digimat multi-scale modeling: the technology & software ... · digimat multi-scale modeling:...

www.e-Xstream.com

DIGIMAT Multi-Scale Modeling:

The Technology & Software Tools for a

Predictive Development of Reinforced

Plastic Parts

Roger A. Assaker

[email protected]

Skype: eX_RAR

Mobile: +32 495 52 56 52

www.e-Xstream.com

http://www.e-xstream.com/



mailto:[email protected]






The Problem: Fiber Orientation & the UnderlyingMaterial Behavior (Simplified)

Friday, August 01, 2008 Copyright© e-Xstream engineering SA, 2003-2007 2

«E»11 «E»22 «E»33

PA Matrix

E = 3150 Mpa

Nu = 0.36

Density = 1220 kg/m3

Glass Fibers

E = 71 200 Mpa

Nu = 0.22

Density = 2600 kg/m3

Weight fraction = 0.33

4,470 MPa 10,600 MPa 4,440 MPa 8,180 MPa 4,440 MPa 12,100 MPa

1

3

Courtesy of SOLVAY

Realistic FEA of the Reinforced Plastic Parts

Software Choice

Geometric nonlinearities (Large Deformations)

Contact

Implicit/Explicit integration

Optimal mesh refinement

Optimal element choice

1st/2nd order

Tet or Hex, Triangle or Quad

Material Reinforced Plastic

Anisotropic

Heterogeneous

Nonlinear

Rate-dependent

Damage

Fatigue

Failure

Etc.Friday, August 01, 2008 Copyright© e-Xstream engineering SA, 2003-2007 3

Which Material Model ?AMODEL AS-4133 - stress-strain curves at 23°C

dumbell specimen and test samples from mold II

0

20

40

60

80

100

120

140

160

180

200

220

240

0.0 0.2 0.4 0.6 0.8 1.0 1.2 1.4 1.6 1.8 2.0 2.2 2.4 2.6 2.8 3.0

strain (%)

str

ess (

MP

a)

dumbell

Mold II - along the flow

Mold II - across the flow

0

20

40

60

80

100

120

140

160

180

200

220

240

260

280

300

320

0,0% 0,5% 1,0% 1,5% 2,0% 2,5% 3,0% 3,5% 4,0% 4,5% 5,0% 5,5% 6,0% 6,5% 7,0% 7,5% 8,0%

Strain

Str

es

s (

N/m

m²)

Damage, 0°, StaticDamage, 0°, 1 s-1Damage, 0°, 100 s-1Damage, 45°, StaticDamage, 45°, 1 s-1Damage, 45°, 100 s-1

TECHNYL C218 V35 Black, 23°C, Eh0

Modeling Needs

How can we design the optimal material ?

What is the relation between the material microstructure (e.g. Fiber content,

length, orientation) and its final properties (e.g. Mechanical, Thermal, Electric,…) ?

How to select the optimal material and use its anisotropy for optimal structure performance ?

What is the link between the material and structure performance ?

How can we optimally process the material and structure ?

What is the relation between the process parameters and product performance ?

How can we achieve these objectives efficiently ?

Predict the composite properties (i.e. Anisotropic, nonlinear, time-dependent, …) as a function of its microstructure.

Predict the product properties as a function the local material microstructure, as induced by the processing conditions (e.g. injection molding, draping,…)


DIGIMAT, The nonlinear multi-scale material & structure

modeling platform

Copyright© e-Xstream engineering, 2007

DIGIMAT to CAE, Coupling Injection Molding to Realistic FEA


Fiber Length Distribution

0

10

20

30

40

50

60

70

80

90

100

25

75

125

175

225

275

325

375

425

475

525

575

625

675

725

775

825

875

925

975

1025

1075

1125

Length [µm]

Nu

mb

er

Phase Material & Composite Microstructure

-Fiber Shape

-Fiber Weight Fraction

-Fiber Length Distribution

Fiber Orientations

Material DesignProcess Design

Structural FEA

Structure Design

Courtesy of Trelleborg & Rhodia

The Multi-Scale Modeling Approach forFiber Reinforced Engineering Thermoplastic


MAP

DIGIMAT

Matrix Properties

Reinforcement Properties

Composite Morphology

Fiber Length/diameter

Fiber Weight/Volume Fraction

Composite

Properties

Structural

Mesh

Fiber

Orientation

Structural FEA

Injection

Mesh

Injection

Mat Prop.

Injection

Process Param.

Fiber

Orientation

Residual

Stresses

Residual

Temperature

Micro/macro

FEA results

Injection Molding

Software

Interaction between DIGIMAT and FEA


FE model level

Nodal coordinates, …

Strain increments,

material state, …

Element level

Material level

Stresses and

material stiffness

Internal forces and element stiffness

e

s

Classical FE process Coupled FE/DIGIMAT process

« In code » model

FE model level

Nodal coordinates, …

Strain increments,

material state, …

Element level

Stresses and

material stiffness

Internal forces and element stiffness

Material level

Mean Field Homogenization

Composite behavior depends explicitly on the:

Behavior of each phase

Inclusion shape (Aspect Ratio)

Inclusion orientation

Inclusion evolution (finite strain)

Friday, August 01, 2008 Copyright© e-Xstream engineering SA, 2003-2007

E Σ

Local phase behavior

(Step 2)

Global behavior

Localization

(Step1)

Averaging

(Step 3)

εr σr

EHx rrr :)(ee

rrr c es :

es :)( rcc

Pros

Fast model preparation/solution

Accurate predictions

Enables fully coupled nonlinear multi-scale Analyses

Cons

Ellipsoidal inclusions

Uniformly distributed inclusions

Average per phase (micro) results

a) b) Matrix

Fibers

ei

si

a) b)

DIGIMAT-MF: Major Capabilities

Materials: Per Phase (of a composite):

Thermo-Elastic: Anisotropic, Temperature dependent.

Elasto-Plastic: Small deformations/Large rotations

• Pressure dependent (Drucker-Prager)

• Continous Damage (4 parameters model)

Visco-elastic: Linear, small deformations/Large rotations

Elasto-Viscoplastic: Large deformations.

Hyperelastic (5 models): Large deformations

Micro-structure:

N-Phase (e.g. fibers + mineral inclusions, distribution of fiber “lengths”)

General Orientation (e.g. Orientation Tensor from MOLDFLOW)

Inclusion Coating (i.e. Fiber/Matrix Interface)

Voids

Micro & Macro Failure Indicators

1st & 2nd Order Incremental Homogenization Methods:

Mori-Tanaka

Interpolative Double Inclusion (High Concentrations/Contrast)

Thermo-Mechanical Static & Dynamic (Impact) Loading

Nonlinear, strongly coupled CAE Interfaces


10

s

1

E0

1

E = E0 (1–D)

e

MAP: 2D & 3D mapping

Mesh Types Mid-plane(Triangles)Shell (Triangle or Quad)

3D (Tet) 3D (Tet or Hex)

Mesh Format Abaqus

ANSYS

LS-DYNA

PAM-CRASH

Mapped Data: fiber Orientation

Initial Stresses

Initial Temperature

Mapping Error Indicators Global (Contour plot)

Local (Histogramme)

Model Scaling/Positioning

Data post-processing Synchronized

Contour or vector plotsFriday, August 01, 2008 Copyright© e-Xstream engineering SA, 2003-2007 11

SOME APPLICATIONS

Friday, 01 August 2008 Copyright© e-Xstream engineering SA, 2003-2006 13

FAT B85 : FE & Material Models (PP-LGF)

DIGIMAT Material Model

PP-Matrix :

• E= 1500 MPa

• = 0.3

Fibres :

• E = 72000 MPa

• = 0.22

• Volume Fraction = 19.46 % (40 % Weight Fraction)

• Aspect ratio : 100 (Long Fibers)

• Orientation : MOLDFLOW 5.1

ABAQUS FEA Model # Elements =12632 (S3R)

# Nodes = 6365

# DOF = 38190

Material : PP-LGF with DIGIMAT 1.6

Initial Stresses: MOLDFLOW 5.1 Courtesy of:

www.renault.com


FAT B85: Structural Stiffness

Fixed

P3 : 20daNP1: 20daN

P2: 20daN

Structural Stiffness (MDA-Test)/Test

P1 -3.75%

P2 +8.07%

P3 -6.97%Courtesy of:


FAT B85: Modal Response

Eigen Freq. (MDA-Test)/Test

7 1.68%

8 0.96%

9 -3.66%

10 -2.76%

11 -1.84%

12 -1.06%

13 -2.83%

14 -4.45%

15 1.46%

16 2.09%

17 0.47%

18 2.21% 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19

Mode Number

Ein

ge

n F

req

ue

nc

y [

Hz]

Test

MDA Predictions

Courtesy of:


3-Point Bending Beam: Problem Definition

ABAQUS FE Model Number of Elements=30,532 Number of Nodes= 31,528 Number of DOF=154,089

Material : PA 35% GF Matrix :

• Young’s modulus: 2.75 E+09 Pa• Poisson ratio : 0.37

• yield_stress = 4.0e+07 Pa

• hardening_model = exponential linear

• hardening_modulus = 3.67e+07

• hardening_exponent = 3.2e+02

• hardening_modulus2 = 3.0e+07• Density: 1.13 E+03 Kg/m3

Fibers :• Young’s modulus: 7.2e+10 Pa• Poisson ratio: 0.22• Density:2.47e+03 Kg/m3

• Weight fraction : 35 % • Aspect ratio :

– 25– Fiber length Distribution

• Fiber Orientation– Moldflow (DSM)

Axial Young's Modulus Vs Fiber Length

0

2000

4000

6000

8000

10000

12000

14000

16000

0 200 400 600 800 1000 1200

Fiber Length [µm]

Ax

ial

Yo

un

g's

Mo

du

lus

[M

Pa

]

Ln = 248

µm

Lw = 321 µm

ELw = 12147

ELn = 11379

EFLD = 10560

LFLD = 197

µm

Fiber Length Distribution

0

10

20

30

40

50

60

70

80

90

100

25

75

125

175

225

275

325

375

425

475

525

575

625

675

725

775

825

875

925

975

1025

1075

1125

Length [µm]

Nu

mb

er

Courtesy of: DSM


3-Point Bending Beam

Courtesy of: DSM

von Mises Stresses

In PAGF

Accumulated Plastic Strain

In PA matrix


Reaction Force / Intrusion

-11000

-9000

-7000

-5000

-3000

-1000

1000

0 1 2 3 4 5 6 7 8 9 10

Intrusion (mm)

Re

ac

tio

n F

orc

e (

N)

DIGIMAT ar=16 ElastoPlastic NLGEOM

DIGIMAT ar=16 Elastic NLGEOM

SHF ar=16 NLGEOM

SHF ar=25 NLGEOM

DIGIMAT AR=16 ElastoPlastic NLGEOM

(Quad mesh)Experience

3-Point Bending Beam: F/d Curves

-35%

Target

Measured RF= 6,883 N

Predicted RF (Tri mesh) = 7,170 N (+4%)

Predicted RF (Quad mesh) = 6720 N (-2.4%)Courtesy of: DSM

Copyright© e-Xstream engineering SA, 2003-2008 19

Multi-Scale Modeling of Passenger Airbag Container

Moldflow’s Injection Molding Mesh: Number of nodes: 584,123

Number of elements: 3,369,976

Element type: C3D4

Abaqus Structural Mesh: Number of nodes: 368,852

Number of elements: 194,794

Element type: C3D10, C3D10M

Material: AKULON K224-PG8 (40% Glass filled

Impact Modied Polyamide)

Matrix: Impact Modified Polyamide• type = elastoplastic

• Young Modulus = 2350 MPa

• Poisson Ratio = 0.38

• Yield stress =30 MPa

Fibers: E-Glass • Type = elastic

• Density = 2.54 E+3

• Young Modulus = 72 000 MPa

• Poisson Ratio = 0.22

• Weight fraction = 40%

• Aspect ratio (L/D) = 20

• Orientation = Moldflow3D (.xml)

Courtesy of: AUTOLIV & DSM

DIGIMAT to Abaqus Nonlinear Multi-ScaleAnalyses: Results & CPU

Loadings Linear Cyclic

CPU Time 82 h 05 m 41 h 13 m

Elapsed time 25 h 56 h 28 m

# CPUs(Opteron 64 bits)

4 1


RF @ Imposed D ExperimentalForce

DIGIMAT to Abaqus Difference

Linear(to 10.5mm)

~ 6477 N 6203.49 N -4.2%

Cyclic(to 7mm)

~ 4765N 3949.18 N -17 %

Courtesy of: AUTOLIV & DSM

Multi-Scale Impact & Failure Simulation

LS-DYNA FE Mesh

3,630 Elements

3,646 Nodes

16 composite shell sections with different thicknesses.

5 integration points across the thickness

Characteristic dimension : 400x50x30 mm.

Loading & BC :

Simply supported beam

Initial velocity: V0 =-5 m/sec


Fixed

Initial velocity

FixedCourtesy of :

Impact & Failure Analysis: Aligned Fibers


Failure Prediction: Effect of Injection Gate Location


Courtesy of :

DIGIMAT to PAM/CRASH: Impact & Failure


Conclusions

Mean Field homogenization is a powerful technology to predict

the anisotropic, nonlinear and rate-dependent behavior of

mutli-phase materials in general and of reinforced plastics in

particular.

Fully coupled multi-scale modeling bridges the gap between

the manufacturing process and the final part performance via

the material microstructure.

DIGIMAT offer the tools and modeling process for the optimal

and predictive design of reinforced plastic parts.

digimat multi-scale modeling: the technology & software ... · digimat multi-scale modeling:...

Documents