An Efficient Methodology for Fracture Characterization and Prediction of DP980 Steels for Crash Application GDIS2018

Transcription

1 An Efficient Methodology for Fracture Characterization and Prediction of DP980 Steels for Crash Application University of Waterloo: Cliff Butcher Honda Research America: Jim Dykeman, Skye Malcolm GDIS2018

2 Project Team Steel Marketing Development: ArcelorMittal: AK Steel: Nucor: Honda Research of Americas: Hesham Ezzat, Dave Anderson Steve Lynes, Tim Lim Kavesary Raghavan Dean Kanelos, Andy Thompson Jim Dykeman, Skye Malcolm University of Waterloo: 2 PI s: Research Team : Cliff Butcher and Mike Worswick Jose Imbert-Boyd Armin Abedini Kenneth Cheong Sante DiCecco Sam Kim Amir Zhumagulov Taamjeed Rahmaan Kaab Omer

3 1. Each Supplier Submits One DP980 to SMDI Sample Bank Project Structure 2. Material identification removed: Sent to UW and Honda for Fracture Characterization from Coupons to Crash 3

4 Study Materials: Composition 980 #1 980 #2 980 #3 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 Mat#3 1.4 Dual Phase C-Mn-Si Zinc Materials can generally be described as DP with fine, uniform microstructure. Grades represent recent optimization in processing / chemistry (but are not Gen 3 level)

5 Engineering Stress (MPa) Bend Angle ( ) - VDA Method DP980 Material Properties: Tensile & VDA Bend Performance SMDI DP980 s SMDI MAT#1 500 SMDI MAT#2 SMDI MAT# Engineering Strain DP980 - Material 1 DP980 - Material 2 DP980 - Material 3 JAC 980Y Performance of these grades is consistent with or above current commercial products Better local formability relative to other DP980 s 5

6 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 6

7 Equivalent Stress (MPa) 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 What is stress level at fracture? Experiment: Tension 400 Hardening data becomes a function of numerical model 200 assumptions... 0 DP980 - Material 1 Transverse Direction Equivalent Plastic Strain 7

8 Equivalent Stress (MPa) 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 Experiment: Converted Shear Experiment: Tension Holloman Model p (MPa) Methodology in Rahmaan, T., et al., Int. J. Impact. Eng., 2017 (in-press) 200 DP980 - Material 1 Transverse Direction Equivalent Plastic Strain 8

9 Equivalent Stress (MPa) Material Characterization at Large Strains and Strain Rates Tensile characterization from to 1000 s s -1 Scale quasi-static data obtained to large strains for strain rates p, p QS F p Efficient experimental method for constitutive characterization Exp s-1 Exp. 10 s-1 Exp. 100 s-1 Exp s-1 Model: s-1 Model: 0.10 s-1 Model: 1 s-1 Model: 10 s-1 Model: 100 s-1 Model: 1000 s Eq. Plastic Strain s -1 DP980 - Material 1 : Transverse Direction 9

10 Major Strain Fracture Characterization Results Conflicting limits provided by different specimen types if thinning correction not applied Hole Expansion (Uniaxial tension) ( -0.44, 0.87) Min. of 4 Tests can describe the fracture locus Mini-shear JIS Uniaxial Dome Smooth Notch Tight Notch V-bend (Plane strain) ( 0, 0.45 ) Biaxial Dome Four Relatively Simple Tests: 1. Mini-shear 2. Hole expansion (reamed) 3. V-Bend 4. Biaxial/Bulge Plane Strain Notch 0.3 Plane Strain Dome Minor Strain

11 Eq. Plastic Strain 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 Material Material 1 Relatively comparable fracture loci Material 3 Mat 2 had the lowest hardening rate, highest hole expansion and v-bend How do we use this for CAE? 11 Experimental Fracture Loci: UW Model DP980 - Transverse Direction Stress Triaxiality

12 Hybrid Experimental-Numerical Approach to Fracture 1. Perform set of characterization tests 2. FE modelling of characterization tests Shear Tensile Central Hole Notch Tension 3. Extract Local Stress Histories from models 4. Damage Model for CAE D - Assume damage model, failure locus form & optimize p d MMC f Ta, i T f Optimization 12

13 For industry, there are more steps remaining Repeat Steps 2-4 for Every Mesh Size of Interest* 2. FE modelling of characterization tests Eventually apply to a practical application not used in the calibration to see if it works *Solid element regularization will be different than shell for shells 3. Extract Local Stress Histories from models 4. Damage Model for CAE D - Assume damage model, failure locus form & optimize p d MMC f Ta, i T f Optimization Only works for 1 mesh size 13

14 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 Solid element models are great for academic research but less so for industry CAE models for forming & crash use plane stress shell elements from mm Extracting the plane stress fracture locus from a calibrated 3-D solid model works in theory in practice the element mechanics are different 14

15 Tensile-Based Fracture Characterization 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 Increasing Triaxiality Miniature Uniaxial Large Notch Intermediate Notch Plane Strain Notch 15 Representative Types of Tensile-Based Fracture Coupons

16 Engineering Stress (MPa) 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 Solid Elements: 0.10 mm mesh Shell Elements: 0.20 mm mesh Solid Elements: 0.20 mm mesh Experiment USIBOR 1500-AS: Bainite 100 Miniature Tensile Geometry Mesh Size: 0.20 mm Engineering Strain 16

17 Force [kn] 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 Shell FE Model Solid FE Model Shell FE model with flat hardening curve Experiment Fracture: Experiment Shell Element Model Solid Element Model Shell Element Model with flat hardening curve 20 Shell elements are appropriate Plane Strain Nakazima Dome Test 10 0 USIBOR 1500-AS: Bainite 75 mm Nakazima Dome Test Mesh Size: 0.20 mm Dome Height [mm] 17

18 Shear Stress (MPa) Shear Stress (MPa) Plane Stress Representation of the Shear Test Mechanics of shear deformation creates a Plane Stress-Plane Strain loading condition Shell elements provide an accurate description DP980 - Material 1: 45 Shear Test Solid Elements Shell Elements Equivalent Plastic Strain USIBOR 1500-AS: Bainite Miniature Shear Geometry Mesh Size: 0.20 mm Comparison of FE models for a tailor hot stamped steel 200 Experiment Average Mesh Size: 0.1 mm von Mises Eq. Strain Exp. & FE comparison of shear test for a DP980 steel

19 An Industrial-Focused CAE Strategy is Required Tensile-based strategy requires solid models: Academic-focused Industrial-focused strategy must be tailored for shell elements JIS Tensile DP980: t = 1.2mm 100% strain T ~ % strain T ~ Solid Model (0.1 mm) Shell Model (0.1 mm) We have identified exp. tests with minimal necking where shell elements can be used Consistent with CAE applications

20 Industrial Strategy for Plane Stress Fracture Characterization 1. Perform 4 Plane Stress Characterization Tests 2. Experimental Fracture Locus - Assume failure locus form & calibrate with 4 points exp UW f Ta, 14 Shear Hole Expansion V-Bend Biaxial Dome Physically meaningful fracture locus: Not an FE construct Knowledge Gap 3. Plane Stress Models with Various Mesh Sizes 4. Regularize Exp. Fracture Locus for CAE - Assume damage model & scale locus with mesh size, R: Efficient Length Scale Transition p d D T f exp exp R, R T, a CAE f f i 20

21 Direct Strategy for Numerical Fracture Characterization 1. Start with the Experimental Fracture Locus - Assume failure locus form & calibrate with 4 points exp UW f Ta, Assume a damage form for nonlinear loading Examples: D JC p T exp f 1 1 GISSMO n n D D exp f T p Non-Proportional Loading D 0 D 1 3. Plane Stress Models & Regularization 4. Apply to Structural Components Static & Dynamic Conditions 3-Point Bend & Axial Crush of a 600 mm Rail: 2.5 mm element mesh 21 What is most efficient method to regularize for components?

22 Eq. Plastic Strain 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 Experimental or Reference Fracture Locus Increasing Mesh Size Increasing Mesh Size Stress Triaxiality 22

23 Regularization Example: Biaxial Dome Test with 2.5 mm mesh Decent load-displacement agreement for isotropic plasticity model ¼ FE Model of Biaxial Dome Test DP980: t = 1.2 mm 23

24 Damage Regularization Effectively Alters the Damage Accumulation Rate to Trigger Fracture at the Exp. Dome Height Exp. Disp exp exp R, R T, a CAE f f i Scale Factor, R Regularized No Regularization 0.1 Element deletion Dome Height (mm) 24

25 However, Regularization is Not That Simple Regularization factor depends upon: 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 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 25

26 4 Geometries for Regularization: 0.1 to 5 mm mesh 1. JIS Tensile (T = 0.33) 2. Uniaxial Dome (T = 0.33) 26 Plane Strain Dome (T = 0.58) Equal Biaxial Dome (T = 0.66)

27 Mesh Regularization Factor Mesh Regularization for DP980: t = 1.2 mm Scale factors vary with element size as the local stress state and strain also change DP980: Material 1 Shell Element: Type 16 with 7 Integration Points Damage Exponent: 2 Relatively constant factors for elements larger than thickness Nakazima Dome: Equal-Biaxial Tension 0.5 Regularization ranking: 1. Biaxial requires least regularization 2. Uniaxial dome 3. JIS & PS Dome are similar & lowest Nakazima Dome: Uniaxial Tension JIS Tensile 0.1 Nakazima Dome: Plane Strain Tension Same trends for other 2 DP980 s Shell Element Size (mm) 27

28 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 28

29 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 29

30 Eq. Plastic Strain Stress-State Dependence: Create a Regularization Locus Calibrate a new failure locus with stress-state regularized values to obtain a regularization locus Experimental Fracture Loci: UW Model DP980 - Material 1: Transverse Direction Experimental Biaxial Dome Regularization: 2.5 mm EB Dome JIS JIS Tensile Regularization: 2.5 mm PS Dome JIS Tensile Stress Triaxiality

31 Eq. Plastic Strain Regularized Fracture Loci for Component Tests 3 Strategies for Component Simulations 1. JIS/PS Dome Regularization 2. Biaxial Dome Regularization 3. Regularization Locus Experimental Fracture Loci: UW Model DP980 - Material 1: Transverse Direction 0.40 Biaxial Dome Regularization: 2.5 mm Regularization Locus: 2.5 mm 31 3-pt Bend Model (half symmetry) Axial Crush Model (full symmetry) 0.10 JIS Tensile Regularization: 2.5 mm Stress Triaxiality

32 Force (kn) 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 DP980 Thickness: 1. 2mm 3-pt - Dynamic JIS JIS JIS Tensile Tensile Reg. Locus Biaxial Experiment Sled Displacement (m) 980 #1 32

33 Force (kn) 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 DP980 Thickness mm Dynamic Axial Crush m/s JIS/PS JIS/PS Dome Dome Experiment 100 Reg. Locus Biaxial Biaxial Sled Displacement (m) 33

34 Eq. Plastic Strain 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 Experimental Fracture Loci: UW Model DP980 - Material 1: Transverse Direction 0.40 Biaxial Dome Regularization: 2.5 mm Regularization locus will give best performance in both coupon and component simulations - much more effort required to develop JIS Tensile Regularization: 2.5 mm Regularization Locus: 2.5 mm Select biaxial dome regularization & apply to other component simulations Stress Triaxiality 34

35 Selected Component Simulations Quasi-Static & Dynamic Loading Cases: 1. No Damage Model 2. Damage Model without Regularization 3. Damage Model with Biaxial Regularization Dynamic Axial Crush Axial Crush Model 35 Dynamic 3-Point Bend 3-pt Bend Model (half symmetry)

36 Force (kn) Force (kn) 3-Pt Bend: Quasi-static & Dynamic: Material 1 (1.2 mm) Force vs. Displacement Results Experimental Fracture Locus + Biaxial Regularization can give good CAE predictions in 3-Pt Bend DP980 Thickness: 1. 2mm 3-pt: QS mm/s No damage model DP980 Thickness: 1. 2mm 3-pt: Dynamic m/s No damage model 15 Damage: No reg. 20 Damage: No reg Damage: Biaxial reg. 5 Damage: Biaxial reg Punch Displacement (m) Sled Displacement (m) Quasi-static 3pt Bend 0.5 mm/s Dynamic 3pt Bend 7.8 mm/s 36

37 3-Pt Bend: Quasi-static & Dynamic: Material 1 (1.2 mm) Energy Absorption Experimental Fracture Locus + Biaxial Regularization can give good CAE predictions in 3-Pt Bend Quasi-static 3pt Bend 0.5 mm/s Dynamic 3pt Bend 7.8 mm/s 37

38 Similar Results for Other DP980 s Quasi-static 3pt Bend 0.5 mm/s DP980 Material 2 DP980 Material 3 38

39 Comparison of Quasi-Static (QS) and Dynamic Axial Crush Numerical deformation modes are markedly different in QS and Dynamic crush: Strain-rate & inertial effects Added a force-based spot weld failure criterion Quasi-Static Dynamic 39 Quasi-Static Dynamic

40 Force (kn) QS Axial Crush: Material 1 Force-Displacement DP980: t = 1.2 mm Axial Crush: 0.5 mm/s No Damage Model 100 Experiment Damage. No Reg With Biaxial Reg & Spot Weld failure Displacement (m) Progressive Folding (No Damage Model) Progressive Folding (Damage without Reg.) Progressive Folding: 3 SW failures (Damage + Reg. + SW Failure 40

41 Dynamic Axial Crush: Material 1 Force-Displacement Mixed Buckling/ Folding (No Damage Model) Mixed Buckling/Folding (Damage without Reg.) Progressive Folding: 1 SW failure (Damage + Reg. + SW Failure) 41

42 Conclusion: Direct Strategy for Numerical Fracture Characterization Established 1. Start with the Experimental Fracture Locus - Assume failure locus form & calibrate with 4 points exp UW f Ta, Assume a damage form for nonlinear loading Examples: D JC p T exp f 1 1 GISSMO n n D D exp f T p Non-Proportional Loading D 0 D 1 3. Plane Stress Models & Regularization 4. Apply to Structural Components Static & Dynamic Conditions 3-Point Bend & Axial Crush of a 600 mm Rail: 2.5 mm element mesh 42

43 Outlook & Future Work Have developed an industrially-focused methodology for efficient fracture characterization 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 43

44 Acknowledgements 44

45 For More Information Cliff Butcher University of Waterloo Jim Dykeman Honda Research America 45