A COMPUTATIONAL APPROACH FOR DESIGNING TRIP STEELs STUDYING

Transcription

1 A COMPUTATIONAL APPROACH FOR DESIGNING TRIP STEELs STUDYING Shengyen Li June 2,

2 Alloy Design 2

3 True stress, MPa Eperimental Results - Fe-0.32C-1.42Mn-1.56Si As-cast 950 C ECAEed 2 passes route C 950 C ECAEed 2 passes route C+ Not optimized two-step heat treatment 950 C ECAEed 2 passes route C+ Optimized Two-step heat treatment True strain, mm/mm

4 Outline Motivations CALPHAD-based Models Mechanical Models Swift Model Irreversible Thermodynamics Genetic Algorithms Artificial Neural Network Conclusion 4

5 The Properties of TRIP Steel Why TRIP Steel Transformation Induced Plasticity Martensitic transformation during plastic deformation contributes to overall ductility How to make TRIP steels Select chemical composition properly Apply two-step heat treatment to manage the carbon content in austenite Target Maimize TRIP effect of the low alloying addition TRIP steel Key Stabilize austenite against the martensitic transformation during heat treatment Suppress the formation of cementite Zhu, Acta Mat.,

6 Two-Step Heat Treatment T T IA IA (1) α ϒ ϒ T BIT α ϒ M (2) BIT time α (3) B M (4) Two-step heat treatment: (1) inter-critical annealing (IA) (2)-(3) bainite isothermal transformation (BIT) (3)-(4) final cooling to room temperature Zhu, Acta Mat.,

7 Two-Step Heat Treatment T T IA IA (1) α ϒ T BIT time Two-step heat treatment: (1) inter-critical annealing (IA) Li, Acta Mat., 2012 Li, Acta Mat.,

8 Two-Step Heat Treatment α γ T T IA IA (1) α ϒ ϒ T BIT α ϒ M (2) BIT time α (3) B M (4) Two-step heat treatment: (1) inter-critical annealing (IA) (2)-(3) bainite isothermal transformation (BIT) (3)-(4) final cooling to room temperature Li, Acta Mat., 2012 Li, Acta Mat.,

9 Displacive Bainitic Transformation The Gibbs free energies of bainitic ferrite and austenite are equal at T J/mole of the strain energy is considered for bainitic transformation as T 0. The non-homogeneous C-distribution sustains the bainitic transformation. The curve is fitting based on database TCFE6 V6.2 The empirical data is obtained from: Chang et al., Met. Mat. Tran. A, 1999 and Zhao et al., J. Mat. Sci.,

10 Heterogeneous Carbon Distribution During the growth some carbon diffuses out of the ferrite grains into the surrounding austenite matri. The higher the temperature of formation, the freer the ferrite is of supersaturated carbon. Zener, 1912 Li, Acta Mat.,

11 Heterogeneous Carbon Distribution (1) BF ΔG BF ΔG time w T0 11

12 Heterogeneous Carbon Distribution (2) DC >> df DC << df 12

13 Heterogeneous Carbon Distribution (3) Caballero et al.,

14 Two-Step Heat Treatment T T IA IA (1) α ϒ ϒ T BIT α ϒ M (2) BIT time α (3) B M (4) Two-step heat treatment: (1) inter-critical annealing (IA) (2)-(3) bainite isothermal transformation (BIT) (3)-(4) final cooling to room temperature Li, Acta Mat., 2012 Li, Acta Mat.,

15 Mechanical Properties Swift Model Jacques et al. Acta Mat., 2007 σ i = K i 1 + ε 0,i ε n i Composition and micro-structure in tensile tests w C w Mn w Si Vf Fer Vf Bai Vf Aus Vf Mar γ w C A B σ = σ i Vf i i σ = σ A + σ B σ A w C γ A B Phase K i, MPa ε 0,i n i Austenite BCC Martensite Austenite BCC Martensite

16 Optimum Heat Treatment for Fe-0.32C-1.42Mn-1.56Si T IA T BIT The temperature domains (Kelvin) for optimizing the heat treatment for Fe-0.32C-1.42Mn-1.56Si Case 3 Case 2 Case 1 16

17 The Predictions for Case 1 Vf(Fer) Vf(Bai) Vf(Aus) Vf(Mar) 17

18 The Predictions for Case 1 (1) Strain, % (2) Strength, MPa (3) WTN, MPa% 18

19 The Predictions for Case 2 Vf(Fer) Vf(Bai) Vf(Aus) Vf(Mar) 19

20 The Predictions for Case 2 (1) Strain, % (2) Strength, MPa (3) WTN, MPa% 20

21 The Predictions for Case 3 (1) Strain, % (2) Strength, MPa (3) WTN, MPa% In T 0 calculations, for most of the microstructures the predicted retained austenite is less than 5%. Therefore, these diagrams include all the predicted microstructures. 21

22 True Stress, MPa True stress, MPa Optimum Heat Treatment for Maimizing Toughness As-cast 950 C ECAEed 2 passes route C 950 C ECAEed 2 passes route C+ Not optimized two-step heat treatment 950 C ECAEed 2 passes route C+ Optimized Two-step heat treatment True strain, mm/mm C 2hr min 780 C 2hr min 772 C 2hr min True Strain, mm/mm 0.20 Eperiments 22

23 Alloy Design Process 23

24 Genetic Algorithms - Schema 24

25 Optimum Composition and Heat Treatment Composition, wt%; Temperature, Kelvin w C w Mn w Si T IA T BIT bits memory for each variable 2. Vf Aus > 5% 3. Total alloying addition is less than 4 wt% individuals in one generation, 1,000 generations 5. Full equilibrium after IA treatment is considered 6. T 0 and para γ θ concepts are utilized 25

26 The Predicted Fitness as Function of Mechanical Properties Objective Function f = Vf Aus w C γ Vf Mar 26

27 Chemical Composition vs Mechanical Properties (1) w C (2) w Mn Fe-0.32C-1.42Mn-XSi (3) w Si 27

28 The Search in 6 Components, 2 Temperatures Domain w C w Mn w Si w Al w Cr w Ni T IA T BIT bits memory for each variable 2. Vf Aus > 5% Composition, wt%; Temperature, Kelvin 3. Total alloying addition is less than 4 wt% individuals in one generation, 10,000 generations 5. Full equilibrium after IA treatment is considered 6. T 0 and para γ θ concepts are utilized 28

29 Predicted Fitness as Function of Mechanical Properties Fitness w C w Mn w Si w Al w Cr w Ni T IA T BIT The predicted mechanical properties of Fe-C-Mn-Si-Al-Cr-Ni and Fe-C-Mn-Si alloys 29

30 Summary Composition Selection CALPHAD Method C Mn Si Al Ni Cr Genetic Algorithm Mechanical Properties Micro-Structure ϒ α B M Mechanical Swift Model Model based on Irreversible Thermodynamics 30

31 Plastic Deformation Model Huang et al., 2008 Huang et al., 2008 During the isothermal plastic deformation, the energy dissipation, de can be attributed to (1) the echange of the energy with the environment, dq; (2) energy consumption by dislocation variation, dw E de = TdS = Cb l dτ = dq + dw E 31

32 Plastic Deformation Model Model cont. 1 Huang et al., 2008 By energy conservation, dq can be calculated dq = du dw M Because of the dislocation: (1) Generation, dw ge ; (2) Glide, dw gl ; (3) Annihilation, dw an dw E = W ge + W gl + W an The energy dissipation can be estimated as: de = 1 2 μb2 dρ in + + τbldρ in μb2 dρ in μb2 dρ in τ in dε 32

33 Plastic Shear Stress Deformation Model cont. 2 To estimate the shear stress (τ), several mechanisms are taken into account: τ = τ 0 + τ s + τ H P + τ in 2 + τ p 2 τ 0 : Peierls force [Irvine1969; Varin1988; Zhao 2007] τ s : solid-solution strengthening [Irvine1969; Varin1988; Zhao 2007] τ H P : Hall-Petch effect [Irvine1969; Varin1988; Zhao 2007] τ in : dislocation strengthening inside the grain τ p : precipitation strengthening 33

34 Plastic Irreversible Deformation Thermodynamics Model cont. 3 Huang et al., 2008 This energy dissipation is also related to (1) the hardness parameter (σ ), (2) flow stress (τ), and (3) strain rate (γ ). It is proposed: TdS = Cb l dτ = 1 2 μb2 dρ in + + τbldρ in μb2 dρ in μb2 dρ in τ in dε dρ in = dρ in + dρ in dρ in = ν 0 γ ep ΔG kt ρ indγ 34

35 Stress-Strain Curves of Steel Alloys Ferrite EI Galindo-Nava et al., Mat. Sci. Eng. A 2012 Ferrite + Martensite PEJ Rivera et al., Mat. Sci. Tech

36 Micro-Structure & Plastic Deformation Ferrite Bainite Plastic-Deformed Bainitic Ferrite Austenite Film Bainitic Ferrite Strengthening Martensite Austenite Strain-Induced Martensite Plastic-Deformed Austenite Austenite Strengthening Plastic-Deformed Martensite The plastic-deformed phases No deformed phases 36

37 Bainite Sub-Unit Size Garcia-Mateo et al., Chemical Driving Force 2. Austenite Yield Strength 3. Temperature C 1 = n ω + θ 1 H 1 = tanh C 1 37

38 Mechanical Response of Bainite Garcia-Mateo et al., 2011 Training Prediction 38

39 Micro-Structure & Plastic Deformation Ferrite Bainite Plastic-Deformed Bainitic Ferrite Austenite Film Bainitic Ferrite Strengthening Martensite Austenite Strain-Induced Martensite Plastic-Deformed Austenite Austenite Strengthening Plastic-Deformed Martensite The plastic-deformed phases No deformed phases 39

40 Parameters for Olson-Cohen Model Olson et al., 1972, Jacques et al., Phil. Mag. A, 2001 Composition O-C Param. α=20, β= Fe-0.13C-1.42Mn-1.50Si α=26, β= α=20, β= α=57, β= Fe-0.16C-1.30Mn-0.38Si α=30, β= α=49, β= Vf α = 1 ep β 1 ep αε n τ YS = τ 0 + κd

41 Parameters for Olson-Cohen Model Vf α = 1 ep β 1 ep αε n Jacques et al., Philosophical Magazine A, 2001 Strain Rate = 2 mm/min = 6.67E-4 s -1 Composition T BIT, K ΔG, J/mol Fe-0.13C-1.42Mn- 1.50Si Fe-0.16C-1.30Mn- 0.38Si O-C Param α=20, β= α=26, β= α=20, β= α=57, β= α=30, β= α=49, β= (b1) α (b2) β 41

42 The Optimum Conditions for TRIP Steels C Mn Si T IA T BIT Ma min MSize Goal 1. Total alloying addition is less than 4 wt% individuals in one generation, 1,000 generations 3. Full equilibrium after IA treatment is considered 4. T 0 and para γ θ concepts are utilized 42

43 Phase Constituent and Performance Vf(Fer) Vf(Mar) Vf(Bai) Vf(Aus) 43

44 Chemical Composition and Performance w(c) w(mn) w(si) 44

45 The Optimum Conditions To achieve 15%-1600MPa, the recommended conditions are (wt%; K): Goal C Mn Si T IA T BIT

46 Conclusion Composition Selection CALPHAD Method C Mn Si Al Genetic Algorithm Mechanical Properties Micro-Structure ϒ α B M Mechanical Model based on Irreversible Thermodynamics 46

47 Selected References 1. S Li et al., Thermodynamic analysis of two-stage heat treatment in TRIP steels 2. S Li et al., Development of a Kinetic Model for Bainitic Isothermal Transformation in Transformation-Induced Plasticity Steels 3. H. Bhadeshia, Bainite in Steels. 4. Caballero et al., Design of Advanced Bainitic Steels by Optimisation of TTT Diagrams and T0 Curves 5. Xu et al., Genetic alloy design based on thermodynamics and kinetics 6. Matsumura et al., Mechanical properties and retained austenite in intercritically heattreated bainite-transformed steel and their variation with Si and Mn additions 7. De Cooman, Structure properties relationship in TRIPsteels containing carbide-free bainite 8. Fan et al., A Review of the Physical Metallurgy related to the Hot Press Forming of Advanced High Strength Steel 47