SPH-flow: Simulate what others can’t
The SPH-flow solver uses the Smoothed Particle Hydrodynamics (SPH) method. This method looks at computational fluid dynamics in a new light. Resulting from years of intense research, SPH-flow reveals new abilities in simulating previously unreachable complex problems.
SPH, an innovative approach
Among these approaches, an innovative method is the Smoothed Particle Hydrodynamics (SPH). With SPH, the long days of tedious meshing operations are gone. The Navier-Stokes equations are no longer discretized on coincident cells but on a set of particles moving along with the fluid.
After almost two decades of intense research and development with our academic and industrial partners, the SPH-flow solver is offering new opportunities by efficiently simulating complex problems never addressed before.
No tedious meshing operations for faster and easier simulation setup
High-quality results from high-level formulations
Local particle refinement for focused simulations
Lagrangian formulation for advection-related physical phenomena
Accurate free-surface tracking
Wide variety of accounted physics
From imposed rigid body motion to coupled FSI simulations
Strongly scalable MPI algorithms for HPC
No tedious meshing operations for faster and easier simulation setupThe meshing procedure in most conventional solvers – especially structured ones – is both delicate and time-consuming. It often represents a significant portion of engineering time. Conversely, the SPH-flow solver does not require any user meshing operation. Its particle generator automatically populates the simulation domain. More information about this feature on page Research Fast and accurate SPH modelling of 3D complex wall boundaries in viscous and non-viscous flows
High-quality results from high-level formulationsSince Monaghan’s initial formula, the SPH method for fluid mechanics has been greatly investigated. After two decades of R&D, the SPH-flow solver not only includes state-of-the-art models, it is being used also as research support by an active scientific community. A cherry-picked combination of advanced models is nowadays behind clear and predictive SPH-flow simulations. Among others, one can emphasize the high-order convective flux computation based on Riemann solvers, the particle rearrangement method using consistent ALE-based shifting and the so-called NFM boundary formulation which remains valid on complex geometry. More information about this feature on page Research SPH accuracy improvement through the combination of a quasi-Lagrangian shifting transport velocity and consistent ALE formalisms
Local particle refinement for focused simulationsWhen users want to focus on identified areas of interest, basic simulations with uniform spatial resolution are not appropriate. Rather, local particle refinement techniques, which involve multiple-sized particles, could be quite helpful for SPH-based simulations. Although the formalism of such methods is relatively easy to implement, the robustness at the coarse/fine interfaces can be sensitive. The SPH-flow solver implements an approach which ensures robustness with alleviated constraints. The robustness and accuracy achieved by locally refined simulations reach the ones of fully refined references, while the CPU cost is drastically lowered. More information about this feature on page Research Analysis and improvements of Adaptive Particle Refinement (APR) through CPU time, accuracy and robustness considerations
Lagrangian formulation for advection-related physical phenomenaThe SPH method is based on a Lagrangian formulation: the flow is described by means of discrete particles which move with the fluid. Under such formulation, the advection term of the Navier-Stokes equations vanishes. This constitutes a substantial benefit, when one considers the challenge that accurate and robust computation poses for Eulerian-based solvers. The SPH-flow solver shows full advantage of this strength when simulating flows driven by advection-related physical phenomena.
Accurate free-surface trackingMany design problems involve fluids whose domains are not known in advance. Those flows may deal with surface deformation, such as waves, or fragmentation and coalescence, such as droplet formation and merging. Thanks to its Lagrangian nature, the SPH-flow solver implicitly tracks the free surfaces. No approximate, diffusing nor CPU-consuming numerical computation is needed to accurately locate the interfaces.
Multi-fluid capacitySome applications rely on physical phenomena involving different fluids. Within the SPH framework, this translates into describing the flow using particles with various physical characteristics (density, viscosity…). The physical properties at and across multi-fluid interfaces, such as velocity continuity and viscous strain, are imposed by suitable particle-to-particle interactions. Like free-surface interfaces, SPH-flow takes advantage of its Lagrangian nature to implicitly and accurately track multi-fluid interfaces.
Wide variety of accounted physicsThe SPH-flow solver can take into account many physical models: viscosity, compressibility, thermal, surface tension, contact angle, non-Newtonian fluids… Different implementations have been used for most of them. From this complete set of numerical alternatives, the SPH-flow solver exhibits a selection of effective, application-specific guidelines.
From imposed rigid body motion to coupled FSI simulationsStructures in motion can affect fluid flows as massively as fluid flows can stress solid structures. Simulating these Fluid–Structure Interactions (FSI) requires not only two solvers – one for fluids and one for structures – but also a coupling strategy: i.e. a protocol to allow those solvers to exchange relevant dynamic information. A proven, precise and robust strategy allows the efficient coupling of the SPH-flow solver with diverse Finite Element Method (FEM) codes including EDF Code_Aster and 3DS SIMULIA Abaqus. More information about this feature on page Research An efficient FSI coupling strategy between Smoothed Particle Hydrodynamics and Finite Element methods
Waves generatorWaves must often be considered when carrying out studies in marine environments. To achieve optimal accuracy, SPH-flow uses a coupling with the ECN/LHEEA HOS-Ocean solver (based on the High-Order Spectral method) to generate and propagate the wave.
Strongly scalable MPI algorithms for HPCBecause millions of particles may be necessary to simulate most industrial applications, SPH-flow parallel computing performance has been optimized to an advanced level. Clear scalability has been proven on up to tens of thousands of processors. Such HPC results make it possible to simulate complex and realistic problems with relevant accuracy and at an affordable CPU cost. More information about this feature on page Research On distributed memory MPI-based parallelization of SPH codes in massive HPC context
- Navier-Stokes or Euler equations
- Weakly compressible and incompressible approaches
- Implicitly tracked free surface
- Viscosity models
- Surface tension models
- Thermic effects
- Multi-fluid capacity
- Various boundary conditions: no-slip/slip, periodicity, inlet/outlet…
- Rigid body with imposed or free motion
- Fluid-structure interaction (FSI)
- Aerodynamics forcing on fluid
- Particles-based method SPH
- Lagrangian or ALE approach
- Convective flux computation based on Riemann solvers
- ALE-based particles rearrangement (shifting)
- NFM and ghost boundary formulations
- Local particle refinement
- 3rd order explicit and semi-implicit time schemes
- Scalable parallel computing based on MPI protocol