particles i and j, given by

\Pi_{ij} = \frac{ -\alpha \mu_{ij} c_{ij} + \beta \mu_{ij}^2 }{\rho_{ij}},

where \mu_{ij} = h (\mathbf{v}_i - \mathbf{v}_j) \cdot (\mathbf{r}_i - \mathbf{r}_j) / (|\mathbf{r}_{ij}|^2 + \epsilon h^2), c_{ij} is the average sound speed, \rho_{ij} is the average density, h is the smoothing length, \alpha and \beta are user-defined coefficients (typically \alpha \approx 1, \beta \approx 2), and \epsilon is a small constant to avoid singularities. This form ensures monotonicity and stability by mimicking the signal speed in shocks, with the linear term providing dissipation proportional to the sound speed and the quadratic term capturing von Neumann-Richtmyer viscosity for strong shocks. The Monaghan-Gingold artificial viscosity, an early influential variant, emphasizes dissipation tied to the signal velocity while minimizing post-shock oscillations, forming the basis for many subsequent implementations in hydrodynamic simulations.^[32] It was developed specifically for shock tube problems and has been widely adopted due to its balance of accuracy and computational efficiency in capturing dissipative physics without excessive numerical diffusion.^[32] Surface tension in SPH is often modeled using the continuum surface force (CSF) approach, which distributes the capillary pressure P = \sigma \kappa—where \sigma is the surface tension coefficient and \kappa is the local mean curvature—as a body force across particles near the interface. Curvature \kappa is approximated by first computing particle surface normals from the divergence of a color field or density gradient, then applying an SPH symmetrized Laplacian to estimate \kappa = \nabla \cdot \hat{\mathbf{n}}, where \hat{\mathbf{n}} is the unit normal. This method effectively reproduces droplet dynamics and coalescence, though it requires careful kernel choices to avoid tensile instabilities at interfaces. Additional effects like heat transfer are included via diffusive terms in the energy equation, approximating the conduction flux \nabla \cdot (k \nabla T) using SPH second derivatives, such as the symmetric Laplacian form \sum_j m_j (k_i + k_j) (T_j - T_i) / \rho_j \nabla_i W_{ij}, where k is thermal conductivity, T is temperature, and W_{ij} is the kernel. This discretization conserves total thermal energy and is particularly useful for problems involving temperature gradients, such as in astrophysical or multiphase flows, by ensuring the flux is anti-symmetric between particles.

Applications

Fluid Dynamics Simulations

Smoothed Particle Hydrodynamics (SPH) has become a prominent method for simulating complex engineering fluid dynamics problems, particularly those involving free-surface flows and multiphase interactions, due to its mesh-free nature that naturally handles large deformations and topology changes.^[33] In applications such as sloshing in tanks, SPH effectively captures violent fluid motions and impacts against tank walls, as demonstrated in simulations of multi-phase sloshing flows where the method accurately reproduces pressure distributions and free-surface evolution under intense excitations.^[34] Similarly, for wave propagation, SPH models in numerical wave tanks simulate the generation and evolution of waves with high fidelity, enabling the study of nonlinear wave interactions in coastal environments.^[35] Droplet impact simulations using SPH further highlight its utility, where the approach resolves the detailed dynamics of liquid drop spreading, rebounding, or splashing on dry or wet surfaces, providing insights into wetting phenomena critical for industrial processes.^[36] Multiphase SPH formulations enhance interface tracking in simulations involving immiscible fluids, employing techniques such as color field methods to maintain sharp interfaces by assigning phase labels to particles and reducing artificial mixing through corrected kernel gradients.^[37] Level set approaches integrated into SPH further refine multiphase modeling by embedding interfaces as zero-level contours of a signed distance function, allowing robust handling of complex topologies like droplet coalescence or bubble entrainment without explicit surface reconstruction.^[38] These methods have been validated in benchmarks showing good agreement with experimental data on interface sharpness and mass conservation during multiphase interactions.^[37] In industrial contexts, SPH via codes like DualSPHysics has been widely adopted for coastal engineering applications, including the modeling of wave overtopping and harbor oscillations, where it simulates real-sea wave impacts on structures with resolutions up to millions of particles for practical-scale problems.^[39] For oil spill modeling, SPH simulates the diffusion and containment of spilled oil under currents and waves, as in studies of boom performance that capture oil entrainment and leakage mechanisms, aiding in the design of effective barriers post-incidents like the 2010 Deepwater Horizon spill.^[40]^[41] Validation against experiments, such as the classic dam-break test, confirms SPH's accuracy; for instance, 3D simulations of dam-break waves impacting obstacles reproduce experimental water levels, pressures, and run-up heights with errors below 10% in benchmark cases.^[42] These validations underscore SPH's reliability for predicting transient free-surface flows in engineering scenarios.

Astrophysics and Solid Mechanics

In astrophysics, Smoothed Particle Hydrodynamics (SPH) has been extensively applied to simulate complex gravitational interactions and dynamical processes, leveraging its ability to handle free boundaries and large density contrasts without a fixed mesh. For galaxy collisions, SPH codes model the tidal disruptions, gas inflows, and star formation triggered by mergers, capturing the evolution of gaseous disks and dark matter halos over cosmic timescales. The GADGET code, a parallel TreeSPH implementation, has been pivotal in these simulations, enabling high-resolution modeling of galaxy interactions in cosmological contexts, such as the formation of elliptical galaxies from major mergers.^[43] SPH also facilitates simulations of planet formation by resolving the collapse of protoplanetary disks and the accretion of planetesimals, where self-gravity and viscous heating play key roles. In these scenarios, hybrid SPH-N-body approaches track the growth of planetary embryos amid turbulent gas flows, providing insights into the initial conditions for solar system architectures. For star cluster dynamics, SPH combined with N-body methods simulates the violent relaxation and mass segregation in young clusters, accounting for stellar feedback and binary interactions that influence cluster survival. The SEREN code exemplifies this by integrating SPH hydrodynamics with direct N-body integration for multi-scale star and planet formation processes.^[44] In solid mechanics, total Lagrangian SPH formulations are employed to model materials undergoing large deformations, where the reference configuration remains fixed, ensuring conservation properties and stability for hyperelastic and elastoplastic behaviors. This approach computes the deformation gradient from particle positions in the initial state, avoiding distortions associated with Eulerian updates. A key component is the constitutive relation for the second Piola-Kirchhoff stress tensor \mathbf{S}, given by the linear elastic model:

\mathbf{S} = 2\mu \boldsymbol{\varepsilon} + \lambda \operatorname{tr}(\boldsymbol{\varepsilon}) \mathbf{I}

where \boldsymbol{\varepsilon} is the Green-Lagrange strain tensor, \mu and \lambda are the Lamé constants, and \mathbf{I} is the identity tensor; this relation extends Hooke's law to finite strains in SPH discretizations.^[45] Fluid-structure interaction (FSI) simulations using SPH often adopt partitioned coupling schemes, where the fluid and solid domains are solved separately and interfaced via iterative data exchange. Penalty methods enforce kinematic continuity at the interface by adding corrective forces proportional to penetration, stabilizing the coupling without requiring a monolithic solver. This technique has been implemented in SPH-FE hybrids to handle deformable solids in dynamic fluid environments, such as wave impacts on elastic barriers.^[46] Representative examples include SPH modeling of asteroid impacts, where high-velocity collisions between rubble-pile bodies are simulated to predict fragment size distributions and ejecta plumes, as in studies of 100-km-scale disruptions at speeds up to 15 km/s. In solid mechanics, recent 2020s simulations have used total Lagrangian SPH to track fracture propagation in brittle materials under dynamic loading, such as rock samples with pre-existing cracks, revealing branching patterns and energy dissipation without mesh entanglement.^[47]^[48]

Advantages, Limitations, and Extensions

Strengths in Mesh-Free Simulations

One of the primary strengths of Smoothed Particle Hydrodynamics (SPH) lies in its adaptive resolution capabilities, achieved through the variation of the smoothing length [h](/page/H+), which allows the method to dynamically adjust spatial resolution based on local flow conditions. This variable [h](/page/H+) enables finer resolution in regions of high gradients, such as near free surfaces or shocks, while coarsening elsewhere to optimize computational efficiency. In adaptive SPH formulations, particle splitting and merging further enhance this adaptability: particles are split into multiple entities when resolution demands increase (e.g., based on a mass ratio threshold), conserving mass and momentum, and merged when density allows, naturally handling fragmentation and coalescence without manual intervention. This approach is particularly effective for simulations involving breaking waves or material breakup, as demonstrated in free-surface flow models where particle spacing varies proportionally with distance from the interface.^[49] SPH also exhibits strong conservation properties, particularly in symmetric formulations that ensure exact conservation of linear and angular momentum, as well as energy, for interpolated linear fields. These formulations derive from variational principles, guaranteeing that the discrete particle system preserves key invariants like circulation and potential vorticity, even under time evolution, without artificial diffusion. For instance, in applications to geophysical fluid dynamics, symmetric pressure gradient terms maintain total energy and momentum balance across the particle ensemble, providing a robust framework for long-term simulations where conservation errors could otherwise accumulate. This exactness for linear fields stems from the method's Lagrangian nature and kernel symmetrization, outperforming some mesh-based methods in maintaining global integrals.^[50] The mesh-free, particle-based structure of SPH facilitates straightforward parallelization, especially on GPUs, owing to the inherent locality of particle interactions—each particle only communicates with a compact neighborhood defined by h. This locality minimizes data dependencies, enabling efficient neighbor searches via spatial hashing or bucketing, which map particles to grid cells for rapid querying. Implementations on multi-GPU systems further exploit this by assigning particles independently across devices with periodic load balancing, achieving near-linear speedups; for example, a simulation with 6.5 million particles in solid mechanics showed a 2.36-fold acceleration on three GPUs compared to a single GPU, demonstrating effective parallelization for large-scale problems. Such scalability makes SPH ideal for high-fidelity, real-time computations in complex dynamics.^[51]^[52] Additionally, SPH excels in modeling free surfaces and moving boundaries, as its Lagrangian particles naturally track interfaces without the need for remeshing or special tracking algorithms required in Eulerian methods. Free surfaces emerge organically from the particle distribution, with surface tension or pressure discontinuities handled via kernel corrections, avoiding distortion issues in deforming domains. For moving boundaries, such as in fluid-structure interactions, particles can represent or ghost boundaries seamlessly, enabling simulations of extreme deformations like those in friction stir extrusion, where material mixing and interfaces are captured accurately over large strains. This remeshing-free approach is crucial for applications involving multiphase flows or impacts, where traditional grids would fail due to topological changes.^[53]

Common Challenges and Improvements

One of the primary challenges in Smoothed Particle Hydrodynamics (SPH) is tensile instability, which leads to unphysical clumping of particles under tensile stress, particularly in simulations involving negative pressures or material failure.^[30] This instability arises from the pairwise nature of SPH interactions and can degrade accuracy in problems like elastic solids or high-strain fluids.^[54] Boundary deficiencies represent another key issue, where particles near free surfaces or solid walls experience incomplete kernel support, resulting in erroneous density estimates and pressure gradients that cause surface tension artifacts or penetration.^[55] Particle clustering, often exacerbated by tensile instability or large smoothing lengths, further complicates simulations by creating numerical voids and reducing resolution in clustered regions.^[56] To address tensile instability, artificial stress formulations have been developed, such as Monaghan's artificial pressure, which introduces a repulsive force proportional to tensile strain to prevent clumping without altering compressive behavior.^[30] Normalized kernels improve boundary handling by renormalizing the kernel sum to unity near boundaries, mitigating deficiency errors and enhancing accuracy in free-surface flows.^[57] The δ+SPH scheme, an extension of the original δ-SPH model, incorporates adaptive diffusive terms in the continuity equation to stabilize weakly compressible flows, reducing oscillations and improving energy conservation in violent impact scenarios.^[58] Hybrid SPH-FEM couplings combine SPH's mesh-free advantages for fluids or fracturing solids with FEM's robustness for structured domains, enabling seamless data transfer at interfaces for fluid-structure interaction problems like metal cutting or casting.^[59] Recent developments as of 2025 include AI-optimized particle distributions, such as differentiable SPH frameworks that leverage machine learning for adjoint optimization and inverse design in fluid dynamics, allowing gradient-based refinement of particle layouts to minimize clustering.^[60] Entropy-stable formulations have also advanced, with schemes ensuring discrete entropy conservation in thermo-elastic simulations to enhance long-term stability in high-strain regimes.^[61] Popular open-source codes implementing these improvements include PySPH, a Python-based framework supporting high-performance SPH with GPU acceleration for general fluid simulations; GPUSPH, a CUDA-optimized solver focused on free-surface flows; and LAMMPS extensions, which integrate SPH for large-scale molecular and continuum mechanics.^[62]^[63]

References

[1]
Smooth Particle Hydrodynamics - an overview | ScienceDirect Topics
Smooth Particle Hydrodynamics (SPH) is a computational method for simulating fluid dynamics, using points as particles, and is a meshless Lagrangian technique.
[2]
What is Smoothed-particle Hydrodynamics Simulation? - Ansys
Oct 11, 2023 · Smoothed-particle hydrodynamics (SPH) is a particle-based, mesh-free method for solid-fluid applications. It discretizes flow into elements ...
[3]
A numerical approach to the testing of the fission hypothesis. - ADS
A finite-size particle scheme for the numerical solution of twoand three-dimensional gasdynamic problems of astronomical interest is described and tested.
[4]
Smoothed particle hydrodynamics: theory and application to non ...
A new hydrodynamic code applicable to a space of an arbitrary number of dimensions is discussed and applied to a variety of polytropic stellar models.
[5]
Smoothed particle hydrodynamics - IOPscience
Jul 5, 2005 · In this review the theory and application of Smoothed particle hydrodynamics (SPH) since its inception in 1977 are discussed.Missing: history | Show results with:history
[6]
(PDF) Smoothed particle hydrodynamics and its applications in fluid ...
This paper is dedicated to a review of the recent developments of SPH method and its typical applications in fluid-structure interactions in ocean engineering.<|control11|><|separator|>
[7]
On methodology and application of smoothed particle ... - NIH
This paper aims to provide a comprehensive overview on the recent advances of SPH method in the fields of fluid, solid, and biomechanics.
[8]
Review of smoothed particle hydrodynamics: towards converged ...
Sep 9, 2020 · While original applications were in astrophysics, early engineering applications showed the versatility and robustness of the method without ...
[9]
The Mathematics of Smoothed Particle Hydrodynamics (SPH ...
In this paper, an overview of the mathematical aspects of the SPH consistency is presented with a focus on the most recent developments.
[10]
Smoothed Particle Hydrodynamics Techniques for the Physics ...
Sep 15, 2020 · Smoothed Particle Hydrodynamics (SPH) is a spatial discretization technique used for simulating fluids, highly viscous materials, and ...
[11]
Particle methods for hydrodynamics - ScienceDirect.com
M A G I , a three-dimensional shock and material response code which is based on smoothed particle hydrodynamics (SPH) is described.
[12]
A GPU accelerated mixed-precision Smoothed Particle ... - arXiv
Nov 9, 2023 · This paper presents a GPU-accelerated SPH framework using FP16 for NNPS with a RCLL algorithm, achieving significant speedups using GPU ...Missing: milestones 2020s
[13]
Hierarchy of reduced Lagrangian models of turbulence
May 5, 2023 · Physics-informed machine learning with smoothed particle hydrodynamics: Hierarchy of reduced Lagrangian models of turbulence. Michael ...
[14]
3. SPH formulation · DualSPHysics/DualSPHysics Wiki - GitHub
May 26, 2023 · 3.2.1 Artificial Viscosity. The artificial viscosity scheme, proposed by [Monaghan, 1992], is a common method within fluid simulation ...
[15]
[PDF] Implicit Formulation for SPH-based Viscous Fluids - GAMMA
In contrast to previous methods that use explicit integration, our method uses an implicit formulation to improve the robustness of viscosity integration, ...<|control11|><|separator|>
[16]
A simplified technique to enforce solid boundary conditions in SPH
The ghost particles approach is easy to implement when the geometry of the boundary has a rectangular (2D) or a cuboid (3D) shape, but difficulties arise when ...
[17]
An arbitrary boundary with ghost particles incorporated in coupled ...
The ghost particles are produced automatically for every fluid particle at each time step, and the properties of ghost particles, such as density, mass and ...
[18]
[PDF] Ghost SPH for Animating Water - UBC Computer Science
We propose a new ghost fluid approach for free surface and solid boundary conditions in Smoothed Particle Hydrodynamics (SPH) liquid simulations.
[19]
Robust solid boundary treatment for compressible smoothed particle ...
Aug 22, 2024 · Dynamic Boundary Condition (DBC) consider multiple layers of ghost particles to model the solid boundaries, as shown in Fig. 1(d). The density ...
[20]
An enhanced treatment of boundary conditions in implicit smoothed ...
Feb 22, 2014 · A repulsive force is also applied for boundary particles to prevent fluid particles from unphysical penetration through solid boundaries. The ...
[21]
Inflow/outflow pressure boundary conditions for smoothed particle ...
Dec 15, 2017 · The boundary condition treatment is one of the main issues in the SPH method. Peculiar difficulties are encountered when considering inflow and ...
[22]
Extrapolation Boundary Conditions for 2‐D Smoothed Particle ...
May 21, 2025 · The defined method utilises three layers of particles, called ghost particles herein, for both inflow and outflow operations. Additionally, ...2 Sph Method And Operators · 4 Numerical Results · 4.4 Efflux Discharge Over A...
[23]
Unsteady open boundaries for SPH using semi-analytical conditions ...
In the framework of Weakly Compressible SPH (WCSPH), using Riemann solvers can partially solve this issue [5], [6], but modelling a complex boundary where the ...
[24]
An approach for permeable boundary conditions in SPH
Nov 1, 2021 · Ferrand et al. Unsteady open boundaries for sph using semi-analytical conditions and Riemann solver in 2d. Comput. Phys. Commun. (2017). E ...
[25]
A Riemann-based two-phase two-layer SPH method for simulating ...
This study proposes a new two-phase two-layer Riemann-smoothed particle hydrodynamics (SPH) numerical model.
[26]
An SPH Projection Method - ScienceDirect.com
The SPH projection method enforces incompressibility by projecting the velocity field onto a divergence-free space using a pressure Poisson equation. It uses a ...
[27]
SPH without a Tensile Instability - ScienceDirect.com
Apr 10, 2000 · In this paper it is shown how the instability can be removed by using an artificial stress which, in the case of fluids, is an artificial pressure.Missing: WCSPH | Show results with:WCSPH
[28]
Shock simulation by the particle method SPH - ScienceDirect.com
The particle method SPH is applied to one-dimensional shock tube problems by incorporating an artificial viscosity into the equations of motion.
[29]
Smoothed particle hydrodynamics (SPH) for complex fluid flows
Jan 10, 2019 · So far the MPS method has been widely applied to modeling complex fluid flows especially free surface flows with fluid-structure interactions.
[30]
Simulating multi-phase sloshing flows with the SPH method
In this paper, the intense sloshing flow is simulated by Smoothed Particle Hydrodynamics (SPH) method successfully.
[31]
Wave Propagation Studies in Numerical Wave Tanks with Weakly ...
Generation and propagation of waves in a numerical wave tank constructed using Weakly Compressible Smoothed Particle Hydrodynamics (WCSPH) are considered here.
[32]
Simulation of liquid drop impact on dry and wet surfaces using SPH ...
This paper describes an advanced numerical method, based on smoothed particle hydrodynamics (SPH), for predicting the detailed outcomes of drop–wall ...Simulation Of Liquid Drop... · 2. Sph Method · 3. Results And Discussions<|separator|>
[33]
An SPH model for multiphase flows with complex interfaces and ...
Feb 15, 2015 · In this paper, an improved SPH model for multiphase flows with complex interfaces and large density differences is developed.
[34]
Smoothed particle hydrodynamics modelling of multiphase flows
Dec 12, 2023 · Smoothed particle hydrodynamics (SPH) is a meshless, particle-based approach that has been increasingly applied for modelling of various fluid-flow phenomena.
[35]
SPH Simulations of Real Sea Waves Impacting a Large-Scale ...
This research represents the first known application of SPH open boundary conditions to model a real-world engineering case.2. Study Case · 3. Numerical Model · 4. Results And Discussion
[36]
Numerical modelling of boom and oil spill with SPH - ScienceDirect
The protection of coastal areas against oil pollution is often addressed with the use of floating booms. These bodies are subject to an empirical design ...
[37]
Research on oil boom performance based on Smoothed Particle ...
Jul 27, 2023 · The DualSPHysics solver is employed for numerical simulations, incorporating optimized SPH techniques and eight different skirt configurations ...
[38]
A Numerical Validation of 3D Experimental Dam-Break Wave ... - MDPI
A numerical investigation of a 3D dam-break impacting a sharp obstacle was carried out with the SPH-based solver DualSPHysics. The DualSPHysics code was tested ...
[39]
The cosmological simulation code gadget-2 - Oxford Academic
We discuss the cosmological simulation code gadget-2, a new massively parallel TreeSPH code, capable of following a collisionless fluid with the N-body method.
[40]
SEREN - A new SPH code for star and planet formation simulations
Feb 3, 2011 · We present SEREN, a new hybrid Smoothed Particle Hydrodynamics and N-body code designed to simulate astrophysical processes such as star and planet formation.
[41]
Thermomechanical total Lagrangian SPH formulation for solid ...
This paper presents a thermomechanical SPH in total Lagrangian formulation to simulate efficiently large deformations thermomechanical problems.
[42]
A 3D SPH–FE coupling for FSI problems and its application to tire ...
The purpose of this work is to analyze the SPH–FE coupling capabilities for modeling efficiently such a complex phenomenon. On the fluid side, the SPH method is ...
[43]
SPH simulations of high-speed collisions between asteroids and ...
Sep 1, 2022 · We computed a standard set of 125 simulations of collisions between representative asteroids and high-speed icy projectiles (comets), in the range 8 to 15 km/s.Missing: impacts | Show results with:impacts
[44]
An improved SPH method for simulating crack propagation and ...
Apr 14, 2023 · In this paper, an improved SPH method is proposed to simulate the brittle fracture process of rock samples with pre-existing cracks. The present ...
[45]
[PDF] Smoothed particle hydrodynamics with adaptive spatial resolution ...
This particle shifting technique is further modified in this work for use with variable smoothing length to improve the particle distribution. The position of a ...
[46]
[PDF] CONSERVATION PROPERTIES OF SMOOTHED PARTICLE ...
Mass conservation is intrinsic to the SPH formulation. It is also well-known that SPH can be derived from a variational principle (see, e.g., [14, 3, 10]); ...
[47]
[PDF] Smoothed Particle Hydrodynamics on GPUs
Abstract In this paper, we present a Smoothed Parti- cle Hydrodynamics (SPH) implementation algorithm on. GPUs. To compute a force on a particle, ...
[48]
[PDF] A novel multi-GPU parallelization paradigm for SPH applied to solid ...
Once the data is received, each GPU can compute its local particle interaction with the MGPU boundary particles. g. Each GPU can integrate its equations. Note: ...Missing: locality | Show results with:locality
[49]
Meshfree simulation and experimental validation of extreme ...
Nov 1, 2021 · The particle representation in SPH also enables a natural description of material interfaces, free surfaces, and moving boundaries during the ...
[50]
SPH without a Tensile Instability - ADS
These problems include the evolution of a region with negative pressure, extreme expansion in one dimension, and the collision of rubber cylinders. To study ...Missing: boundary | Show results with:boundary
[51]
On the correction of the boundary deficiency in SPH for the frictional ...
Dec 7, 2013 · In this study, the cause of boundary deficiency in the SPH simulation for frictional contact is analysed. ... Finally, numerical tests are carried ...
[52]
[PDF] Momentum conserving methods that reduce particle clustering in SPH
Jan 1, 2014 · In this paper we present two remedies for particle clustering in SPH. Since particle clustering is the consequence of a diminishing kernel.
[53]
Improving smoothed particle hydrodynamics with an integral ...
We develop and test a fully conservative SPH scheme based on a tensor formulation that can be applied to simulate astrophysical systems.<|separator|>
[54]
A generalized density dissipation for weakly compressible smoothed ...
Aug 16, 2024 · Sun et al.27,28 incorporated particle shifting technique (PST) into the δ-SPH and proposed δ-plus-SPH, which demonstrated very good stability ...
[55]
Hybrid SPH-FEM solver for metal cutting simulations on the GPU ...
This paper presents a hybrid Smoothed Particle Hydrodynamics (SPH) – Finite Element Method (FEM) solver for metal cutting simulations on Graphics Processing ...
[56]
https://pure.tue.nl/ws/files/3858217/376670351851652.pdf
[57]
Entropy-Stable Smooth Particle Hydrodynamics for Thermo-Elasticity
This paper presents a novel Smooth Particle Hydrodynamics computational framework for the simulation of large strain fast solid dynamics in thermo-elasticity.
[58]
GPUSPH
GPUSPH was the first implementation of SPH to run entirely on GPU with CUDA and aims to provide a basis for cutting edge SPH simulations.Missing: PySPH LAMMPS
[59]
[PDF] The implementation of Smooth Particle Hydrodynamics in LAMMPS ...
Jul 17, 2011 · This document describes the implementation of the Smooth Particle Hydrodynamics (SPH) method within the Large-scale Atomic/Molecular ...Missing: PySPH | Show results with:PySPH