Three-dimensional mesh adaptation in PFEM Preprint

Thomas Leyssens, Jonathan Lambrechts, Jean-François Remacle · 2025

Read Three-dimensional mesh adaptation in PFEM Preprint by Thomas Leyssens, Jonathan Lambrechts, Jean-François Remacle (2025) in the Tower of Babel, a free online library of public-domain works preserved in clean, searchable text. No sign-up. The full text is served as a plain-text file at a predictable URL, where text is present: https://antarctic-labs.com/tower-of-babel/arxiv-2512-20347.txt.

Introduction

Thomas Leyssens, Jonathan Lambrechts, and Jean-François Remacle improve the geometry handling of the Particle Finite Element Method. PFEM simulates free-surface flows, such as splashing liquids, by moving mesh nodes like particles and rebuilding the mesh constantly. In three dimensions, detecting the new domain shape and keeping elements healthy is hard.

Three contributions make the method practical. The domain is reconstructed from the advected boundary of the previous step instead of the classic alpha-shape procedure. Refinement proceeds in two stages, first on the boundary surface, then by quality-based insertion in the bulk. A boundary manager handles complex geometries that earlier versions could not. Demonstrations on chaotic free-surface problems show the approach holding accuracy while the fluid changes shape and topology.

Catalog record

CREATORThomas Leyssens, Jonathan Lambrechts, Jean-François Remacle
YEAR2025
COLLECTIONPAPERS
FORMATTXT
FILE SIZE47 KB
VERSION1.0
LICENSEhttp://creativecommons.org/licenses/by/4.0/
RIGHTSCC BY
SOURCEarXiv

Source: https://arxiv.org/abs/2512-20347

Browse the full library · About the Tower of Babel

Full text

The opening of the work, from the library's hosted text:

Three-dimensional mesh adaptation in PFEM Preprint Thomas Leyssens, Jonathan Lambrechts, Jean-François Remacle Abstract Chaotic free surface flows are challenging problems to simulate numerically, mainly due to the significant changes in geometry and frequent topological changes. Methods that track the evolution of the fluid in a Lagrangian formulation are a natural choice. One such method is the Particle Finite Element Method (PFEM). As a hybrid particle-based and mesh-based method, PFEM leverages advantages from both approaches. The equations of motion are solved on a mesh using the finite element method and the obtained velocity field is used to displace the nodes of this mesh, considered as particles carrying all the relevant information across time steps. To avoid element distortion, the mesh is frequently re-generated. This introduces some challenges: How can the new shape of the domain be detected? How can the quality of the elements be kept acceptable? Can adaptive mesh refinement increase the accuracy and efficiency of the solver? Can PFEM simulations be performed in the presence of complex boundary geometries? In this work, three contributions to the geometry and mesh component of PFEM are introduced for three-dimensional free surface flow simulations. First, we propose a different domain reconstruction approach than the classically used α \alpha -shape procedure, namely through the use of the advected boundary from the previous time step as a predicate to represent the new shape of the domain. Second, an adaptive refinement procedure is proposed in two steps: refinement of the boundary surface followed by quality-based node insertion in the bulk. Third, an approach for managing boundaries in complex geometries is presented. A series of applications is shown to demonstrate the interest of the approach. 1 Introduction The particle finite element method, or PFEM, is an interesting approach for free surface flow simulations. Defined through a Lagrangian finite element framework coupled to an automatic surface tracking algorithm, the method allows to simulate strong topological changes of the simulated domain. These features make it a method of choice for simulations such as fluid-structure interactions [ 3 ] , landslide modelling [ 11 ] , and melt-pool dynamics [ 10 ] . In PFEM, the domain evolution is described by Lagrangian particles, while the governing equations are discretized using the finite element method, benefiting from its mature and well-understood mathematical foundations. In contrast to methods such as the material point method (MPM) [ 6 ] or the particle-in-cell method [ 2 ] , which rely on a fixed background mesh over which particles are advected, PFEM operates on a mesh constructed from the Lagrangian particles themselves. The particles become the nodes of the finite element mesh, and the equations of motion are solved on this mesh using a classical piecewise linear finite element discretization. The positions of the particles are subsequently updated by integrating the computed velocity field in time. Keeping the same set of particles throughout the simulation is not feasible, since even in incompressible flows, a shearing motion will progressively transform an initially well-organized particle distribution into a highly distorted and unusable configuration. As a result, the particle distribution must be updated whenever necessary. In the present work, this update is performed at every time step. Instead of operating directly on the particle cloud, PFEM performs the particle update by acting on the mesh obtained from the Delaunay triangulation of the particles, which is considerably more robust. Repeated mesh regeneration endows PFEM with considerable flexibility, particularly through its ability to naturally accommodate topological changes of the fluid domain, but it also represents a major computational challenge. In practice, the computational cost of triangulating a point cloud, even with several million points, is negligible compared to that of solving the incompressible Navier–Stokes equations on the resulting mesh [ 27 ] . The difficulty lies instead in the fact that a Delaunay triangulation fills the convex hull of the point cloud, and this convex hull provides a particularly poor oracle for determining the actual shape of a fluid domain represented by a set of particles. Moreover, as discussed above, the mesh must subsequently be adapted in order to restore a well-balanced and harmonious particle distribution. Finally, the set of particles defining the fluid domain evolves within a bounded region, typically delimited by rigid walls. When a particle reaches such a rigid boundary, an appropriate boundary condition must be enforced. It therefore becomes clear that a robust PFEM formulation must address three key geometrical challenges: 1. an accurate and efficient oracle to determine the shape of the fluid domain, 2. a mesh adaptation strategy to update the particle distribution, 3. a general and flexible representation of the solid boundaries confining the fluid motion. The objective of this work is to present three innovative approaches for tackling these challenges. First, we propose a new oracle to represent the fluid domain after re-meshing. The classical \alpha -shape approach has many known limitations, resulting in mass errors [ 12 ] . While retaining the algorithmic efficiency and robustness of the Delaunay triangulation, we use the advected boundary from the previous time step as a predicate to define the updated shape of the domain. This boundary representation, in the form of a deformed triangulation, is not necessarily watertight; nevertheless, it can be used as a robust oracle to answer the fundamental question of whether a point \boldsymbol{x} lies inside or outside the fluid domain. To this end, we draw inspiration from recent work on fast winding numbers [ 1 ] . Second, we present a 3D adaptive mesh refinement technique for PFEM. Using this technique, the quality of the elements remains sufficiently high at all times, and local mesh adaptation allows to concentrate the degrees of freedom of the simulation in the regions of interest, for example, along the free surface. The approach consists of first refining the bounding surface elements, followed by refinement of the internal volume elements. This mesh adaptation procedure is based on a size field, allowing the user have control over the mesh. Third, we propose a robust treatment of solid boundaries that fully decouples the representation of the fluid domain from that of the solid geometry. Solid walls are neither meshed nor topologically connected to the fluid domain. Instead, all interactions with the solid boundaries rely exclusively on two geometric queries. The first query is a ray–solid intersection operator. Given a point \boldsymbol{x} and a direction \boldsymbol{d} , it returns the first intersection of the ray (\boldsymbol{x},\boldsymbol{d}) with the solid boundary, if such an intersection exists. This operator is used to identify fluid faces that are in contact with solid walls. The second query is an orthogonal projection operator, which maps a point \boldsymbol{x} onto its closest point on the solid boundary. This operator is used to enforce boundary conditions and to reposition particles near walls. Because the solid geometry is accessed only through these two queries, it may be represented in a wide variety of forms, including surface meshes, triangle soups, level-set representations, or distance fields. This makes the approach robust with respect to imperfect or non-watertight solid geometries and significantly simplifies the handling of fluid–solid interactions. In the following sections, section 2 first presents the physical context. The core of the paper then follows in section 3

Read the complete text