computer-aided design for additive manufacturing: … · quality control ship to mexico • slice...

Computer-Aided Design for Additive Manufacturing: Can We Exploit Shape and Material

Complexity Capabilities?

David W. Rosen

George W. Woodruff School of Mechanical EngineeringGeorgia Institute of Technology

Atlanta, GA [email protected] 404-894-9668

Georgia Tech, Atlanta, GA

Georgia Tech

Georgia Tech

• 20,000 undergraduate and graduate students

• 800+ faculty• >$300M in research funding• >$300M in research funding• Located in downtown Atlanta• GT was the Olympic Village

for 1996 Summer Games

Mechanical Engineering

• Includes ME, Nuclear Engr., Health Physics

• 85 faculty• 85 faculty• 1850 undergraduate students• 820 graduate students• 11 Research Groups (CAEDesign, Mfg,

Acoustics, MEMS, Bio, etc.)• Top 5 program

Overview• Additive Manufacturing

• Cellular Structures

– Construction

– Optimization

• Integrate Materials into Computer-Aided Design

– Simultaneous product-material-process design

– Process-structure-property relationships

• Exposure Controlled Projection Lithography

Additive Manufacturing• Class of manufacturing processes that build

parts one layer at a time.

• Stereolithography, Selective Laser Sintering, Fused Deposition Modeling, ...

AM Unique Capabilities

• Shape Complexity

• Material Complexity

• Hierarchical Complexity

• Functional Complexity

Hearing Aid Shells

CreateImpression

Scan Impression

Create CAD model

Assemble components

Fabricate shell

Invisalign Manufacturing Process

Timeline

Receiving & Dental Data Treat Aligner Order Entry Laboratory Acquisition Operations Fabrication

SLA MoldFabrication

Prepare & Slice

• STL File Generat ion• Packing (orient & place)• Support Generat ionPrepare & Slice

SLA Build

Quality Control

Ship to Mexico

• Support Generat ion• Slice Data

• Laser (Scanning)• Re-coat (dip in resin)• Sweep (remove excess

resin)• Part Removal & Post-

cure

• Verify test parts• Conduct individual mold

inspect ion

• Kit in transportation box



– Construction

– Optimization





Cellular StructuresWhen modern man builds large loadWhen modern man builds large load--bearing structures, he uses dense solids; bearing structures, he uses dense solids;

steel, concrete, glass. steel, concrete, glass. When nature does the same, she generally uses When nature does the same, she generally uses cellular materialscellular materials; wood, ; wood,

bone, coral. bone, coral. There must be a reason for it.There must be a reason for it.

-- Michael F. Ashby, Anthony Evans, et al.; Michael F. Ashby, Anthony Evans, et al.; Metal Foams: A Design GuideMetal Foams: A Design Guide

Bone Structure Human Skull

[Gibson, 2001]Linear cellular alloy parts [McDowell, 2004]

Design For Additive ManufacturingLow volume ratio structure

(Adaptive Material Distribution)

Multi-functionality:Structure + Heat Transfer

Multi-functionality: Structure + Acoustics

Combustor liner

Design and CAD Methods

00.5

1

00.5

1

00.5

1

0

0.5

1

0

0.5

00.5

1

0

0.5

1

0

0.5

CAD Representations

• Purely geometry in CAD systems• Boundary Representation solid modeling

– all geometric details are always represented– 1-2000 geometric entities is limit

• Parametric dimensions– adjustable geometry

• Complicated topology

Freedom of CreationLoughborough Univ.

Conformal Lattice Structures

16

18

18

Paramount Industries, Inc.©2009

0

2

40

5

6

8

10

12

14

Z

X

0 2 4024

6

8

10

12

14

16

UAV & Micro-UAV Examples

Application to UAV Fuselage

Size Matching & Scaling (SMS)

1 6

4 3

2

5

Greg Graf’s MS Thesis (April 2009)

Micro Air Vehicle: Problem Definition

TailFixed support

xy

FTailFMotor

F

19

Fpayload

F Motor(N) 5.9

F Tail (N) 2.7

F Payload (N/mm2) 0.1

Target Volume (mm3) 100,000

Total Unit-Cell Count 214

Optimized MAV Fuselage



– Construction

– Optimization





Structure-Property Relationship

Har

dnes

s

Distance from surface ( µµµµm)

Har

dnes

s

Low carbon steel Materials Chemistry and Physics, 112:1099-1105, 200 8.

Motivation

Macro-scale

Decomposition(lower resolution)

CAD system

Property

11 12 1

12 22 2

66 12

0

0

0 0 2

S S

S S

S

σε σ σ

γ

= =

S

Micro-scale

RadonTransform

WaveletTransform

MicrostructureImage

ComputationalMaterials Design

MethodsMicrostructureModel

• Fiber properties• Matrix properties

Material library

DATABASE

Process Design

Find: process variables

Satisfy: process constraints

Minimize: time, cost

Materials Design

Find: volume fractions, grain size, shape

Satisfy: compatibility constraints

Maximize: energy absorption, mech properties

Part/Product Design

Find: dimension values

Satisfy: stress, strain

Maximize: energy absorption

50

Design Problem Formulation

Multi-Objective Optimization Methods

Simultaneous Product / Material / Process Design Me thods

Common Computer-Aided Design Models for Parts and Microstructure

Encoding of microstructure using Surfacelet coefficients

“Zoom-in” and “Zoom-out” operations enabled by Surfacelet model

CAD methods enabled by geometric model based on wav elets and new surfacelets

100

150

200

250

Process ↔ Structure ↔ Property ↔ Performance

Dual-Rep Approach• Common geometric model for macro

geometry and microstructure.• Wavelets support multi-resolution

modeling. – Extend to multi-scale.– Represent distributions of material, Ω

δΩ

properties– inefficient in representing curve and

surface singularities.

• Surfacelet: proposed complement to wavelets for representing boundaries.

Ω

( ) ( )( )

− − ⋅ + ⋅ + ⋅ − = + − ⋅ + ⋅

1 2

2

1/ 2 1 1, , , , , 2

2

cos cos cos sin sin

sin cosa b r r

r x y z ba a

r x ya b

b a b a by y

a ar

Physical CaP-PHB

a) Radon transformθ

µ

b) inverse Radon transforma) Radon transformθ

µ

θ

µ

b) inverse Radon transform

Wavelet decomposition level 4Wavelet decomposition level 4

1 µµµµm1 µµµµm

120~125°135~140°80~95°45~50°

0~3° &

179~180°

θ

µ

c) inverse wavelet transform d) inverse Radon transform

Wavelet decomposition level 4

120~125°135~140°80~95°45~50°

0~3° &

179~180°120~125°135~140°120~125°135~140°80~95°80~95°45~50°45~50°

0~3° &

179~180°0~3° &

179~180°

θ

µ

θ

µ

c) inverse wavelet transform d) inverse Radon transform

Wavelet decomposition level 4

Three-CaP PHB fiber

• Compute Mechanical Property From microstructure– Resultant elastic modulus matrix*

3.93 0.793 0

– Rule-of mixture : Eeff =3.45 GPa– Inverse rule-of mixture: Eeff =1.14 Gpa

3.93 0.793 0

1 9 0.793 3.26 0

0 0 1.59

e

=

E

*Kalidindi, S.R. and J.R. Houskamp 2007

IN100 Example

0

2

4

6

0

2

4

6

0

2

4

6

8

10

z (µ

m)

0 10 20 30 40 50 60 70 800

10

20

30

40

50

60

70

80

1

2

3

4

5

6

7

8

9

10

11

12

13

14 1516

17

18

19

20

21

22

23

24

25

26 27

8

10

8

10x (µm)

y (µm)

0 10 20 30 40 50 60 70 80

a b

d

c

e

Computational Materials Design

LaserMelting

Figure courtesy Dr. Surya Kalidindi, Drexel Univers ity



– Construction

– Optimization





Exposure Controlled Projection Lithography

Laser Computer

Collimating

Imaging lens

Resin vat

Engineered diffuser

UV irradiation

Reaction

chamberCured

Part

Glass slide

x

z

DMD

Collimating lens

A more realistic model to predict the cured shape for a given exposure profile is required

Exposure (E) Cured height (z)

??

ECPL Process Overview

UV LED Source

Beam Conditioning System

Projection System

Resin Chamber

System

DMD

Mirror

Samples Fabricated with ECPL• Lenses ranging from 100µm to 10mm in a

variety of shapes, with sag heights ranging from 80µm to 300µm

July 3, 2012 33

Real Time Monitoring System –System Overview

Resin chamber

Detector

Laser

UV Source (365nm)

Beam

system

Beam conditioning

systemDMD™

Projection system

EExposure xposure CControlled ontrolled PProjection rojection LLithographyithography Patent Pending

Real Time Monitoring System (Discrete point)

Patent Pending

• Optical Path length, L = nt• Round trip optical path length:(4ngtg + 2nctc)• Change in optical thickness by photocuring, ∆L=2∆nctc

Working Principle

Detectorc c

• Phase shiftØ=2 π ∆L / λØ=4 π ∆nc tc / λ

July 3, 2012 36

Detector

Laser

Patent Pending

tc, nc

• n : refractive index• t : thickness tg, ng

Experimental Results

July 3, 2012 37

Snapshot of measuring the cured

part from confocal microscopeEstimated

HeightMeasured

Height

105 µm 100 µm

64 µm 66 µm

Conclusions• Cellular materials are basis for lightweight structure

design: repeated units arrayed along surfaces or fill volumes.

• TrussCreator NX: plug-in for commercial CAD that enables design, FEA, optimization of lattice structure.

• Take Advantage of AM geometric freedom.• Take Advantage of AM geometric freedom.

• Size Matching & Scaling optimization method is efficient (2 variables) and effective.

• Progress toward material (process-structure-property relationships) integrated into Computer-Aided Design systems.

Paramount Industries, Inc.©2009

Roadmap for Additive Manufacturing Workshop

March 2009, US NSF, ONR funded

http://www.wohlersassociates.com/roadmap2009.html

Thank you!Acknowledge:

RPMI Industry Members: 3D Systems, CIBA Vision, Ford, Pratt & Whitney, Paramount Industries

NSF IIS-0120663, DMI-0522382, CMMI-1030385

Post-Docs:Scott JohnstonFei Ding

Students:• Hongqing Wang (PhD, 3D Systems)• Benay Sager (PhD, McKinsey)Fei Ding

GT Collaborator:Yan Wang

404-894-9668 [email protected]

• Benay Sager (PhD, McKinsey)• Chris Williams (PhD, Virginia Tech)• Jane Chu (MS, Michelin)• Greg Graf (MS, Rockwell) • Sarah Engelbrecht (MS, Northrop-Grumman)• Namin Jeong, Patrick Chang, Jason Nguyen• Amit Jariwala

computer-aided design for additive manufacturing: … · quality control ship to mexico • slice...

Documents