Parallel simulation and adaptive mesh refinement for 3D elastostatic contact mechanics problems between deformable bodies
EPALLE Alexandre 1,2 , RAMIÈRE Isabelle 1 , LATU Guillaume 1 , LEBON Frédéric 2 · 2025
Read Parallel simulation and adaptive mesh refinement for 3D elastostatic contact mechanics problems between deformable bodies by EPALLE Alexandre 1,2 , RAMIÈRE Isabelle 1 , LATU Guillaume 1 , LEBON Frédéric 2 (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-2511-20142.txt.
Introduction
Epalle, Ramière, Latu, and Lebon scale contact mechanics simulations across a thousand cores. Their 2025 paper combines adaptive mesh refinement with parallel computing. Contact between deformable bodies is stubbornly nonlinear and non-smooth. Fine meshes are needed only near the contact zone. Refining everywhere wastes enormous computing power.
Their algorithm pairs contact nodes directly, node to node. A penalty method enforces the contact constraints. Non-conforming h-refinement sharpens quadrilateral and hexahedral meshes locally. An estimate-mark-refine loop decides where refinement pays. Paired contact nodes are kept on the same MPI process. That placement slashes communication between parallel tasks. Super-parametric elements preserve curved surfaces during refinement. Even coarse starting meshes detect contact zones well. Hertzian contact problems in 2D and 3D validated accuracy. Strong scaling held up to 1,024 cores in tests. Problems with tens of millions of unknowns become feasible.
Catalog record
| CREATOR | EPALLE Alexandre 1,2 , RAMIÈRE Isabelle 1 , LATU Guillaume 1 , LEBON Frédéric 2 |
|---|---|
| YEAR | 2025 |
| COLLECTION | PAPERS |
| FORMAT | TXT |
| FILE SIZE | 96 KB |
| VERSION | 1.0 |
| LICENSE | http://creativecommons.org/licenses/by/4.0/ |
| RIGHTS | CC BY |
| SOURCE | arXiv |
Source: https://arxiv.org/abs/2511-20142
Browse the full library · About the Tower of Babel
Full text
The opening of the work, from the library's hosted text:
Parallel simulation and adaptive mesh refinement for 3D elastostatic contact mechanics problems between deformable bodies EPALLE Alexandre 1,2 , RAMIÈRE Isabelle 1 , LATU Guillaume 1 , LEBON Frédéric 2 Abstract Parallel implementation of numerical adaptive mesh refinement (AMR) strategies for solving 3D elastostatic contact mechanics problems is an essential step toward complex simulations that exceed current performance levels. This paper introduces a scalable, robust, and efficient algorithm to deal with 2D and 3D elastostatics contact problems between deformable bodies in a finite element framework. The proposed solution combines a treatment of the contact problem by a node-to-node pairing algorithm with a penalization technique and a non-conforming h-adaptive refinement of quadrilateral/hexahedral meshes based on an estimate-mark-refine approach in a parallel framework. One of the special features of our parallel strategy is that contact paired nodes are hosted by the same MPI tasks, which reduces the number of exchanges between processes for building the contact operator. The mesh partitioning introduced in this paper respects this rule and is based on an equidistribution of elements over processes, without any other constraints. In order to preserve the domain curvature while hierarchical mesh refinement, super-parametric elements are used. This functionality enables the contact zone to be well detected during the AMR process, even for an initial coarse mesh and low-order discretization schemes. The efficiency of our contact-AMR-HPC strategy is assessed on 2D and 3D Hertzian contact problems. Different AMR detection criteria are considered. Various convergence analyses are conducted. Parallel performances up to 1024 cores are illustrated. Furthermore, memory footprint and preconditionners performance are analyzed. † † highlights: Combinaison of adaptive mesh refinement and contact in a parallel finite element framework Non-conforming h-adaptive refinement on quadrilateral and hexahedral meshes Super-parametric elements to deal with curved geometries during the AMR process Mesh partitioning ensuring contact paired nodes on the same MPI tasks Very satisfactory strong scalability up to 1024 cores. 1 Introduction Efficient numerical solving of contact mechanics problems remains challenging. These problems are locally highly non-linear and non-smooth. They often have the particularity of concentrating very high stress in the contact area. Thus, the numerical simulation of this type of problems is computationally demanding, especially if a local accurate solution is expected. Dealing with contact between deformable solids, the finite element method (FEM) [ 27 ] remains the most widely used numerical method [ 94 ] for solving the associated problem, although others approaches have been more recently introduced. Let us cite for example the virtual element method (VEM) [ 31 , 90 ] and its extension to contact [ 91 , 65 ] as well as the Hybrid High Order (HHO) [ 34 , 35 ] and its adaptation to (frictional) contact problems [ 26 ] . The scaled boundary finite element method (SBFEM) [ 83 ] , combining the advantages of the FEM and of the boundary element method (BEM), has also been recently applied for contact analysis in [ 54 ] . Finally, let us mention the work on isogeometric contact [ 32 ] which considers contact formulations within the framework of isogeometric analysis (IGA) [ 29 ] . Several numerical strategies have been developed over the years to deal with constraints at contact interfaces [ 61 , 66 , 94 ] . When using the FEM, the penalization method [ 62 ] , the Lagrange multiplier method [ 61 ] , the augmented Lagrangian method [ 82 ] , the Nitsche method [ 93 , 24 ] or mortar-based formulation [ 14 ] are commonly employed to design robust and efficient algorithms. For more details about these strategies, the reader is referred to [ 94 ] . Furthermore, solving contact problems with accuracy requires a high degree of mesh refinement in the contact zones. Building uniformly refined mesh for such problems is still prohibitively computationally expensive, especially for 3D geometries. One way to overcome this limit is to rely on adaptive mesh refinement (AMR) methods [ 87 ] . These techniques, which have been well known for many years, enable the mesh to be locally adjusted to a desired accuracy, thereby reducing the total number of degrees of freedom compared to uniformly refined meshes. Popular adaptive techniques for reducing discretization error include h-adaptive [ 33 , 36 ] (mesh-step refinement), p-adaptive [ 10 , 8 ] (order of basis function increase), hp-adaptive [ 99 , 8 ] (combination of both previous) and local multigrid [ 19 , 11 , 60 ] approaches (geometric refinement by adding local refined meshes). Adaptive mesh refinement is generally coupled with a posteriori error estimators [ 87 ] in order to automatically determine the elements to be refined, and sometimes also to automatically stop the refinement process [ 98 , 67 ] . Many works have focused on the application of AMR techniques in the context of contact mechanics problems, mainly in the FEM framework. Let us cite for example [ 92 , 18 , 47 , 6 ] whose authors proposed dedicated a posteriori error estimators for contact problems that have been applied with conforming h-adaptive refinement. In particular, [ 92 ] modifies the Zienkiewicz and Zhu (ZZ) estimator [ 98 ] to take into account contact contributions. In their turn, the works of [ 80 , 77 ] focus on mesh adaptivity with non-conforming h-refinement methods. Adaptation of local multigrid methods for contact problems have been studied in [ 16 , 67 ] . Let us mention that the authors of [ 67 ] successfully applied the standard ZZ estimator in each body separately as for multimaterial problems. Most of the above methods have only been applied on 2D geometries. It should also be noted that some of them only concern rigid-deformable contact problems (Signorini problems) [ 18 , 80 , 77 , 6 ] . For its part, the extension of SBFEM to contact problems [ 54 ] has been directly introduced in a non-conforming h-adaptive mesh framework based on a quadtree decomposition with a dedicated SBFEM error indicator [ 84 ] . Concerning the VEM, the authors in [ 2 ] provide a methodology to combine AMR and 2D contact problems. Two types of AMR techniques are studied on 2D contact problems: non-conforming h-adaptive meshes or conforming locally refined Voronoï cell meshes. The mesh is locally refined in a predefined contact zone until a geometric criterion on the refined contact area is reached. The extension to 3D contact problems of the VEM with AMR based on Voronoï type elements is proposed in [ 28 ] . However, for highly singular problems such as those arising from contact mechanics, the discretization can become very fine locally, and the number of unknowns prohibitive even with AMR. It is therefore also desirable to use high-performance computing (HPC) methods for faster processing and for handling large-scale problems. The fast solution of partial differential equations with the FEM on adaptively refined meshes via parallelization strategies has already been widely explored, see [ 42 , 86 , 71 , 85 , 20 , 21 ] to cite but a few. A large set of these strategies lies on refinement-tree based partitioning methods and hence non-conforming h-adaptive approaches (with hanging nodes). In addition, it has been recently presented in