An Efficient Methodology for Fracture Characterization and ...

GDIS2018

An Efficient Methodology for Fracture Characterization and Prediction of DP980 Steels for Crash Application

University of Waterloo: Honda Research America:

Cliff Butcher Jim Dykeman, Skye Malcolm

#GDIS | #SteelMatters 2

Project Team

Steel Marketing Development: Hesham Ezzat, Dave Anderson ArcelorMittal: Steve Lynes, Tim Lim AK Steel: Kavesary Raghavan Nucor: Dean Kanelos, Andy Thompson Honda Research of Americas: Jim Dykeman, Skye Malcolm

University of Waterloo:

PI’s: Cliff Butcher and Mike Worswick Research Team : Jose Imbert-Boyd

Armin Abedini Kenneth Cheong Sante DiCecco Sam Kim Amir Zhumagulov Taamjeed Rahmaan Kaab Omer


Project Structure 1. Each Supplier Submits One DP980 to SMDI Sample Bank

2. Material identification removed:

Sent to UW and Honda for Fracture Characterization from Coupons to Crash


Study Materials: Composition 980 #1

SMDI Grades Thickness (mm) Type General Composition Coating

980 Mat#1 1.2 Dual Phase C-Mn-Si Bare

980 Mat#1 1.6 Dual Phase C-Mn-Cr-Mo-Nb Zinc

980 Mat#3 1.4 Dual Phase C-Mn-Si Zinc

980 #2 980 #3

Grades represent recent optimization in processing / chemistry (but are not Gen 3 level)

Materials can generally be described as DP with fine, uniform microstructure.


0

10

20

30

40

50

60

70

80

90

100

110

120

DP980 -

Material 1

DP980 -

Material 2

DP980 -

Material 3

JAC 980Y

Ben

d A

ng

le ( )

-V

DA

Met

ho

d

DP980 Material Properties: Tensile & VDA Bend Performance

Performance of these grades is consistent with or above current commercial products

Better local formability relative to other DP980’s

SMDI DP980’s

0

100

200

300

400

500

600

700

800

900

1000

1100

1200

0.00 0.02 0.04 0.06 0.08 0.10 0.12 0.14 0.16

En

gin

eeri

ng

Str

ess

(MP

a)

Engineering Strain

SMDI MAT#1

SMDI MAT#2

SMDI MAT#3


Project Goals

1. Characterize properties of various Dual Phase 980 grades selected by Steel

Marketing Development Institute (Blind Study)

2. Investigate optimized fracture testing methodology for Advanced High Strength

Steel Industrial Friendly and Efficient Methods Required (GDIS 2017)

3. Perform experimental axial and bend crush experiments and assess

fracture performance (GDIS 2017)

4. Numerical characterization for CAE application to dynamic tests (GDIS 2018)

Efficient methods needed to transition from coupons to crash simulations


Recap: Material Characterization at Large Strains

• Limited hardening data

available in tensile tests

• Inverse FE modeling used to

identify hardening at large

strains for fracture

• Hardening data becomes a

function of numerical model

assumptions... 0

200

400

600

800

1000

1200

1400

1600

0.00 0.10 0.20 0.30 0.40 0.50 0.60 0.70 0.80

Eq

uiv

ale

nt

Str

ess

(M

Pa)

Equivalent Plastic Strain

Experiment: Tension

DP980 - Material 1Transverse Direction

What is stress

level at fracture?


0

200

400

600

800

1000

1200

1400

1600

0.00 0.10 0.20 0.30 0.40 0.50 0.60 0.70 0.80

Eq

uiv

ale

nt

Str

ess

(M

Pa

)Equivalent Plastic Strain

Experiment: Converted Shear

Experiment: Tension

Holloman Model

DP980 - Material 1Transverse Direction

0.074

(MPa) 1392 p

Recap: Material Characterization at Large Strains

• UW developed simple method to

use tensile & shear test data to

obtain hardening to large strain

levels

• DP980 data to 60% strain!

• Not related to FE model

Methodology in Rahmaan, T., et al.,

Int. J. Impact. Eng., 2017 (in-press)


800

900

1000

1100

1200

1300

1400

0 0.02 0.04 0.06 0.08 0.1E

qu

ivale

nt

Str

ess

(M

Pa)

Eq. Plastic Strain

Exp. 0.001 s-1 Exp. 10 s-1 Exp. 100 s-1

Exp. 1000 s-1 Model: 0.001 s-1 Model: 0.10 s-1

Model: 1 s-1 Model: 10 s-1 Model: 100 s-1

Model: 1000 s-1

1000 s-1

0.001 s-1

DP980 - Material 1: Transverse Direction

Material Characterization at Large Strains and Strain Rates

• Tensile characterization from

0.001 to 1000 s-1

• Scale quasi-static data obtained

to large strains for strain rates

• Efficient experimental method for

constitutive characterization

,p p QS pF


Fracture Characterization Results

0

0.1

0.2

0.3

0.4

0.5

0.6

0.7

0.8

0.9

1

-0.7 -0.5 -0.3 -0.1 0.1 0.3 0.5

Ma

jor

Str

ain

Minor Strain

JIS

Mini-shear Smooth

Notch

Tight

Notch

Uniaxial

Dome

Plane Strain

Dome

Biaxial Dome

Plane Strain

Notch

V-bend (Plane strain) ( 0, 0.45 )

Hole Expansion (Uniaxial tension)

( -0.44, 0.87)

• Conflicting limits provided by different specimen types if thinning correction not applied

Min. of 4 Tests can describe

the fracture locus

Four Relatively Simple Tests:

1. Mini-shear

2. Hole expansion (reamed)

3. V-Bend

4. Biaxial/Bulge


Experimental Fracture Loci: 3 DP980’s

Four tests can be used to generate physically-meaningful fracture loci

Not the product of a simulation exercise – Real material performance can be assessed

• Relatively comparable fracture loci

• Mat 2 had the lowest hardening rate,

highest hole expansion and v-bend

• How do we use this for CAE?

0.00

0.20

0.40

0.60

0.80

1.00

1.20

-0.10 0.00 0.10 0.20 0.30 0.40 0.50 0.60 0.70

Eq

. P

last

ic S

train

Stress Triaxiality

Experimental Fracture Loci: UW ModelDP980 - Transverse Direction

Material 2

Material 1

Material 3


Hybrid Experimental-Numerical Approach to Fracture

1. Perform set of characterization tests

Shear Tensile Central Hole Notch Tension

2. FE modelling of characterization tests

3. Extract Local Stress Histories from models 4. Damage Model for CAE

- Assume damage model, failure locus form & optimize

p

f

dD

T

,MMC

f iT a

Optimization


For industry, there are more steps remaining…

2. FE modelling of characterization tests

3. Extract Local Stress Histories from models 4. Damage Model for CAE

- Assume damage model, failure locus form & optimize

p

f

dD

T

,MMC

f iT a

Optimization

Repeat Steps 2-4 for

Every Mesh Size of Interest*

*Solid element regularization will be different than shell for shells

Eventually…apply to a practical application not

used in the calibration to see if it works

Only works for 1 mesh size


Choice of CAE Characterization Tests

1. Perform set of characterization tests

Shear Tensile Central Hole Notch Tension

Tensile-Based Characterization Tests are Employed

X – Strong localization

X – Through-Thickness Strain Gradients

X – Fractures at mid-thickness

No DIC strain measurement

X – Requires 3-D solid elements

X – Requires fine mesh: ~ 0.10 mm

X – Non-linear 3-D stress state develops

CAE models for forming & crash use plane stress shell elements from 0.5 – 7.0 mm

Solid element models are great for academic research but less so for industry

Extracting the plane stress fracture locus from a calibrated 3-D solid model

works in theory…in practice the element mechanics are different


Tensile-Based Fracture Characterization

Plane Strain Notch Intermediate Notch Miniature Uniaxial

Increasing

Triaxiality

Large Notch

Relatively simple tests that most labs can perform and are comfortable with

Since sheet is thin, the logic is that these samples are plane stress….

Deformation rapidly localizes, violating plane stress assumption but creating a desired change in the stress state

Representative Types of Tensile-Based Fracture Coupons


Comparison of Solid & Shell Models for Tensile CAE Coupons

Shell models cannot resolve strong local thinning and localization Overestimate the stress response, underestimates strain

Methods exist to add damage-induced softening to improve the shell solutions. Not a damage issue but element type.

Can create problems for cases when shells are appropriate

0

100

200

300

400

500

600

700

800

900

0.00 0.05 0.10 0.15 0.20 0.25

En

gin

eeri

ng S

tres

s (M

Pa)

Engineering Strain

USIBOR® 1500-AS: BainiteMiniature Tensile GeometryMesh Size: 0.20 mm

Solid Elements:0.20 mm mesh

Shell Elements: 0.20 mm mesh

Experiment

Solid Elements: 0.10 mm mesh


“Plane Stress Friendly” Characterization

Shell element models for sheet metal forming and structural component models can be very accurate

Use of Nakazima dome tests for CAE characterization is more consistent with the end-use applications

Plane Strain Nakazima Dome Test

0

10

20

30

40

50

60

0 5 10 15 20 25Fo

rce

[kN

]Dome Height [mm]

Shell FE Model

Solid FE Model

Shell FE model with flat hardening curve

Experiment

Fracture: Experiment

USIBOR® 1500-AS: Bainite75 mm Nakazima Dome TestMesh Size: 0.20 mm

Solid Element Model

Shell Element Model

Shell Element Model with flat hardening curve

Shell elements are

appropriate


Plane Stress Representation of the Shear Test

0

200

400

600

800

1000

0.00 0.10 0.20 0.30 0.40 0.50 0.60 0.70 0.80

Sh

ear S

tress

(M

Pa)

von Mises Eq. Strain

Experiment Average

Mesh Size: 0.1 mm

DP980 - Material 1: 45 Shear Test

Mechanics of shear deformation creates a Plane Stress-Plane Strain loading condition

Shell elements provide an accurate description

0

100

200

300

400

500

600

700

0.00 0.10 0.20 0.30 0.40 0.50 0.60 0.70 0.80 0.90

Sh

ear

Str

ess

(MP

a)

Equivalent Plastic Strain

USIBOR® 1500-AS: BainiteMiniature Shear GeometryMesh Size: 0.20 mm

Solid Elements

Shell Elements

Comparison of FE models for a tailor hot stamped steel Exp. & FE comparison of shear test for a DP980 steel


100% strain

T ~ 0.70

15% strain

T ~ 0.33

An Industrial-Focused CAE Strategy is Required

Solid Model

(0.1 mm)

Shell Model

(0.1 mm)

Tensile-based strategy requires solid models: Academic-focused

Industrial-focused strategy must be tailored for shell elements

We have identified exp. tests with minimal necking where shell

elements can be used Consistent with CAE applications

JIS Tensile

DP980: t = 1.2mm


Industrial Strategy for Plane Stress Fracture Characterization

1. Perform 4 Plane Stress Characterization Tests

Shear Hole

Expansion V-Bend Biaxial Dome

3. Plane Stress Models with Various Mesh Sizes 4. Regularize Exp. Fracture Locus for CAE

- Assume damage model & scale locus with mesh size, R:

p

f

dD

T

exp exp, ,CAE

f f iR R T a

2. Experimental Fracture Locus

- Assume failure locus form & calibrate with 4 points

exp

1 4,UW

f T a

Physically meaningful

fracture locus: Not an FE construct

Efficient

Length

Scale

Transition

Knowledge

Gap


“Direct” Strategy for Numerical Fracture Characterization

3. Plane Stress Models & Regularization

2. Assume a damage form for nonlinear loading

exp

pJC

f

DT

Non-Proportional

Loading

1D

11

exp

GISSMO pn

f

nD D

T

0D

Examples:

1. Start with the Experimental Fracture Locus


exp

1 4,UW

f T a

4. Apply to Structural Components

3-Point Bend & Axial Crush of a 600 mm Rail:

2.5 mm element mesh

Static & Dynamic Conditions

What is most efficient method to regularize for components?


Length-Scale Transition from Coupons to Components: Regularization

Large elements cannot resolve the stress and strain as accurately

Local strain at fracture generally decreases with element size

0.00

0.20

0.40

0.60

0.80

1.00

1.20

-0.20 0.00 0.20 0.40 0.60E

q. P

last

ic S

train

Stress Triaxiality

Experimental or Reference Fracture Locus

Increasing Mesh Size

Increasing Mesh Size


Regularization Example: Biaxial Dome Test with 2.5 mm mesh

¼ FE Model of Biaxial Dome Test

Decent load-displacement agreement

for isotropic plasticity model

DP980: t = 1.2 mm


Regularization Effectively Alters the “Damage” Accumulation Rate

to Trigger Fracture at the Exp. Dome Height

0.0

0.1

0.2

0.3

0.4

0.5

0.6

0.7

0.8

0.9

1.0

0 5 10 15 20 25 30 35 40 45

Dam

age

Dome Height (mm)

Scale

Factor,

R

No Regularization Regularized

Failure @

Exp. Disp

Element deletion

exp exp, ,CAE

f f iR R T a


However, Regularization is Not That Simple…

1. Coupon geometry

2. Element type: some geometries are poorly described by shells

3. Deformation mode: Bending mode is not well described by large elements relative to stretching mode

4. Stress State: Uniaxial tension is different than biaxial tension

Regularization factor depends upon:

Regularization atones for any experimental and modelling sins

Issues of modelling taste Different fracture methodologies can lead to similar

results in component tests after each is regularized…


4 Geometries for Regularization: 0.1 to 5 mm mesh

1. JIS Tensile

(T = 0.33)

2. Uniaxial Dome

(T = 0.33)

Plane Strain Dome (T = 0.58)

Equal Biaxial Dome (T = 0.66)


0.0

0.1

0.2

0.3

0.4

0.5

0.6

0.7

0.8

0.9

1.0

0 1 2 3 4 5 6

Mesh

Regu

lariz

ati

on

Facto

r

Shell Element Size (mm)

Nakazima Dome: Equal-Biaxial Tension

Nakazima Dome: Uniaxial Tension

Nakazima Dome: Plane Strain Tension

JIS Tensile

DP980: Material 1Shell Element: Type 16 with 7 Integration PointsDamage Exponent: 2

Mesh Regularization for DP980: t = 1.2 mm

Scale factors vary with element size

as the local stress state and strain also

change

Relatively constant factors for elements

larger than thickness

Regularization ranking:

1. Biaxial requires least regularization

2. Uniaxial dome

3. JIS & PS Dome are similar & lowest

Same trends for other 2 DP980’s


Regularization Results: 2.5 mm Mesh*

1. Plane Strain Dome ≈ JIS Tensile (severe regularization – lower bound for CAE)

2. Equal Biaxial Dome – Least amount of regularization (upper bound for CAE)

3. Uniaxial Dome – Similar but a bit higher than PS Dome and JIS

*Component models built

with a 2.5 mm mesh size


Regularization Results: 2.5 mm Mesh*

Expect Choice of Regularization Factor to be Application Dependent 1. For Moderately Uniform Sheet Stretching PS Dome/JIS/UT Dome

2. For Bending, Folding, Crash Nakazima Biaxial Dome

*Component models built

with a 2.5 mm mesh size


Stress-State Dependence: Create a Regularization Locus

Calibrate a new failure locus with stress-state regularized values to obtain a regularization locus

0.00

0.20

0.40

0.60

0.80

1.00

1.20

-0.10 0.00 0.10 0.20 0.30 0.40 0.50 0.60 0.70

Eq

. P

last

ic S

train

Stress Triaxiality

Experimental Fracture Loci: UW ModelDP980 - Material 1: Transverse Direction

JIS Tensile Regularization: 2.5 mm

Experimental

Biaxial Dome Regularization: 2.5 mm

PS Dome ≈ JIS Tensile

EB Dome

JIS


Regularized Fracture Loci for Component Tests

3 Strategies for Component Simulations

1. JIS/PS Dome Regularization

2. Biaxial Dome Regularization

3. Regularization Locus

0.00

0.10

0.20

0.30

0.40

0.50

0.60

-0.10 0.00 0.10 0.20 0.30 0.40 0.50 0.60 0.70

Eq

. P

last

ic S

tra

in

Stress Triaxiality



Regularization Locus: 2.5 mm


3-pt Bend Model (half symmetry)

Axial Crush Model

(full symmetry)


0

5

10

15

20

25

30

0 0.02 0.04 0.06 0.08 0.1

Forc

e (

kN)

Sled Displacement (m)

DP980Thickness: 1. 2mm3-pt - Dynamic - 7.8

Experiment

Dynamic 3pt Bend of DP980: Force-Disp.

JIS/PS Dome Regularization is too aggressive: Premature failure

EB Dome and Reg. Locus performed well: Local bending & folding

980 #1

JIS Tensile

Experiment

0

5

10

15

20

25

30

0 0.02 0.04 0.06 0.08 0.1

Forc

e (

kN)



Biaxial

JIS Tensile

0

5

10

15

20

25

30

0 0.02 0.04 0.06 0.08 0.1

Forc

e (

kN)



Reg. Locus

JIS Tensile

Biaxial


Dynamic Axial Crush of DP980: Force-Disp.

JIS/PS Dome Regularization is too aggressive:

Near Instant failure that changes folding mode and leads to the high peak force

EB Dome and Reg. Locus perform well: Local bending & folding

0

100

200

300

400

500

600

0.00 0.01 0.02 0.03 0.04 0.05 0.06 0.07 0.08 0.09 0.10

Forc

e (

kN)


DP980Thickness - 1.2 mmDynamic Axial Crush - 7.8 m/s

Experiment JIS/PS Dome

0

100

200

300

400

500

600

0.00 0.01 0.02 0.03 0.04 0.05 0.06 0.07 0.08 0.09 0.10

Forc

e (

kN)



Experiment

Biaxial

JIS/PS Dome

0

100

200

300

400

500

600

0.00 0.01 0.02 0.03 0.04 0.05 0.06 0.07 0.08 0.09 0.10

Forc

e (

kN)



Experiment

Biaxial

JIS/PS Dome

Reg. Locus


Regularization for CAE Summary

Biaxial dome regularization can rapidly convert the exp. fracture locus to a CAE model for

components

- shells are valid for biaxial dome simulation

- can use large mesh sizes

- simple test & model

Regularization locus will give best performance

in both coupon and component simulations

- much more effort required to develop

0.00

0.10

0.20

0.30

0.40

0.50

0.60

-0.10 0.00 0.10 0.20 0.30 0.40 0.50 0.60 0.70

Eq

. P

last

ic S

tra

in

Stress Triaxiality



Regularization Locus: 2.5 mm


Select biaxial dome regularization &

apply to other component simulations


Selected Component Simulations

3-pt Bend Model (half symmetry)

Axial Crush

Model

Quasi-Static & Dynamic Loading Cases:

1. No Damage Model

2. Damage Model without Regularization

3. Damage Model with Biaxial Regularization

Dynamic Axial Crush

Dynamic 3-Point Bend


3-Pt Bend: Quasi-static & Dynamic: Material 1 (1.2 mm)

Dynamic 3pt Bend – 7.8 mm/s Quasi-static 3pt Bend – 0.5 mm/s

Experimental Fracture Locus + Biaxial Regularization can give good CAE predictions in 3-Pt Bend

Force vs. Displacement Results

0

5

10

15

20

25

30

0 0.01 0.02 0.03 0.04 0.05 0.06 0.07 0.08 0.09 0.1

Forc

e (

kN)


DP980Thickness: 1. 2mm3-pt: Dynamic - 7.8 m/s

No damage model

Damage: No reg.

Damage: Biaxial reg.

0

5

10

15

20

25

0 0.01 0.02 0.03 0.04 0.05

Forc

e (

kN)

Punch Displacement (m)

DP980Thickness: 1. 2mm3-pt: QS - 0.5 mm/s

No damage model

Damage: No reg.

Damage: Biaxial reg.


3-Pt Bend: Quasi-static & Dynamic: Material 1 (1.2 mm)

Dynamic 3pt Bend – 7.8 mm/s Quasi-static 3pt Bend – 0.5 mm/s

Experimental Fracture Locus + Biaxial Regularization can give good CAE predictions in 3-Pt Bend

Energy Absorption


Similar Results for Other DP980’s

DP980 – Material 2

Quasi-static 3pt Bend – 0.5 mm/s

DP980 – Material 3


Comparison of Quasi-Static (QS) and Dynamic Axial Crush

Quasi-Static Dynamic

Quasi-Static Dynamic

Numerical deformation modes are markedly different in QS and Dynamic crush: Strain-rate & inertial effects

Added a force-based spot weld failure criterion


QS Axial Crush: Material 1

Progressive Folding

(Damage without Reg.)

Progressive Folding

(No Damage Model)

Progressive Folding:

3 SW failures

(Damage + Reg. + SW Failure)

Force-Displacement

0

50

100

150

200

250

300

350

400

0 0.02 0.04 0.06 0.08 0.1 0.12 0.14

Forc

e (

kN)

Displacement (m)

DP980: t = 1.2 mmAxial Crush: QS @ 0.5 mm/s

Experiment

0

50

100

150

200

250

300

350

400

0 0.02 0.04 0.06 0.08 0.1 0.12 0.14

Forc

e (

kN)

Displacement (m)

DP980: t = 1.2 mmAxial Crush: QS @ 0.5 mm/s

No Damage Model

Damage. No Reg.

With Biaxial Reg

& Spot Weld failure


Dynamic Axial Crush: Material 1

Mixed Buckling/Folding

(Damage without Reg.)

Mixed Buckling/

Folding

(No Damage Model)

Progressive Folding:

1 SW failure

(Damage + Reg. + SW Failure)

Force-Displacement


Conclusion: “Direct” Strategy for Numerical Fracture Characterization Established

3. Plane Stress Models & Regularization

2. Assume a damage form for nonlinear loading

exp

pJC

f

DT

Non-Proportional

Loading

1D

11

exp

GISSMO pn

f

nD D

T

0D

Examples:

1. Start with the Experimental Fracture Locus


exp

1 4,UW

f T a

4. Apply to Structural Components

3-Point Bend & Axial Crush of a 600 mm Rail:

2.5 mm element mesh

Static & Dynamic Conditions


Outlook & Future Work

The results are promising but much work remains:

• Application to sheet metal forming with severe non-proportional loading

• Application to sheet metal forming through to crash of an AHSS component

• Spot weld failure and potential un-zipping of weld groups

• Improve physics of damage model

• Need some physics to help guide regularization

Have developed an industrially-focused methodology for efficient fracture

characterization


Acknowledgements


For More Information

Cliff Butcher

University of Waterloo

[email protected]

Jim Dykeman

Honda Research America

[email protected]

mailto:[email protected]

An Efficient Methodology for Fracture Characterization and ...

Documents

Transcript of An Efficient Methodology for Fracture Characterization and ...