cubit mesh generation project technical history of hex mesh generation scott a. mitchell sandia...
TRANSCRIPT
CUBIT Mesh Generation Project
Technical History of Hex Mesh Generation
Technical History of Hex Mesh Generation
Scott A. Mitchell
Sandia National Labs
Sandia is a multiprogram laboratory operated by Sandia Corporation, a Lockheed Martin Company,for the United States Department of Energy under contract DE-AC04-94AL85000.
CUBIT Mesh Generation Project
Technical History of Hex Mesh Generation
Outline
Introduction Relationship Map
» Sweep-like techniques» 3d techniques» Synthesis
Where are we now?» Hopes
CUBIT Mesh Generation Project
Technical History of Hex Mesh Generation
Introduction
CUBIT Mesh Generation Project
Technical History of Hex Mesh Generation
Introduction
What I will cover What I won’t cover Why is hex meshing hard
CUBIT Mesh Generation Project
Technical History of Hex Mesh Generation
What I will cover
Asked to give technical talk Unstructed hex meshing algorithms and ideas
» Relationships between algorithms– Technical threads from one technique to another
Focus on assembly meshing problems» Multiple interlocking parts
Focus on connectivity problem» 3d Algorithms» A little sweeping
CUBIT Mesh Generation Project
Technical History of Hex Mesh Generation
What I won’t cover
Tools, packaging, ease of use… Nodal positioning for good quality
» Bias and sizing control» Except as approach to connectivity
problem I’ll get some names right, but not all,
and leave some people out. Sorry, no offense intended.
CUBIT Mesh Generation Project
Technical History of Hex Mesh Generation
What I won’t cover
Techniques involving» Mixed elements
– Hex-dominant methods
» Non-conforming meshes– Overset grids
» Primitives and Special algorithms for restricted classes of geometries
– Sphere-like assemblies (Hex3d at LANL…)– Air around airplanes, cars,…
CFD block primitives
Medial-axis and Midpoint Subdivision
CUBIT Mesh Generation Project
Technical History of Hex Mesh Generation
Why is Hex Meshing Hard?
Connectivity of tet meshing is “easy”» Delaunay triangulations
– Given set of points, can connect them up to form triangles, tetrahedra, d-simplices
– Empty-sphere property shows you how– Caveats
Good quality is difficult in 3 dimensions Boundary constraints is difficult in 3 dimensions
» Octree (subdivision) techniques– Keep dividing polyhedron until you get simplices
No new points on the boundary are necessary
» Advancing front techniques– Remaining space always dividable into tetrahedra
CUBIT Mesh Generation Project
Technical History of Hex Mesh Generation
Why is Hex Meshing Hard?
Quads harder than tris (i.e. even 2d is harder)» Can’t just divide polygons into quads
– New points on the boundary may be necessary
» In practice, an even number of boundary edges can be divided into quads (with new interior points).
– Necessary and sufficient
? ?
q q
bdyeeq int24Euler:
CUBIT Mesh Generation Project
Technical History of Hex Mesh Generation
3D Approaches
Assume: assembly problem» Fixed boundary mesh
Geometry first» constrained by topology
Topology first» unconstrained
– fix-up afterwards
» constrained by geometry
CUBIT Mesh Generation Project
Technical History of Hex Mesh Generation
Relationship Map
Will refer to these map for context throughout talk
CUBIT Mesh Generation Project
Technical History of Hex Mesh Generation
Sweep Relationship Map
Sweeping Node Placement Strategies
2D2.
5D2.
75D
Good Source SurfaceMesh Strategies
1-1 Sweeping / Projection / Rotation (lots)
M-1 Sweeping (several)
M-M Sweeping
Multi-axisCooper Tool Grafting
Cooper Tool
Inside-out3d
CUBIT Mesh Generation Project
Technical History of Hex Mesh Generation
3D Relationship MapPaving/Plastering
Whisker Weavingstate list
STC dual structurerediscovered
Whisker Weavingcurve contraction
STC folds
Existence Proofs
Dual cycleElimination& Shelling
Recursive bisection
Linear-size
Local reconnection and smoothing
Flipping, (planar faces)
Prepare the boundary mesh:splicing
Refinement andcoarsening
Schneider’s open problem
Geode
Hex-to-void
clea
n-up
Geode betweenshells
geometry first topology firstTHEXcons
truc
tion
disc
over
y
Sweep-liketechniques
THEX
samitch: be careful to imply conceptual relationship, not cause and effect of efforts or people
samitch: be careful to imply conceptual relationship, not cause and effect of efforts or people
? Hex dominant?
Dual-ParticleMethods
Inside-outtechniques
Slicing doughnuts
CUBIT Mesh Generation Project
Technical History of Hex Mesh Generation
3D Relationship MapPaving/Plastering
Whisker Weavingstate list
STC dual structurerediscovered
Whisker Weavingcurve contraction
STC folds
Existence Proofs
Dual cycleElimination& Shelling
Recursive bisection
Linear-size
Local reconnection and smoothing
Flipping, (planar faces)
Prepare the boundary mesh:splicing
Refinement andcoarsening
Schneider’s open problem
Geode
Hex-to-void
clea
n-up
Geode betweenshells
geometry first topology firstTHEXcons
truc
tion
disc
over
y
Sweep-liketechniques
THEX
? Hex dominant?
Dual-ParticleMethods
Inside-outtechniques
Slicing doughnuts
CUBIT Mesh Generation Project
Technical History of Hex Mesh Generation
Paving/Plastering
~12 years ago Blacker, Stephenson, Cass, Benzley Paving
» 2d advancing front– Advance based on geometric criteria– Merge connectivity when fronts collide
Theme: geometry first, connectivity second
CUBIT Mesh Generation Project
Technical History of Hex Mesh Generation
Paving
Paving•Advancing Front: Begins with front at boundary•Forms rows or single of elements based on front angles•Must have even number of intervals for all-quad mesh
(Blacker,92)(Cass,96)
CUBIT Mesh Generation Project
Technical History of Hex Mesh Generation
Paving
Paving•Advancing Front: Begins with front at boundary•Forms rows of elements based on front angles•Must have even number of intervals for all-quad mesh
(Blacker,92)(Cass,96)
CUBIT Mesh Generation Project
Technical History of Hex Mesh Generation
Paving
Paving•Advancing Front: Begins with front at boundary•Forms rows of elements based on front angles•Must have even number of intervals for all-quad mesh
(Blacker,92)(Cass,96)
Form new row and check for
overlap
CUBIT Mesh Generation Project
Technical History of Hex Mesh Generation
Paving
Paving•Advancing Front: Begins with front at boundary•Forms rows of elements based on front angles•Must have even number of intervals for all-quad mesh
(Blacker,92)(Cass,96)
Insert “Wedge
”
CUBIT Mesh Generation Project
Technical History of Hex Mesh Generation
Paving
Paving•Advancing Front: Begins with front at boundary•Forms rows of elements based on front angles•Must have even number of intervals for all-quad mesh
(Blacker,92)(Cass,96)
Seams
CUBIT Mesh Generation Project
Technical History of Hex Mesh Generation
Paving
Paving•Advancing Front: Begins with front at boundary•Forms rows of elements based on front angles•Must have even number of intervals for all-quad mesh
(Blacker,92)(Cass,96)
Close Loops and smooth
CUBIT Mesh Generation Project
Technical History of Hex Mesh Generation
Paving
Interesting / relevant details for later– Single loop: if seaming, keep two resulting loops even
– Multiple loops? Finesse by having each loop even at all stage
?
?
CUBIT Mesh Generation Project
Technical History of Hex Mesh Generation
Plastering
~10 years ago, Blacker, Canann, Stephenson Paving worked so well, why not try paving in
3d? = Plastering» Start with fixed surface mesh» Add proto-hexes (faces) one by one, conforming
to bdy» Seam together nearby points, edges, faces
– Build hexes
CUBIT Mesh Generation Project
Technical History of Hex Mesh Generation
Plastering
Kind of worked» Could fill up most of space
CUBIT Mesh Generation Project
Technical History of Hex Mesh Generation
Plastering
Difficulties» Advancing fronts rarely seemed to close
– Seams and Wedges (knives) Blacker et al.
– Ted Blacker to Pete Murdoch Find some way of closing up remaining region
Drive or collapse later
Remove from front for now
CUBIT Mesh Generation Project
Technical History of Hex Mesh Generation
3D Relationship MapPaving/Plastering
Whisker Weavingstate list
STC dual structurerediscovered
Whisker Weavingcurve contraction
STC folds
Existence Proofs
Dual cycleElimination& Shelling
Recursive bisection
Linear-size
Local reconnection and smoothing
Flipping, (planar faces)
Prepare the boundary mesh:splicing
Refinement andcoarsening
Schneider’s open problem
Geode
Hex-to-void
clea
n-up
Geode betweenshells
geometry first topology firstTHEXcons
truc
tion
disc
over
y
Sweep-liketechniques
THEX
? Hex dominant?
Dual-ParticleMethods
Inside-outtechniques
Slicing doughnuts
CUBIT Mesh Generation Project
Technical History of Hex Mesh Generation
Plastering as hex dominant
Years later» Knives are ok finite elements, as are pyramids
– Leave knives in? – vaporware
» Hex-tet interface, fill remaining void with pyramids and tets– Geode transition template– Developed but not used
We’ll leave this thread and come back to it later– Ted Blacker to Pete Murdoch
(Meyers, Stephenson, Benzley) Find some way of closing up remaining region
CUBIT Mesh Generation Project
Technical History of Hex Mesh Generation
3D Relationship MapPaving/Plastering
Whisker Weavingstate list
STC dual structurerediscovered
Whisker Weavingcurve contraction
STC folds
Existence Proofs
Dual cycleElimination& Shelling
Recursive bisection
Linear-size
Local reconnection and smoothing
Flipping, (planar faces)
Prepare the boundary mesh:splicing
Refinement andcoarsening
Schneider’s open problem
Geode
Hex-to-void
clea
n-up
Geode betweenshells
geometry first topology firstTHEXcons
truc
tion
disc
over
y
Sweep-liketechniques
THEX
? Hex dominant?
Dual-ParticleMethods
Inside-outtechniques
Slicing doughnuts
CUBIT Mesh Generation Project
Technical History of Hex Mesh Generation
STC Dual Structure Rediscovered for Engr.
Dual of tri / tet mesh » is just the dual!
Dual of quad / hex mesh» is structure of arrangement of curves or surfaces» Murdoch, 16 September 1993
CUBIT Mesh Generation Project
Technical History of Hex Mesh Generation
STC Dual Structure Rediscovered
Rediscovered a couple of times» Physicists: divide high-dimensional space into cubes to
study relativity– A.B. Zamalodchikov: Tetrahedron equations and the relativistic
S-matrix of straight strings in 2+1 dimensions, Comm. Math. Phys. 79, 489-505 (1981)
» Mathematics: arrangements– B. Grünbaum: Arrangements and Spreads, Reg. Conf. Ser. in
Math. no 10, Amer. Math Soc. (1972)
» Topology: geometric– I. R. Aitchison, J. H. Rubinstein: An introduction to polyhedral
metrics of nonpositive curvature on 3-manifolds, Geometry of manifolds, ed S Donaldson and C Thomas, Cambridge Univ. Press 1990
CUBIT Mesh Generation Project
Technical History of Hex Mesh Generation
Plastering’s downfall
» Seaming and connecting together based on geometry won’t create this nice structure.
CUBIT Mesh Generation Project
Technical History of Hex Mesh Generation
3D Relationship MapPaving/Plastering
Whisker Weavingstate list
STC dual structurerediscovered
Whisker Weavingcurve contraction
STC folds
Existence Proofs
Dual cycleElimination& Shelling
Recursive bisection
Linear-size
Local reconnection and smoothing
Flipping, (planar faces)
Prepare the boundary mesh:splicing
Refinement andcoarsening
Schneider’s open problem
Geode
Hex-to-void
clea
n-up
Geode betweenshells
geometry first topology firstTHEXcons
truc
tion
disc
over
y
Sweep-liketechniques
THEX
? Hex dominant?
Dual-ParticleMethods
Inside-outtechniques
Slicing doughnuts
CUBIT Mesh Generation Project
Technical History of Hex Mesh Generation
Murdoch’s thesis (1995)
“The Spatial Twist Continuum (S.T.C.) is a dual representation of the hexahedral mesh. The S.T.C. method of representing a mesh inherently captures the global connectivity constraint of all hexahedral meshes and thereby allows the construction of the mesh in a robust ordered or random manner”
Ideas» Geometric curvature and continuity of twist planes
» Envision mostly regular arrangement, with careful attention paid to few places of irregular connectivity, and how to combine them together
CUBIT Mesh Generation Project
Technical History of Hex Mesh Generation
Murdoch’s vaporware STC fold algorithm
“…a valid hexahedral mesh will have a valid S.T.C., the question is then whether to create the S.T.C. first or use it as a tool to guide the internal mesh connectivity”» Suggested creating surface and volume mesh simultaneously by
inserting geometric twist planes one by one, guided by desired mesh density functions
» Seemed hard– Could you do that for an assembly? Or just single volume?– How to define geometry of twist-planes?– How to ensure dual of surface mesh captured the topology of the
model? Compare to octree, inside out method by Schneiders (which hadn’t been done yet): ½ time spent on surface isomorphism!
CUBIT Mesh Generation Project
Technical History of Hex Mesh Generation
Murdoch’s vaporware STC fold algorithm
» To my knowledge, no serious investment in simultaneous surface and volume approach. They all create the surface mesh, then try to get the volume mesh
– If it worked for fixed surface mesh, it would solve the assembly problem– Refinement and coarsening, mesh improvement, flipping, sheet
pushing = after initial arrangement, mostly topology– Ostenson’s idea– Thuston’s minimal energy surface (soap bubble) idea to define
geometry– Get a good surface mesh – visit later
Shepherd’s splicing Mattias Muller-Hanneman WW surface mesh fixup WW add geode templates
» Surface mesh first, then volume, turned out to be hard, too!
CUBIT Mesh Generation Project
Technical History of Hex Mesh Generation
3D Relationship MapPaving/Plastering
Whisker Weavingstate list
STC dual structurerediscovered
Whisker Weavingcurve contraction
STC folds
Existence Proofs
Dual cycleElimination& Shelling
Recursive bisection
Linear-size
Local reconnection and smoothing
Flipping, (planar faces)
Prepare the boundary mesh:splicing
Refinement andcoarsening
Schneider’s open problem
Geode
Hex-to-void
clea
n-up
Geode betweenshells
geometry first topology firstTHEXcons
truc
tion
disc
over
y
Sweep-liketechniques
THEX
? Hex dominant?
Dual-ParticleMethods
Inside-outtechniques
Slicing doughnuts
CUBIT Mesh Generation Project
Technical History of Hex Mesh Generation
Whisker Weaving State List
Instead of Plastering hexes, » With so much structure capturing these
implicit constraints, why not start with the STC, then dualize to get hexes?
» Blacker, Tautges, Mitchell
CUBIT Mesh Generation Project
Technical History of Hex Mesh Generation
Whisker Weaving State List Ideas
Main ideas of state list Whisker Weaving » Start with fixed surface mesh
– This defines “boundary loops” of the sheets = twist-planes
green red blue purple
CUBIT Mesh Generation Project
Technical History of Hex Mesh Generation
Whisker Weaving State List Ideas
» Add dual of hex (STC vertex) from bdy inward, one at a time
– Compare to plastering…– Each sheet is 2d: did we reduce this 3d problem to a set
of 2d problems? I think not…although it is easier to understand and visualize than free-floating primal edges.
crossing = weaving
CUBIT Mesh Generation Project
Technical History of Hex Mesh Generation
Whisker Weaving State List Ideas
Now it’s different than Plastering» Try to get good local connectivity
– Connectivity Rules about where to add, when to seam, … Easily identify seaming = joining cases Rarely join two loops together
» Keep sheets as disk
– Hmm, knives=wedges sometimes show up at the end of chords where loops self-intersect…try tracing a knife
» Geometric rules: consider geometric dihedral angles of faces near surface mesh, no flat hexes
» Heroic effort by Tautges to build rules for good all-hexes and close the weave
– In practice, often unclosed region left-over, like Plastering– In practice, very complicated sheets
CUBIT Mesh Generation Project
Technical History of Hex Mesh Generation
WW complicated sheets
edge w/ 10 hexes
• One or two sheets going through most of the surface quads• Related to expected connectivity of random graphs on lattice?• Red = knives (wedges)
CUBIT Mesh Generation Project
Technical History of Hex Mesh Generation
3D Relationship MapPaving/Plastering
Whisker Weavingstate list
STC dual structurerediscovered
Whisker Weavingcurve contraction
STC folds
Existence Proofs
Dual cycleElimination& Shelling
Recursive bisection
Linear-size
Local reconnection and smoothing
Flipping, (planar faces)
Prepare the boundary mesh:splicing
Refinement andcoarsening
Schneider’s open problem
Geode
Hex-to-void
clea
n-up
Geode betweenshells
geometry first topology firstTHEXcons
truc
tion
disc
over
y
Sweep-liketechniques
THEX
? Hex dominant?
Dual-ParticleMethods
Inside-outtechniques
Slicing doughnuts
CUBIT Mesh Generation Project
Technical History of Hex Mesh Generation
Schneiders Open Problem
Can anyone create an all-hex mesh conforming to the following surface mesh?
http://www-users.informatik.rwth-aachen.de/~roberts/open.html
» He didn’t really say why you’d want to. But, if you could, then for any model, you could Plaster a while, add these, fill with tets, then divide into all-hex mesh.
» Are there “bad” surface meshes that aren’t fillable?
CUBIT Mesh Generation Project
Technical History of Hex Mesh Generation
3D Relationship MapPaving/Plastering
Whisker Weavingstate list
STC dual structurerediscovered
Whisker Weavingcurve contraction
STC folds
Existence Proofs
Dual cycleElimination& Shelling
Recursive bisection
Linear-size
Local reconnection and smoothing
Flipping, (planar faces)
Prepare the boundary mesh:splicing
Refinement andcoarsening
Schneider’s open problem
Geode
Hex-to-void
clea
n-up
Geode betweenshells
geometry first topology firstTHEXcons
truc
tion
disc
over
y
Sweep-liketechniques
THEX
? Hex dominant?
Dual-ParticleMethods
Inside-outtechniques
Slicing doughnuts
CUBIT Mesh Generation Project
Technical History of Hex Mesh Generation
Existence Proofs
Parallel thread, unknown at the time to Murdoch, Blacker, Mitchell» Mac Casel, of PDA, posting to sci-math:
– Under what conditions can you hex-mesh a polyhedron without modifying its boundary?
» Bill Thurston (world’s best topologist)– 25 Oct 1993, 39 days after Murdoch rediscovered the STC,
knew about “STC” all along, answers in 1 ½ pages:– If the dual curves (loops) intersect an even number of times,
then there is a topological hex mesh– “As for the geometric question: My first guess is it can be
done geometrically if it can be done topologically, but it looks very tricky…. Which question are you really trying to answer?”
CUBIT Mesh Generation Project
Technical History of Hex Mesh Generation
Existence Proofs
Mitchell’s proof, very similar to Thurston’s outline, a couple years later (mid 1995)» Map surface mesh to a sphere (smooth)» Form STC loops – smooth closed curves» Extend loops into closed surfaces» Fix arrangement of surfaces to avoid degenerate elements» Dualize to form hexes» Map back to original object.
CUBIT Mesh Generation Project
Technical History of Hex Mesh Generation
Necessary Conditions
CUBIT Mesh Generation Project
Technical History of Hex Mesh Generation
Aside: What about tet meshing?
Tet’s have 4 (even) triangles, so all tet meshes are bounded by an even number of tets, too!
But triangle’s have 3 (odd) edges, so any triangle surface mesh has same parity of triangles as bounding edges» Any triangle mesh of a closed volume is even
1,3 2,4 Any triangle mesh of top hassame parity as any triangle mesh of bottom.Hence sum is even
CUBIT Mesh Generation Project
Technical History of Hex Mesh Generation
Sufficient Conditions
Every pair of loops on sphere intersect each other even number of times
A single loop can self-intersect an even or odd number of times» Recall trouble with knives and self-intersecting loops…
CUBIT Mesh Generation Project
Technical History of Hex Mesh Generation
Sufficient Conditions
Topology theorems» For any single curve with even # self-intersections,
can construct a surface» For any pair of curves with odd # self-
intersections, can construct a surface» (Proof is constructive: rumor that non-constructive
is false)
CUBIT Mesh Generation Project
Technical History of Hex Mesh Generation
Sufficient Conditions
If you dualized a very coarse arrangement, you’d get degenerate hexes.Easy to fix with extra balls.
CUBIT Mesh Generation Project
Technical History of Hex Mesh Generation
Sufficient Conditions
Do similar buffering sheet next to surface mesh
CUBIT Mesh Generation Project
Technical History of Hex Mesh Generation
Schneiders Open Problem Closed?
Using these kind of techniques, a dozen people have constructed a “mesh” for the open problem. » One self-intersecting curve, 8 times. 2 curves.» Non-degenerate meshes have > 100 hexes» Poor quality
– But positive jacobians last IMR! 1.0E-4)
CUBIT Mesh Generation Project
Technical History of Hex Mesh Generation
What about non-spheres?
“Map surface mesh to a sphere (smooth)” David Eppstein told me about Thurston’s
proof. Bill Thurston and I were going to write joint
paper. Thurston worked a couple days and decided
rigorously extending to non-balls was harder than it looked, lost interest. So single-author paper
CUBIT Mesh Generation Project
Technical History of Hex Mesh Generation
What about non-spheres?
Necessary: even surface mesh and some edge-chains is necessary. Volume property.
Is it sufficient?» Nate Folwell showed exponential number of hexes
necessary for knotted holes
CUBIT Mesh Generation Project
Technical History of Hex Mesh Generation
What about non-spheres?
Algorithms» V.A. Gasilov (Moscow) reduce topology of
non-spheres to collection of spheres.– “slicing doughnuts”– Mesh with tets, grow a maximal tet collection
that is a ball and doesn’t self-touch, boundary is a cut.
» Jeff Erickson’s recent paper “Optimally cutting a surface into a disk” may be relevant?
CUBIT Mesh Generation Project
Technical History of Hex Mesh Generation
3D Relationship MapPaving/Plastering
Whisker Weavingstate list
STC dual structurerediscovered
Whisker Weavingcurve contraction
STC folds
Existence Proofs
Dual cycleElimination& Shelling
Recursive bisection
Linear-size
Local reconnection and smoothing
Flipping, (planar faces)
Prepare the boundary mesh:splicing
Refinement andcoarsening
Schneider’s open problem
Geode
Hex-to-void
clea
n-up
Geode betweenshells
geometry first topology firstTHEXcons
truc
tion
disc
over
y
Sweep-liketechniques
THEX
? Hex dominant?
Dual-ParticleMethods
Inside-outtechniques
Slicing doughnuts
CUBIT Mesh Generation Project
Technical History of Hex Mesh Generation
Whisker Weave via Curve Contraction
Implemented the existence proof for curves w/out self-intersections, sphere surfaces. (Nate Folwell, Mitchell)» To get surface, take loop and extend topologically next to the
remaining void» Pre-process surface mesh to get rid of self intersections.
– Face collapse and refinement.
Always solves the topology problem (existence proof is a proof, after all…), but rarely the geometric problem
– Thurston: “As for the geometric question: My first guess is it can be done geometrically if it can be done topologically, but it looks very tricky…. Which question are you really trying to answer?”
CUBIT Mesh Generation Project
Technical History of Hex Mesh Generation
Whisker WeavingSelf-Intersection
Removal
Face collapse
and pillowing
CUBIT Mesh Generation Project
Technical History of Hex Mesh Generation
Whisker WeavingSelf-Intersection
Removal
Bad surface mesh quality in practice
CUBIT Mesh Generation Project
Technical History of Hex Mesh Generation
Whisker Weaving via Curve Contraction
Small, isolated regions of poor quality» 30 hexes sharing one node» Revisit later
CUBIT Mesh Generation Project
Technical History of Hex Mesh Generation
3D Relationship MapPaving/Plastering
Whisker Weavingstate list
STC dual structurerediscovered
Whisker Weavingcurve contraction
STC folds
Existence Proofs
Dual cycleElimination& Shelling
Recursive bisection
Linear-size
Local reconnection and smoothing
Flipping, (planar faces)
Prepare the boundary mesh:splicing
Refinement andcoarsening
Schneider’s open problem
Geode
Hex-to-void
clea
n-up
Geode betweenshells
geometry first topology firstTHEXcons
truc
tion
disc
over
y
Sweep-liketechniques
THEX
? Hex dominant?
Dual-ParticleMethods
Inside-outtechniques
Slicing doughnuts
CUBIT Mesh Generation Project
Technical History of Hex Mesh Generation
Dual Cycle Elimination
Matthias Muller-Hannemann Similar to curve contraction algorithm Advantages
» No fix-up, instead prevents bad cases– Sufficient conditions to eliminate a cycle without
introducing degenerate or flat elements– (but sometimes not possible to make progress)
» No self-intersecting curves for arrangements– German CAD tool geometry decomposer into 4-sided
surfaces, and well structured surface mesher
CUBIT Mesh Generation Project
Technical History of Hex Mesh Generation
Dual Cycle Elimination
Basic principle» Cycles eliminated one by one
– As WW curve contraction
CUBIT Mesh Generation Project
Technical History of Hex Mesh Generation
Dual Cycle Elimination
Cycle elimination order must be a shelling» With other algebraic rules
– sufficient to avoid degeneracies
» E.g. forbidden cycle– Maybe a different order is a shelling
CUBIT Mesh Generation Project
Technical History of Hex Mesh Generation
Dual Cycle Elimination
Remove self-intersecting loops ahead of time» Dice one level
– High element count
» Replace self-intersections with template– Good quality compared to WW approach
» Works for assembly with shared surfaces– Doesn’t introduce new self-intersections
CUBIT Mesh Generation Project
Technical History of Hex Mesh Generation
Dual Cycle Elimination
Application success» Human mandible
– Good quality hex mesh– Had to do some manual
work to identify good surface mesh.
CUBIT Mesh Generation Project
Technical History of Hex Mesh Generation
3D Relationship MapPaving/Plastering
Whisker Weavingstate list
STC dual structurerediscovered
Whisker Weavingcurve contraction
STC folds
Existence Proofs
Dual cycleElimination& Shelling
Recursive bisection
Linear-size
Local reconnection and smoothing
Flipping, (planar faces)
Prepare the boundary mesh:splicing
Refinement andcoarsening
Schneider’s open problem
Geode
Hex-to-void
clea
n-up
Geode betweenshells
geometry first topology firstTHEXcons
truc
tion
disc
over
y
Sweep-liketechniques
THEX
? Hex dominant?
Dual-ParticleMethods
Inside-outtechniques
Slicing doughnuts
CUBIT Mesh Generation Project
Technical History of Hex Mesh Generation
Recursive Bisection
Calvo and Idelsohn in Argentina» Anyone want to pay for their visit?
Select a cycle for elimination = insert a layer of hexes» Instead of layer next to boundary (as WW or
Shelling), it cuts the model in two.– Layer has connectivity of one of the sides– Maybe better geometrically?
» Two halves resolved independently– Topological “monsters” (degeneracies) as WW
CUBIT Mesh Generation Project
Technical History of Hex Mesh Generation
Bisection v.s. Cycle Elimination
Recursive bisection Shelling / WW
CUBIT Mesh Generation Project
Technical History of Hex Mesh Generation
3D Relationship MapPaving/Plastering
Whisker Weavingstate list
STC dual structurerediscovered
Whisker Weavingcurve contraction
STC folds
Existence Proofs
Dual cycleElimination& Shelling
Recursive bisection
Linear-size
Local reconnection and smoothing
Flipping, (planar faces)
Prepare the boundary mesh:splicing
Refinement andcoarsening
Schneider’s open problem
Geode
Hex-to-void
clea
n-up
Geode betweenshells
geometry first topology firstTHEXcons
truc
tion
disc
over
y
Sweep-liketechniques
THEX
? Hex dominant?
Dual-ParticleMethods
Inside-outtechniques
Slicing doughnuts
CUBIT Mesh Generation Project
Technical History of Hex Mesh Generation
Dual Particle Methods Analyst group at SNL tired of waiting for WW-like approaches, …start
with 2d» Jung, Dohrmann, Witkowski, Wolfenbarger, Gerstle, Mitchell, Panthaki,
Segalman» Wanted something more mathematical» Solve geometric placement and connectivity simultaneously» Throw dual particles into a domain
– Thought there would be an Euler’s formula that would say how many
» Position them using optimization (minimum energy configuration)» Connect them up, dualize to get quads
CUBIT Mesh Generation Project
Technical History of Hex Mesh Generation
Dual Particle Methods
Physics forces between particles» Attraction, repulsion, damping, torque
CUBIT Mesh Generation Project
Technical History of Hex Mesh Generation
Dual Particle Methods
Forces are local, didn’t always close up nicely» Heuristic local
reconnection» Particle add &
remove» Helped in 2d
CUBIT Mesh Generation Project
Technical History of Hex Mesh Generation
Dual Particle Methods
Initial position dual particles based on solution to Laplace equation.» Almost as good as paving, when it worked…
random Laplace
CUBIT Mesh Generation Project
Technical History of Hex Mesh Generation
Dual Particle Methods – 3D
3D - Leung» Position particles as before» Write down all the constraints from existence proof
» Solve constraints via discrete optimization– Add dual geometric constraints as they are violated (no cells interpenetrate)
– Huge running time
– Mostly solved topology of hex mesh. Geometric quality not good enough.
CUBIT Mesh Generation Project
Technical History of Hex Mesh Generation
Dual Particle Methods – 3D
CUBIT Mesh Generation Project
Technical History of Hex Mesh Generation
Dual Particle Methods – 3D
Geometric problems still- interpenetrations
CUBIT Mesh Generation Project
Technical History of Hex Mesh Generation
3D Relationship MapPaving/Plastering
Whisker Weavingstate list
STC dual structurerediscovered
Whisker Weavingcurve contraction
STC folds
Existence Proofs
Dual cycleElimination& Shelling
Recursive bisection
Linear-size
Local reconnection and smoothing
Flipping, (planar faces)
Prepare the boundary mesh:splicing
Refinement andcoarsening
Schneider’s open problem
Geode
Hex-to-void
clea
n-up
Geode betweenshells
geometry first topology firstTHEXcons
truc
tion
disc
over
y
Sweep-liketechniques
THEX
? Hex dominant?
Dual-ParticleMethods
Inside-outtechniques
Slicing doughnuts
CUBIT Mesh Generation Project
Technical History of Hex Mesh Generation
Linear-size Hex Meshes
Eppstein» Mitchell-Thurston algorithms produce too many
hexes» n = number of surface quads
O(n) on faceO(n) on each side
O(n) on edge
O(n1/2) on edge
CUBIT Mesh Generation Project
Technical History of Hex Mesh Generation
Linear-Size Hex Meshes
Steps
Hex-shaped boundary tiles
THex interiorO(n) tets
Original shape
CUBIT Mesh Generation Project
Technical History of Hex Mesh Generation
Linear-Size Hex Meshing
For each tile» Fill with hexes
– Matching to create a couple of types of tiles, all satisfy sufficient conditions of existence proofs
– Solves topology problem, not geometry problem
– Anybody figured out exactly how to fill?I didn’t…like Schneiders open problem
CUBIT Mesh Generation Project
Technical History of Hex Mesh Generation
3D Relationship MapPaving/Plastering
Whisker Weavingstate list
STC dual structurerediscovered
Whisker Weavingcurve contraction
STC folds
Existence Proofs
Dual cycleElimination& Shelling
Recursive bisection
Linear-size
Local reconnection and smoothing
Flipping, (planar faces)
Prepare the boundary mesh:splicing
Refinement andcoarsening
Schneider’s open problem
Geode
Hex-to-void
clea
n-up
Geode betweenshells
geometry first topology firstTHEXcons
truc
tion
disc
over
y
Sweep-liketechniques
THEX
? Hex dominant?
Dual-ParticleMethods
Inside-outtechniques
Slicing doughnuts
CUBIT Mesh Generation Project
Technical History of Hex Mesh Generation
Geode - motivation
Recall linear-hex Recall Plastering, hex to void
CUBIT Mesh Generation Project
Technical History of Hex Mesh Generation
Geode Motivation
3d problem
How to conformally interface in 3d?
CUBIT Mesh Generation Project
Technical History of Hex Mesh Generation
Geode
There is a transition template that someone (me) has filled with ok-quality hexes» Based on constructing sheets, interleaving
Hexesdivided into 8
Tetsdivided into 4 hexes
CUBIT Mesh Generation Project
Technical History of Hex Mesh Generation
Geode – 2d
CUBIT Mesh Generation Project
Technical History of Hex Mesh Generation
Geode – 3d
Hex side
Tet side
CUBIT Mesh Generation Project
Technical History of Hex Mesh Generation
Geode – 3d
CUBIT Mesh Generation Project
Technical History of Hex Mesh Generation
Geode between shells
Ideal template has scaled jacobian 0.26» Comparable to THex mesh
In practice, almost never get well-shaped layer» Good for single layer on planar surface…
thexTHexgeode
geode
mesh with hexes
hex
CUBIT Mesh Generation Project
Technical History of Hex Mesh Generation
Geode
Why geode and not other templates?» Self-intersections? Then why not 4-triangle geode:
» We know a few things that make a surface mesh hard to fill with hexes, but we still don’t know how to make it easy.
CUBIT Mesh Generation Project
Technical History of Hex Mesh Generation
Geode Eliminates Self-Intersecting Loops!
Unpublished» Recall Dual Cycle Elimination
– 1. Dice– 2. Replace self-int with template
» Instead, – 2. Add one geode template (not layer)– Remaining region has no self-intersection– Conjecture: after dicing, every surface mesh of any volume admits a
compatible, linear-size, positive jacobian hexahedral mesh.
Inside of volume
CUBIT Mesh Generation Project
Technical History of Hex Mesh Generation
Geode and Linear-Size
What’s so special about dicing? Is quality going to be that good?
» Doubt it would be that practicle» Revisit loops that are self-nearby later
CUBIT Mesh Generation Project
Technical History of Hex Mesh Generation
3D Relationship MapPaving/Plastering
Whisker Weavingstate list
STC dual structurerediscovered
Whisker Weavingcurve contraction
STC folds
Existence Proofs
Dual cycleElimination& Shelling
Recursive bisection
Linear-size
Local reconnection and smoothing
Flipping, (planar faces)
Prepare the boundary mesh:splicing
Refinement andcoarsening
Schneider’s open problem
Geode
Hex-to-void
clea
n-up
Geode betweenshells
geometry first topology firstTHEXcons
truc
tion
disc
over
y
Sweep-liketechniques
THEX
? Hex dominant?
Dual-ParticleMethods
Inside-outtechniques
Slicing doughnuts
CUBIT Mesh Generation Project
Technical History of Hex Mesh Generation
Prepare the Boundary Mesh
Recall WW, Shelling, and Geode » approaches to preparing the surface mesh to
admit a (good quality?) hex mesh
CUBIT Mesh Generation Project
Technical History of Hex Mesh Generation
Algorithms - splicing
Mesh Editing and Splicing
Initial GridRemove self intersecting chord –
Leave boundary nodes fixed
CUBIT Mesh Generation Project
Technical History of Hex Mesh Generation
Algorithms - splicing
Mesh Editing and Splicing
Intermediate mesh – needs new chord to reconcile boundary
New non self intersecting chord inserted to reconcile boundary nodes
CUBIT Mesh Generation Project
Technical History of Hex Mesh Generation
Surface Mesh Modification - Splicing
Undesirable mesh interval
Remove unwanted quads Completed MeshMesh from W76 AFF
The procedure:
CUBIT Mesh Generation Project
Technical History of Hex Mesh Generation
Fixing Surface Mesh for WW
• No self-intersections, but pretty convoluted• Remove it!
CUBIT Mesh Generation Project
Technical History of Hex Mesh Generation
Fixing Surface Meshes
In progress» Automation
How general of a method will it be?» What’s properties are really sufficient for
the surface mesh?
CUBIT Mesh Generation Project
Technical History of Hex Mesh Generation
3D Relationship MapPaving/Plastering
Whisker Weavingstate list
STC dual structurerediscovered
Whisker Weavingcurve contraction
STC folds
Existence Proofs
Dual cycleElimination& Shelling
Recursive bisection
Linear-size
Local reconnection and smoothing
Flipping, (planar faces)
Prepare the boundary mesh:splicing
Refinement andcoarsening
Schneider’s open problem
Geode
Hex-to-void
clea
n-up
Geode betweenshells
geometry first topology firstTHEXcons
truc
tion
disc
over
y
Sweep-liketechniques
THEX
? Hex dominant?
Dual-ParticleMethods
Inside-outtechniques
Slicing doughnuts
CUBIT Mesh Generation Project
Technical History of Hex Mesh Generation
Where are we today?
Summary» Nothing yet is
– Boundary mesh conforming– Automatic– Good quality– Applicable to all geometries
Matthias very good for mandible-like problems
CUBIT Mesh Generation Project
Technical History of Hex Mesh Generation
Best hopes
Suspect that only certain surface meshes admit a good-quality mesh
Combined geom & topology techniques Difficult to get tight necessary and
sufficient conditions -> sufficient might be doable
CUBIT Mesh Generation Project
Technical History of Hex Mesh Generation
Current Activity
Show why hexes are hard» Flipping
Fix existing mesh in unstructured way» Past efforts not promising
Semi-structured sweep-like techniques» Grafting, Coopering
Improve existing mesh in semi-structured way» Looks promising
Surface and volume all-at-once looking better» Mesh cutting
CUBIT Mesh Generation Project
Technical History of Hex Mesh Generation
3D Relationship MapPaving/Plastering
Whisker Weavingstate list
STC dual structurerediscovered
Whisker Weavingcurve contraction
STC folds
Existence Proofs
Dual cycleElimination& Shelling
Recursive bisection
Linear-size
Local reconnection and smoothing
Flipping, (planar faces)
Prepare the boundary mesh:splicing
Refinement andcoarsening
Schneider’s open problem
Geode
Hex-to-void
clea
n-up
Geode betweenshells
geometry first topology firstTHEXcons
truc
tion
disc
over
y
Sweep-liketechniques
THEX
? Hex dominant?
Dual-ParticleMethods
Inside-outtechniques
Slicing doughnuts
CUBIT Mesh Generation Project
Technical History of Hex Mesh Generation
Flipping
Bern and Eppstein Goals
» Find set of operations sufficient for all desired topological changes
» Relate local improvement to problems of mesh existence
» Computational Geometry contributions to meshing, welcome complement to engineering approaches
CUBIT Mesh Generation Project
Technical History of Hex Mesh Generation
Flipping Triangle Meshes
Triangle meshes» Two types of flips» Equivalent to gluing tetrahedron on top of existing
triangle mesh, take non-touching tris» Flip graph is connected
CUBIT Mesh Generation Project
Technical History of Hex Mesh Generation
Flipping Quads
Flip graph is connected » Up to parity» For disk surfaces
no
no
CUBIT Mesh Generation Project
Technical History of Hex Mesh Generation
Flipping Tetrahedra
Flip graph connected for» Topological tets, unknown for geometric tets» Swap top and bottom of 4d simplex
CUBIT Mesh Generation Project
Technical History of Hex Mesh Generation
Flipping Hexestop and bottom of 4D
cuboid
CUBIT Mesh Generation Project
Technical History of Hex Mesh Generation
Flipping Hexes
Some flips preserve flatness of faces» Does warped boundary mesh imply
warped hexes?» Beyond flatness, can we someday
determine the maximum quality of a hex mesh matching a given surface mesh?
CUBIT Mesh Generation Project
Technical History of Hex Mesh Generation
3D Relationship MapPaving/Plastering
Whisker Weavingstate list
STC dual structurerediscovered
Whisker Weavingcurve contraction
STC folds
Existence Proofs
Dual cycleElimination& Shelling
Recursive bisection
Linear-size
Local reconnection and smoothing
Flipping, (planar faces)
Prepare the boundary mesh:splicing
Refinement andcoarsening
Schneider’s open problem
Geode
Hex-to-void
clea
n-up
Geode betweenshells
geometry first topology firstTHEXcons
truc
tion
disc
over
y
Sweep-liketechniques
THEX
? Hex dominant?
Dual-ParticleMethods
Inside-outtechniques
Slicing doughnuts
CUBIT Mesh Generation Project
Technical History of Hex Mesh Generation
Local Reconnection and Smoothing
Knupp’s untangle smoothing» Is able to get positive jacobians in interior of
meshes that we know admit them– Was unable to untangle WW meshes– Issues
Provable? Perfect implementation?
Idea » Combine reconnection and smoothing for
unstructured 3d hex meshes– As done in every other class of unstructured meshers…
CUBIT Mesh Generation Project
Technical History of Hex Mesh Generation
Local Reconnection and Smoothing
Find local area of poor quality Classify type of problem, possible improvements Find best improvement
» Swap local connectivity via primitives» Locally smooth» Assess quality» Undo
Apply best improvement, if any
CUBIT Mesh Generation Project
Technical History of Hex Mesh Generation
Local Reconnection and Smoothing
Flips, but dual interpretation» Might not be flips (non-parity preserving)
for non-ball volumes
Moving sheetsto untangle
Also sheet pushing ideas for local element size control
CUBIT Mesh Generation Project
Technical History of Hex Mesh Generation
Local Reconnection and Smoothing
Connecting sheets
Inserting sheets
CUBIT Mesh Generation Project
Technical History of Hex Mesh Generation
Local Reconnection and Smoothing
Most-untangled sheets not always the best quality Wasn’t able to improve specific small examples, by
any manual means» Just push problem around
Process didn’t work» Search space very large» Progress possible but not monotonic? (unlike quad, tri, tet)
– Space of valid hex meshes too coarse, not well connected?
» Progress not possible?
CUBIT Mesh Generation Project
Technical History of Hex Mesh Generation
3D Relationship MapPaving/Plastering
Whisker Weavingstate list
STC dual structurerediscovered
Whisker Weavingcurve contraction
STC folds
Existence Proofs
Dual cycleElimination& Shelling
Recursive bisection
Linear-size
Local reconnection and smoothing
Flipping, (planar faces)
Prepare the boundary mesh:splicing
Refinement andcoarsening
Schneider’s open problem
Geode
Hex-to-void
clea
n-up
Geode betweenshells
geometry first topology firstTHEXcons
truc
tion
disc
over
y
Sweep-liketechniques
THEX
? Hex dominant?
Dual-ParticleMethods
Inside-outtechniques
Slicing doughnuts
CUBIT Mesh Generation Project
Technical History of Hex Mesh Generation
Sweep Relationship Map
Sweeping Node Placement Strategies
2D2.
5D2.
75D
Good Source SurfaceMesh Strategies
1-1 Sweeping / Projection / Rotation (lots)
M-1 Sweeping (several)
M-M Sweeping
Multi-axisCooper Tool Grafting
Cooper Tool
Inside-out3d
CUBIT Mesh Generation Project
Technical History of Hex Mesh Generation
Sweeping Types
One-to-One SweepMany-to-Many
Multisweep & Coopering
Many-to-One Sweep
CUBIT Mesh Generation Project
Technical History of Hex Mesh Generation
MultiSweep
•Create virtual surfaces w/ imprints•Surface mesh the combination•Sweep the underlying volumes
CUBIT Mesh Generation Project
Technical History of Hex Mesh Generation
MultiSweep
Image of target curvespartitions source surfaces
Image created using • matrix (affine transformation)• or weighted winslow smoothingone mesh layer at a time
The image is a mesh image depends on resolution, premesh
CUBIT Mesh Generation Project
Technical History of Hex Mesh Generation
Multisweep – in progress
CUBIT Mesh Generation Project
Technical History of Hex Mesh Generation
Cooper Tool
Like multisweep» Nice language of caps, barrels» Different projection methods
– All layers at once
» Intersection of projections different– …
CUBIT Mesh Generation Project
Technical History of Hex Mesh Generation
Multi-Axis Coopering
Blacker and Miyoshi Break model into tree
CUBIT Mesh Generation Project
Technical History of Hex Mesh Generation
Multi-Axis Coopering
“Cut” section to sweep in new direction» Cuts can intersect and overlap - tricky
CUBIT Mesh Generation Project
Technical History of Hex Mesh Generation
Mesh grafting
CUBIT Mesh Generation Project
Technical History of Hex Mesh Generation
3D Relationship MapPaving/Plastering
Whisker Weavingstate list
STC dual structurerediscovered
Whisker Weavingcurve contraction
STC folds
Existence Proofs
Dual cycleElimination& Shelling
Recursive bisection
Linear-size
Local reconnection and smoothing
Flipping, (planar faces)
Prepare the boundary mesh:splicing
Refinement andcoarsening
Schneider’s open problem
Geode
Hex-to-void
clea
n-up
Geode betweenshells
geometry first topology firstTHEXcons
truc
tion
disc
over
y
Sweep-liketechniques
THEX
? Hex dominant?
Dual-ParticleMethods
Inside-outtechniques
Slicing doughnuts
CUBIT Mesh Generation Project
Technical History of Hex Mesh Generation
Refinement and Coarsening
Shepherd, Borden,…
Boundary Refinement
Mesh Coarsening
Sheet Refinement& Cleaving
CUBIT Mesh Generation Project
Technical History of Hex Mesh Generation
Sheet Insertion
Modify mesh to capture geometry, through STC sheets
CUBIT Mesh Generation Project
Technical History of Hex Mesh Generation
The problem is to fit the mesh on the left to the volume on the right. The rest of the volumes in the model can then be swept away from this volume.
Mesh Cutting & Sheet Insertion
CUBIT Mesh Generation Project
Technical History of Hex Mesh Generation
Step 1: Capturing the first imprint.
CUBIT Mesh Generation Project
Technical History of Hex Mesh Generation
Step 2: Capturing the other two imprints
CUBIT Mesh Generation Project
Technical History of Hex Mesh Generation
Step 3: Capturing the Cutouts
CUBIT Mesh Generation Project
Technical History of Hex Mesh Generation
Mesh Cutting
Questions » Can we do this for assemblies?» How wild of shapes can it handle?» Compare vs. Schneiders thesis» Compare vs. Murdoch thesis
CUBIT Mesh Generation Project
Technical History of Hex Mesh Generation
Conclusions
Technical history of unstructured meshing ideas» Common themes» Lots left to do and explore
– Enough background ideas to start research
» Probably not much low-hanging fruit left– Hope of practical tools before hex meshing is obsolete
» Gap between theory (understanding) and practice– closing a bit