Carlos A. Felippa

36
University of Colorado - Dept of Aerospace Engrg. Sci. & Center for Aerospace Structures - caswww.colorado.edu Carlos A. Felippa Recent Advances in Finite Recent Advances in Finite Element Templates Element Templates Department of Aerospace Engineering Sciences and Center for Aerospace Structures University of Colorado at Boulder Boulder, CO 80309, USA Presentation to the CST 2000 September 8, 2000, Leuven, Belgium

description

Recent Advances in Finite Element Templates. Carlos A. Felippa. Department of Aerospace Engineering Sciences and Center for Aerospace Structures University of Colorado at Boulder Boulder, CO 80309, USA. Presentation to the CST 2000 September 8, 2000, Leuven, Belgium. Outline. - PowerPoint PPT Presentation

Transcript of Carlos A. Felippa

Page 1: Carlos A. Felippa

University of Colorado - Dept of Aerospace Engrg. Sci. & Center for Aerospace Structures - caswww.colorado.edu

Carlos A. Felippa

Recent Advances in Finite Recent Advances in Finite Element TemplatesElement Templates

Recent Advances in Finite Recent Advances in Finite Element TemplatesElement Templates

Department of Aerospace Engineering Sciences and Center for Aerospace Structures

University of Colorado at BoulderBoulder, CO 80309, USA

Presentation to the CST 2000September 8, 2000, Leuven, Belgium

Page 2: Carlos A. Felippa

University of Colorado - Dept of Aerospace Engrg. Sci. & Center for Aerospace Structures - caswww.colorado.edu

• High Performance (HP) elements

• Templates: concept, “genetics”

• Constraints on template parameters families

• Kirchhoff Plate Triangle (KPT) template benchmarks

• Conclusions

OutlineOutline

Page 3: Carlos A. Felippa

University of Colorado - Dept of Aerospace Engrg. Sci. & Center for Aerospace Structures - caswww.colorado.edu

Evolution of FEMEvolution of FEM

Page 4: Carlos A. Felippa

University of Colorado - Dept of Aerospace Engrg. Sci. & Center for Aerospace Structures - caswww.colorado.edu

High Performance Elements - Definition

High Performance Elements - Definition

See written paper for discussion of “simple”, “engineering accuracy” “arbitrary” and “coarse”

Simple elements that deliver results of engineering accuracy with arbitrary coarse meshes(Felippa & Militello, 1989)

Page 5: Carlos A. Felippa

University of Colorado - Dept of Aerospace Engrg. Sci. & Center for Aerospace Structures - caswww.colorado.edu

• 1965-date: Old Tricks– Incompatible shape functions, reduced & selective integration

• 1968-date: More Scientific – Hybrid/mixed elements, enhanced strain, stabilized elements

• 1975-date: Physically Based– Free Formulation, Assumed Natural Strain

• Author’s Approach– A mixture of above, ending with templates (next slide)

Approaches to the Construction of HP Elements

Approaches to the Construction of HP Elements

Page 6: Carlos A. Felippa

University of Colorado - Dept of Aerospace Engrg. Sci. & Center for Aerospace Structures - caswww.colorado.edu

The Road to TemplatesThe Road to Templates

Page 7: Carlos A. Felippa

University of Colorado - Dept of Aerospace Engrg. Sci. & Center for Aerospace Structures - caswww.colorado.edu

Templates are: parametrized forms of element level FEM equations that satisfy:

(C) Consistency– The Individual Element Test (IET) of Bergan and Hanssen

(1975) is identically passed

(S) Stability– Element operators (stiffness, mass, etc) have correct rank

(I) Observer invariance

(P) Contain free parameters

Template DefinitionTemplate Definition

Page 8: Carlos A. Felippa

University of Colorado - Dept of Aerospace Engrg. Sci. & Center for Aerospace Structures - caswww.colorado.edu

Example: Stiffness of BE Plane Beam Element

Example: Stiffness of BE Plane Beam Element

Rank 1 Rank 1+

Page 9: Carlos A. Felippa

University of Colorado - Dept of Aerospace Engrg. Sci. & Center for Aerospace Structures - caswww.colorado.edu

Fundamental Decomposition of Template Stiffness Matrix

Fundamental Decomposition of Template Stiffness Matrix

Page 10: Carlos A. Felippa

University of Colorado - Dept of Aerospace Engrg. Sci. & Center for Aerospace Structures - caswww.colorado.edu

• The set of free parameters is the template signature

• The number of free parameters can be reduced by applying behavioral constraints to produce element families

• Specific elements instances are obtained by assigning numerical values to the free parameters of a family

• Elements with the same signature, possibly derived through different methods, are called clones

Template “Genetics”Template “Genetics”

Page 11: Carlos A. Felippa

University of Colorado - Dept of Aerospace Engrg. Sci. & Center for Aerospace Structures - caswww.colorado.edu

Kirchhoff Plate Bending Triangle (KPT) Template

Kirchhoff Plate Bending Triangle (KPT) Template

Page 12: Carlos A. Felippa

University of Colorado - Dept of Aerospace Engrg. Sci. & Center for Aerospace Structures - caswww.colorado.edu

Identifiers ofExisting KPT Elements

Identifiers ofExisting KPT Elements

Page 13: Carlos A. Felippa

University of Colorado - Dept of Aerospace Engrg. Sci. & Center for Aerospace Structures - caswww.colorado.edu

Signatures of Existing KPT Elements

Signatures of Existing KPT Elements

Page 14: Carlos A. Felippa

University of Colorado - Dept of Aerospace Engrg. Sci. & Center for Aerospace Structures - caswww.colorado.edu

Linear Constraints on Template Parameters

Linear Constraints on Template Parameters

• Observer Invariance– Equations invariant wrt node numbering; symmetries preserved

• Aspect Ratio Insensitivity– Energy ratio remains bounded as element aspect ratio(s) goes to infinity

– Avoids “aspect ratio locking”

• Energy orthogonality– Always used in older work, nowadays optional

Page 15: Carlos A. Felippa

University of Colorado - Dept of Aerospace Engrg. Sci. & Center for Aerospace Structures - caswww.colorado.edu

Aspect Ratio Insensitivity for KPTs

Aspect Ratio Insensitivity for KPTs

Page 16: Carlos A. Felippa

University of Colorado - Dept of Aerospace Engrg. Sci. & Center for Aerospace Structures - caswww.colorado.edu

Configurations to Apply ARI Constraints

Configurations to Apply ARI Constraints

Page 17: Carlos A. Felippa

University of Colorado - Dept of Aerospace Engrg. Sci. & Center for Aerospace Structures - caswww.colorado.edu

Linear Constraints for KPT Template

Linear Constraints for KPT Template

Page 18: Carlos A. Felippa

University of Colorado - Dept of Aerospace Engrg. Sci. & Center for Aerospace Structures - caswww.colorado.edu

Three KPT Families AppearThree KPT Families Appear

Page 19: Carlos A. Felippa

University of Colorado - Dept of Aerospace Engrg. Sci. & Center for Aerospace Structures - caswww.colorado.edu

Quadratic Constraints on Template Parameters

Quadratic Constraints on Template Parameters

• Morphing– Next slides

• Higher order patch tests

• Mesh distortion insensitivity

• Others under study– Lack of directionality in wave propagation

Page 20: Carlos A. Felippa

University of Colorado - Dept of Aerospace Engrg. Sci. & Center for Aerospace Structures - caswww.colorado.edu

Morphing ConceptMorphing Concept

Page 21: Carlos A. Felippa

University of Colorado - Dept of Aerospace Engrg. Sci. & Center for Aerospace Structures - caswww.colorado.edu

Morphing DOF MatchingMorphing DOF Matching

Page 22: Carlos A. Felippa

University of Colorado - Dept of Aerospace Engrg. Sci. & Center for Aerospace Structures - caswww.colorado.edu

Importance of Morphing in Aerospace Structures

Importance of Morphing in Aerospace Structures

Page 23: Carlos A. Felippa

University of Colorado - Dept of Aerospace Engrg. Sci. & Center for Aerospace Structures - caswww.colorado.edu

F-16 Aeroelastic Structural ModelF-16 Aeroelastic Structural Model

Present model:150000 Nodes,6 DOFs/Node

Page 24: Carlos A. Felippa

University of Colorado - Dept of Aerospace Engrg. Sci. & Center for Aerospace Structures - caswww.colorado.edu

F-16 Exterior Surface ZoomF-16 Exterior Surface Zoom

95% ofelements areHPSHEL318 DOF shells

Page 25: Carlos A. Felippa

University of Colorado - Dept of Aerospace Engrg. Sci. & Center for Aerospace Structures - caswww.colorado.edu

F-16 Internal Structure ZoomF-16 Internal Structure Zoom

Some solid elements (bricks & tetrahedra) used for “wing fingers”

Page 26: Carlos A. Felippa

University of Colorado - Dept of Aerospace Engrg. Sci. & Center for Aerospace Structures - caswww.colorado.edu

Interesting New KPT Element Instances

Interesting New KPT Element Instances

Page 27: Carlos A. Felippa

University of Colorado - Dept of Aerospace Engrg. Sci. & Center for Aerospace Structures - caswww.colorado.edu

Signatures of Interesting New KPT Elements

Signatures of Interesting New KPT Elements

Page 28: Carlos A. Felippa

University of Colorado - Dept of Aerospace Engrg. Sci. & Center for Aerospace Structures - caswww.colorado.edu

“Genealogy” of Existing & New Elements

“Genealogy” of Existing & New Elements

Page 29: Carlos A. Felippa

University of Colorado - Dept of Aerospace Engrg. Sci. & Center for Aerospace Structures - caswww.colorado.edu

Benchmark: SS Square Plate Under Central Load

Benchmark: SS Square Plate Under Central Load

Page 30: Carlos A. Felippa

University of Colorado - Dept of Aerospace Engrg. Sci. & Center for Aerospace Structures - caswww.colorado.edu

Benchmark: Clamped Square Plate Under Central Load

Benchmark: Clamped Square Plate Under Central Load

Page 31: Carlos A. Felippa

University of Colorado - Dept of Aerospace Engrg. Sci. & Center for Aerospace Structures - caswww.colorado.edu

Benchmark: Uniformly Loaded CantileverBenchmark: Uniformly Loaded Cantilever

Page 32: Carlos A. Felippa

University of Colorado - Dept of Aerospace Engrg. Sci. & Center for Aerospace Structures - caswww.colorado.edu

Benchmark: End Shear Loaded Cantilever

Benchmark: End Shear Loaded Cantilever

Page 33: Carlos A. Felippa

University of Colorado - Dept of Aerospace Engrg. Sci. & Center for Aerospace Structures - caswww.colorado.edu

Benchmark: Twisted Ribbon (Robinson’s Test)

Benchmark: Twisted Ribbon (Robinson’s Test)

Page 34: Carlos A. Felippa

University of Colorado - Dept of Aerospace Engrg. Sci. & Center for Aerospace Structures - caswww.colorado.edu

The KPT Benchmark Score So FarThe KPT Benchmark Score So Far

• No “best instance for all seasons” has emerged

• HCTS (new), MDIT1 (new), AQR1 (old) are best overall performers

• Bending- and Twist-Exact instances outperform others in cases favoring beam and twib morphing

• Mesh distortion insensitity associated with = 1

Page 35: Carlos A. Felippa

University of Colorado - Dept of Aerospace Engrg. Sci. & Center for Aerospace Structures - caswww.colorado.edu

Conclusions 1: Advantages of Template Approach

Conclusions 1: Advantages of Template Approach

• One routine does all possible elements

– Advantageous in benchmarking

• HP families emerge naturally

• Can be customized to problem at hand

– Static, vibration, buckling, wave propagation ...

• Signatures detect clones

Page 36: Carlos A. Felippa

University of Colorado - Dept of Aerospace Engrg. Sci. & Center for Aerospace Structures - caswww.colorado.edu

Conclusions 2: Difficulties of Template Approach

Conclusions 2: Difficulties of Template Approach

• Heavy symbolic manipulations required

– On present computers, restricted to 1D and simple 2D configurations

• Mathematical framework needed

– In particular, precise connection between template constraints and global

errors