Alignment and orthogonality in anisotropic metric-based mesh adaptation

Cartesian or structured grids are usually preferred in the numerical simulation community as they have many interesting properties: natural high-order solutions, fast numerical solvers and high robustness. However, when complex geometries are used, robust Cartesian grids are less straightforward to implement, many research exist on the subject in order to intersect a complex geometry with a Cartesian grid.1 As far as mesh adaptation is concerned, reaching a high level of anisotropy remains a challenge as the directions are always aligned with the natural axes. On a di↵erent perspective, unstructured mesh generation is now a mature field of research where many methods are able to generate a 3D volume mesh with respect to a prescribed surface mesh. The most common methods are based on the Delaunay-kernel with boundary recovery,6,12,14 on a frontal approach24,35 or on a combination of both with local reconnections.33 All these techniques have been extended in the framework of anisotropic unstructured mesh adaptation. For robustness issues, most of the strategies rely on local mesh modification operators.8,10,11,20,22,37,40 If these strategies produce highly anisotropic tetrahedra, they fail to naturally produce elements that are aligned with the eigenvectors of the provided metric field. Unfortunately, element shape and alignment with the metric field also impacts accuracy and e ciency of the solution process. One well-known CFD example is the mesh generation of the boundary layer (BL) regions where physical variable gradients vary by orders of magnitude in di↵erent directions. In BL regions the characteristics of a mesh generated using anisotropic unstructured mesh adaptation are not always ideal. Viscous BL regions near a surface often have very stringent numerical requirements as they involve high-gradient and non-linear physics that usually includes turbulence. These regions are known a priori and ideally suited to a pseudostructured approach that generates elements with an advancing layer/normal type process.19,23,30,33,38 The resulting mesh is highly aligned, precisely spaced and very structured in at least the normal direction. The characteristics of a pseudo-structured type mesh are ideal for the BL regions. However, such methods are not compatible with adaptive approaches eliminating the adaptation of the boundary layer region and they usually lead to poor quality mesh near geometry singularities or in the transition between the BL region and the overall field, even if strategies have been proposed to solve this issues.4,5, 30,34 Consequently, the quest of ⇤Researcher, adrien.loseille@inria.fr †Professor, marcum@cavs.msstate.edu ‡Senior Researcher, frederic.alauzet@inria.fr

[1]  Frédéric Alauzet,et al.  High-order sonic boom modeling based on adaptive methods , 2010, J. Comput. Phys..

[2]  F. Hermeline,et al.  Triangulation automatique d'un polyèdre en dimension $N$ , 1982 .

[3]  Timothy J. Baker,et al.  Three dimensional mesh generation by triangulation of arbitrary point sets , 1987 .

[4]  Frédéric Alauzet,et al.  Continuous Mesh Framework Part I: Well-Posed Continuous Interpolation Error , 2011, SIAM J. Numer. Anal..

[5]  Shahyar Pirzadeh,et al.  Viscous unstructured three-dimensional grids by the advancing-layers method , 1994 .

[6]  Mahkame Sharbatdar,et al.  Anisotropic mesh adaptation : recovering quasi-structured meshes , 2013 .

[7]  C.R.E. de Oliveira,et al.  Tetrahedral mesh optimisation and adaptivity for steady-state and transient finite element calculations , 2001 .

[8]  Paul-Louis George,et al.  3D transient fixed point mesh adaptation for time-dependent problems: Application to CFD simulations , 2007, J. Comput. Phys..

[9]  Rainald Löhner Generation of Unstructured Grids Suitable for Rans Calculations , 1999 .

[10]  Eric J. Nielsen,et al.  Validation of 3D Adjoint Based Error Estimation and Mesh Adaptation for Sonic Boom Prediction , 2006 .

[11]  Paul-Louis George,et al.  Fully automatic mesh generator for 3D domains of any shape , 1990, IMPACT Comput. Sci. Eng..

[12]  R. Löhner,et al.  Generation of viscous grids at ridges and corners , 2009 .

[13]  Frédéric Alauzet,et al.  Optimal 3D Highly Anisotropic Mesh Adaptation Based on the Continuous Mesh Framework , 2009, IMR.

[14]  Frédéric Alauzet,et al.  Unstructured Mesh Generation Using Advancing Layers and Metric-Based Transition for Viscous Flowfields , 2013 .

[15]  Adrien Loseille,et al.  On 3D Anisotropic Local Remeshing for Surface, Volume and Boundary Layers , 2009, IMR.

[16]  Adrien Loseille,et al.  Serial and Parallel Mesh Modification Through a Unique Cavity-Based Primitive , 2013, IMR.

[17]  Paul-Louis George,et al.  ASPECTS OF 2-D DELAUNAY MESH GENERATION , 1997 .

[18]  Thierry Coupez,et al.  Génération de maillage et adaptation de maillage par optimisation locale , 2000 .

[19]  Frédéric Alauzet,et al.  Aligned Metric-based Anisotropic Solution Adaptive Mesh Generation , 2014 .

[20]  Frédéric Alauzet,et al.  High Order Sonic Boom Modeling by Adaptive Methods , 2009 .

[21]  C. Dobrzynski,et al.  Anisotropic Delaunay Mesh Adaptation for Unsteady Simulations , 2008, IMR.

[22]  D. Mavriplis An advancing front Delaunay triangulation algorithm designed for robustness , 1993 .

[23]  Paul-Louis George,et al.  Creation of internal points in Voronoi's type method. Control adaptation , 1991 .

[24]  Paul-Louis George,et al.  Construction d'un maillage 3-D anisotrope localement structuré , 2003 .

[25]  D. Ait-Ali-Yahia,et al.  Anisotropic mesh adaptation for 3D flows on structured and unstructured grids , 2000 .

[26]  Todd R. Michal,et al.  Anisotropic Mesh Adaptation Through Edge Primitive Operations , 2012 .

[27]  Adrien Loseille,et al.  Boundary Layer Mesh Generation and Adaptivity , 2011 .

[28]  Frédéric Alauzet,et al.  An L∞―Lp mesh-adaptive method for computing unsteady bi-fluid flows , 2010 .

[29]  Paresh Parikh,et al.  Generation of three-dimensional unstructured grids by the advancing-front method , 1988 .

[30]  D. F. Watson Computing the n-Dimensional Delaunay Tesselation with Application to Voronoi Polytopes , 1981, Comput. J..

[31]  Kazuhiro Nakahashi,et al.  Unstructured Mesh Generation For Viscous Flow Computations , 2002, IMR.

[32]  Adrien Loseille,et al.  Anisotropic Adaptive Simulations in Aerodynamics , 2010 .

[33]  Michael J. Aftosmis,et al.  Adaptive Cartesian mesh generation , 1998 .

[34]  Mark S. Shephard,et al.  3D anisotropic mesh adaptation by mesh modification , 2005 .

[35]  Adrien Loseille,et al.  Metric-orthogonal Anisotropic Mesh Generation☆ , 2014 .

[36]  David L. Marcum,et al.  Unstructured Grid Generation Using Automatic Point Insertion and Local Reconnection , 2002 .

[37]  N. Weatherill,et al.  Unstructured grid generation using iterative point insertion and local reconnection , 1995 .

[38]  Rainald Löhner,et al.  Three-dimensional grid generation by the advancing front method , 1988 .

[39]  Adrian Bowyer,et al.  Computing Dirichlet Tessellations , 1981, Comput. J..

[40]  Carlo L. Bottasso,et al.  Anisotropic mesh adaption by metric‐driven optimization , 2004 .