New Approaches for Shape & Topology Optimization for ...New Approaches for Shape & Topology Optimization for Crashworthiness keynote lecture ... State-of-the-Art in Shape & Topology

Technische Universität München

New Approaches for

Shape & Topology Optimization

for Crashworthinesskeynote lecture

Prof. Dr. Fabian DuddeckTechnische Universität München

[email protected]


Acknowledgements

May-11 Duddeck 2

My PhD students:

S. Hunkeler, M. Rayamajhi, A. Berger

Queen Mary University of London (QMUL)

K. Volz, T. Dietrich, J. Fender

Technische Universität München (TUM)

C. Ortmann TUM in collaboration with HAW Hamburg

My research partners

K.-U. Bletzinger and his team from the TUM

H. Zimmer, M. Prabhuwaingankar, and the team of SFE GmbH

L. Rota, M. Zarroug (PSA Peugeot-Citroen)

A. Schumacher (HAW Hamburg) and E. Schelkle (ASC-S)

H. Müllerschön (DYNAmore GmbH)

F. Dirschmid, G. Fichtinger, O. Meyer, A. Übelacker, M. Zimmermann (BMW)

J. Will (DYNARDO GmbH)

T. Pyttel (ESI GmbH & FH Gießen-Friedberg)


Outline

May-11 Duddeck 3

1. New Challenges

2. State-of-the-Art in Shape & Topology Optimization

for Crashworthiness

3. New Approaches for Shape & Topology Optimization

for Crashworthiness

4. Conclusions


Outline

May-11 Duddeck 4

1. New Challenges


for Crashworthiness


for Crashworthiness

4. Conclusions


May-11 Duddeck 5

New Challenges: New Structural Concepts!

ETC/ACC Technical Paper 2009/4F. Hacker, R. Harthan, F. Matthes, W. Zimmer

Aixam MegaMega E-City

BMWE-mini

???

Bolloré/PininfarinaBlue Car

BYDE6

DaimlerSmart ed

General MotorsVolt

HeuliezWill

Mitsubishii-Miev

NissanE-Cube

RevaG-Wiz

Electric Cars

SubaruR1e

Tesla MotorsRoadster

Th!nkCity

VolkswagenTwin Drive


www.loremo.com

May-11 Duddeck 6

New Challenges: New Structural Concepts!


May-11 Duddeck 7

New Challenges: Electromobility

www.llb.mw.tum.dewww.mute-automobile.de


May-11 Duddeck 8

New Challenges: Electromobility

www.llb.mw.tum.dewww.mute-automobile.de


May-11 Duddeck 9

New Challenges: New Materials (CFRP)

www.bmw-i.de


May-11 Duddeck 10

Lightweight Design (e.g. SuperLightCar EU Project)

M. Stehlin, Volkswagen 2008

2005 2006 2007 2008 2009

Materials & Manufacturing Technologies

Simulation & Testing

Design Concepts

SP2

Technology-Benchmarking

Forming

Joining

Assembly

SP1

SP3

SLC Body Concept

1) Steel intensive

2) Multi material, economic (ULBC)

3) Multi material, advanced (SLBC)

< 2,5 €/kg

< 5 €/kg

< 10 €/kg

Life Cycle Analysis (LCA) ■ Cost assessment ■ Virtual and physical Testing

End of 2008Complete

Multi-MaterialBody structure

Prototype Demonstration

SP4


May-11 Duddeck 11

New Approach: Integral Lightweight Design

SFE GmbH & Porsche

+ Optimal Structure(Size, Shape & Topology)

Established: Multi-material, forming, joining assembly, life-cycle, etc.

+ Optimal Package & Design(New Concepts)

SFE GmbH


Concept

Material Steel Aluminium Composites

Topology

Shape

Size (thickness)

May-11 Duddeck 12

Shape and Topology Optimization for Crash

Zimmer, Hänschke


Topology Opt. for Crash (1): Ground Structure

May-11 Duddeck 13

• Design variables: size of cross-section areas

• Objectives: acceleration, displacement, compliance

Ground structure approach

Domain ismeshed by

beams.

Pedersen (2004)


Topology Opt. for Crash (1): Ground Structure

May-11 Duddeck 14

Rigid wall (sliding) Passenger cabin

Car mass750 kg

Engine75 kg

Tire

Pedersen (2004)

• Implicit FEM• Analytic sensitivities• No contact, no inertia• Low generality in geometry


Topology Opt. for Crash (2): Equivalent Static Loads

May-11 Duddeck 15

Communal structurePartly comm. structureTopology variationSpecific structure

Mass-beam model for size, shape and topology optimisation

• Nonlinear crash cases represented by equivalent linear static load cases

• Only beam elements and single masses

• For optimisation: size variablesdimensions of the cross-sections (gauge, height, width, orientation)

• Shape variables:Connection points of the beams

Torstenfelt B and Klarbring A (2007): Conceptual optimal design of modular car product families using simultaneous size, shape and

topology optimization. Finite Elem. in Analysis & Design 43, 1050 – 1061.


Shape Opt. for Crash (1): Morphing

May-11 Duddeck 16

Morphing can only realise small geometrical modifications

a More flexible shape optimization methods required

Morphing

boxes

Georgios 2009


Shape Opt. for Crash (1): Beyond Morphing

May-11 Duddeck 17

SFE CONCEPT model SFE CONCEPT FE model

H. Zimmer, M. Prabhuwaingankar, F. Duddeck (2009): Topology & geometry based structure optimization using implicit parametric models and LSOPT. 7th Europ. LS-DYNA Conf., Salzburg, Austria.

© SFE GmbH

2011


Outline

May-11 Duddeck 18

1. New Challenges


for Crashworthiness


for Crashworthiness

4. Conclusions


Topology Opt. for Crash (3): Hybrid Cellular Automata

May-11 Duddeck 19

Crash analysis

no

Convergence test

yes

Initial design

Update material distribution

using HCA ruleΔx = f(U,U*)

x(k+1) = x(k)+Δx

U(x(k))

Final design

x(k+1)

A Tovar & JE Renaud (2008): Altair HyperWorks Symposium



May-11 Duddeck 20

Linear-static

Nonlinearstatic

Nonlineardynamic

v0=40 m/s

0 50 100 150 2000

2

4

6

8

10

12

14

16

x 105

linear

nonlinear

dynamic

A Tovar & JE Renaud (2008): Altair HyperWorks Symposium



May-11 Duddeck 21

Hunkeler (PhD thesis), Duddeck (QMUL)


Topology Opt. for Crash (4): Graph-based Method

May-11 Duddeck 22

A Schumacher & C Ortmann (2011): CARHS Automotive Grand Challenge

• Setup of different designs with „library

parts“, e.g. beams, panels

• Each library part has a representation as

an attributed mathematical (sub-)graph

• Library parts can be connected to make

up structures

• Topological changes are applied as

changes to the design graph

• Graph based topology optimization is

understood as optimization of the design

graph topology

• Shape optimization is applied in an inner

loop in each topological iteration



May-11 Duddeck 23




May-11 Duddeck 24


quadrangular

panel

beam

triangular panel

coordinate vertex and

structural link


Duddeck 25

SFE CONCEPT for Shape and Topology Optimization

SFE CONCEPT: Data Flow

Parametric models revolutionize the VPD

SFE CONCEPT Parametric Model

Points and Lines

Cross Sections

Joints & Beams

Welds, Adhesives

FE Meshes

Freeform Surfaces

Loading, Etc.

Beads, Stamps, Ribs

May-11


May-11 Duddeck 26

SFE CONCEPT for Shape and Topology Optimization




May-11 Duddeck 27

Parameterization for Optimization (Two Types)



Maintaining geometrical and topological compatibility is

nearly impossible

Maintaining geometrical and topological compatibility is automatic and very easy

Challenges in optimization

Explicit parameterization(CAD)

Implicit parameterization(SFE CONCEPT)


May-11 Duddeck 28


Parametric models revolutionize the VPDChallenges in optimization

1.1)

1.2)

1.1)

2.)

• Multi-flange assignment• Joining assignment

• Reparametrising the map-targets• Identifying the flanges/joining

?

Withoutsurface map

Withsurface map

Following information is lost: Automatically done:

Parameterization for Optimization (Two Types)


May-11 Duddeck 29

Combined Shape and Topology Optimization



Objective function& constraints

Design variables

Optimizer(e.g. LS/OPT, in house)

SFE CONCEPT

Designvariables

FE mesh &Boundaryconditions

Batchmode

FE SOLVERLS-DYNA, NASTRAN,

PAM-CRASH, PERMASRADIOSS, ABAQUS

DV-file

ASCII output

Optimization loop

DV in ASCIIformat

Modified design

variables

Script for modelupdate & FE

mesh creation

FE mesh &boundary conditions

Problem definition

(only done once)

Design variables inoptimizer

format

ASCII output


Algorithmic Aspects

May-11 Duddeck 30

Hybrid and hierarchical methods for multi-level optimization


May-11 Duddeck 31

Combined Shape and Topology Optimization


Design SpacePackage Opt.

Topology Opt.Principle Load Paths

SFE CONCEPTGeometry Model

Shape & Topologychanges

Refinements


Optimization


Topology Optimization for Crash (ESLM)

May-11 Duddeck 32

• Equivalent static load method

(ESLM) and standard topology

optimization (linear FEM).

• Equivalent loads are defined for

different time steps based on

energy considerations.

• Result is embedded in an overall

loop “topology – shape – size”

optimization.

• Result corresponds well to

reference designs

• New design alternatives can be

identified K. Volz (PhD Dissertation)TU München & BMW (2011)

Package Design space

Optimisation model

Result (topol. opt.)

Transfer to shape opt.



May-11 Duddeck 33

K. Volz (Dissertation)TU München & BMW (2011)

First load phase Second load case Third load case



May-11 Duddeck 34



Combined Shape & Topology Opt. (SFE CONCEPT)

May-11 Duddeck 35



Conclusions

May-11 Duddeck 36

1. New approaches for topology optimizaton (crash):

• Equivalent Static Load Method (ESLM)

• Hybrid Cellular Automata (HCA)

• Graph-based method

2. New approaches for shape optimization (crash)

• Implicit parameterization (SFE CONCEPT)

• Direct parameterization (CAD-based)

• Morphing, etc.

3. Combined package / material / structure (shape &

topology) optimization required.


End.

May-11 Duddeck 37

Trains

Shipping

AerospaceWhite goods

Offshore

Front door

Lift gate

Radiatorsupport

Hood

Frontsub-frame

Automotive

SFE GmbH


Contact

May-11 Duddeck 38

Prof. Dr. Fabian DUDDECKFG Computational MechanicsTechnische Universität München

[email protected]

mailto:[email protected]

Documents

New Approaches for Shape & Topology Optimization for ...New Approaches for Shape & Topology Optimization for Crashworthiness keynote lecture ... State-of-the-Art in Shape & Topology