Review of. and Preliminary Crack Simulations. Objectives

Transcription

1 ANNUAL REPORT 2010 UIUC, August 12, 2010 Review of Longitudinal Facial Cracking and Preliminary Crack Simulations Lance C. Hibbeler and Hemanth Jasti (Ph.D. Student) (MSME Student) Department of Mechanical Science and Engineering University of Illinois at Urbana-Champaign Objectives Review mechanisms for longitudinal cracks in continuously cast steel Simulate crack formation using thermo- mechanical FE model University of Illinois at Urbana-Champaign Metals Processing Simulation Lab Lance C. Hibbeler 2

2 Cracks in Continuously Cast Steel Cracks form by combination of Surface Cracks (initiated in the mold) Transverse corner Transverse surface Longitudinal midface Longitudinal off corner Star 1) tensile stress and 2) metallurgical embrittlement Internal cracks (initiated at solidification front) Midway Straightening Pinch roll Diagonal Triple point Off corner (subsurface) Radial streaks Centerline BG Thomas Longitudinal Facial Cracks (LFCs) Front 100 mm Top Some LFCs have depressions Some don t Brimacombe & Sorimachi, Met Trans B, 8B, 1977, pp University of Illinois at Urbana-Champaign Metals Processing Simulation Lab Lance C. Hibbeler 4

3 Longitudinal Facial Cracks in Continuous Casting Front Top 100 mm Outer Flat Outer Curve Inner Curve Inner Flat Inner Curve Outer Curve Outer Flat Mode Symbol Location Notes I Inside Curve Funnel only; depression-type; excessive bending II Outside Curve Funnel only; depression-type; excessive NF taper III No Preference Jagged, short cracks; heat transfer related IV Near SEN Fluid-flow related V Off-Corner WF Inadequate NF taper University of Illinois at Urbana-Champaign Metals Processing Simulation Lab Lance C. Hibbeler 5 tal LFC Brea akouts (%) Percen ntage of Tot Corus IJmuiden Plant Experience LFC Breakout Locations Data provided by A. Kamperman of Corus Inner Flat Inner Curve Outer Curve Outer Flat Mode IV Mode I Mode II January August 2007 Mode V Distance from Center (mm) Each interval is 10 mm. Shows the locations of depression-type LFC s that caused breakouts University of Illinois at Urbana-Champaign Metals Processing Simulation Lab Lance C. Hibbeler 6

4 Mold Wall Mold Wall Longitudinal Facial Cracking Depression Mechanism (Type I, II) Root cause is non-uniform heat transfer Initiate nonuniformity (shell depression) Variations in slag rim thickness at meniscus Gap from necking (mold friction issues) Gap from buckling (excessive NF taper) Depression causes: Lower heat flux Higher shell temperature Thinner shell Grain growth (larger grains) More brittle behavior Stress and strain concentrations Amplifies if the shell buckles Combination causes cracks Tensile inelastic strain exceeds critical value cracks form University of Illinois at Urbana-Champaign Metals Processing Simulation Lab Lance C. Hibbeler 7 Mechanism: Longitudinal Facial Cracks High tensile strains & stresses in the solidifying shell at the meniscus, due to high heat transfer and/or non-uniform shell growth. Mainly thermal in origin Influencing factors (worse with): peritectic steels ( %C) high S level or low Mn/S ratio < 25 high or variable casting speed Metal level fluctuations Mold powder, taper, oscillation problems Overcooling in sprays Insufficient submold support BG Thomas Poor alignment (especially between mold & submold) University of Illinois at Urbana-Champaign Metals Processing Simulation Lab Lance C. Hibbeler 8

5 Causes of Nonuniform Mold Heat Transfer Level fluctuations (fluid flow problems, too-shallow submergence depth, etc.) Mold hotface variations around perimeter at meniscus Mold water slots (slot variations, cold mold water) Mold water quality (local l scale plugging a channel, etc. cause local l variations Superheat variations Abrupt speed changes Excessive heat removal (makes variations more likely) Insufficient heat SEN preheat causing meniscus bridging (Robinson 1994) BG Thomas University of Illinois at Urbana-Champaign Metals Processing Simulation Lab Lance C. Hibbeler 9 High Temperature Embrittlement BG Thomas University of Illinois at Urbana-Champaign Metals Processing Simulation Lab Lance C. Hibbeler 10

6 Mechanism of Longitudinal Cracking Metal level Casting direction Side view (mainly Mode III) Mold wall shell Top view BG Thomas University of Illinois at Urbana-Champaign Metals Processing Simulation Lab Lance C. Hibbeler 11 Location of Crack Formation start Strand surface x 1 x 1 x 2 Liquidus Solidus Internal crack end x 2 BG Thomas Casting Direction High Temperature Zone of Low Ductility University of Illinois at Urbana-Champaign Metals Processing Simulation Lab Lance C. Hibbeler 12

7 Longitudinal Corner Cracks (Type V) Brimacombe & Sorimachi, Met Trans B, 8B, 1977, pp University of Illinois at Urbana-Champaign Metals Processing Simulation Lab Lance C. Hibbeler 13 Longitudinal Corner Cracks (Type V) Mechanism: Hoop stresses around large corner gap due to locally thin, embrittled shell at corner allow internal cracks to propagate through Influencing factors (worse with): Large corner radius Insufficient taper (generates corner gap in upper mold which reduces heat transfer there) Steel with %C, S>0.035%; P>0.035% Solution: Decrease corner radius to 3-4mm Optimize taper (use double or parabolic design) BG Thomas University of Illinois at Urbana-Champaign Metals Processing Simulation Lab Lance C. Hibbeler 14

8 Depression Mechanisms Inside curve: Friction + bending pins the shell at the transition points Mold may induce buckling if local l shell shrinkage is not enough to match the mold perimeter length change 3 1 Outside Curve Friction + bending pins the shell at the inside/outside curve transition point Excessive NF taper causes the shell to lift off the mold surface, reducing heat transfer Bending (funnel and ferrostatic pressure) causes tensile stress on surface, leads to necking 2 University of Illinois at Urbana-Champaign Metals Processing Simulation Lab Lance C. Hibbeler 15 Excessive Narrow Face Taper Shell under compression once the narrow face comes into good contact with the shell Occurs earlier with deeper crowns Tends to cause buckling, leads to other problems 0.5 Shell Mold Excessive NF Taper tress (MPa) Average St mm Crown 30 mm Crown 40 mm Crown 40 mm 30 mm 20 mm Gap Time (s) University of Illinois at Urbana-Champaign Metals Processing Simulation Lab Lance C. Hibbeler 16

9 Stress Profiles and Histories Through Thickness in Flat Regions St tress (MPa) s 4 s 6 s 8 s 10 s Tensile Stress Compressive Stress After initial tensile load, surface stays in compression Solidification front is always under tensile loading Net stress through-thickness h thi is always zero Distance Beneath Shell Surface (mm) Soft delta-ferrite unable to carry a -2 substantial load -4-6 Stress peaks after transition to the -8 austenite phase -10 (MPa) Stress 0 0.0mm 1.5mm 3.0mm 4.5mm 6.0mm 7.5mm By mold exit, only the first 2 mm of the shell are in compression Time (s) University of Illinois at Urbana-Champaign Metals Processing Simulation Lab Lance C. Hibbeler 17 Surface Stress Around Perimeter: Effect of Funnel Width mm Outer Funnel Width 950 mm Outer Funnel Width Shell Surfac ce Stress (M MPa) s 4 s 6 s 8 s 10 s Distance from Mold Centerline (mm) Stress component perpendicular to temperature gradient (tangent to mold surface) University of Illinois at Urbana-Champaign Metals Processing Simulation Lab Lance C. Hibbeler 18

10 Stress Near Solidification Front: Effect of Funnel Width 750 mm Outer Funnel Width 950 mm Outer Funnel Width Peak stress (in γ phase) (MPa) Ma aximum Sh hell Stress ( s 10 s 4 s 6 s 2 s Distance from Mold Centerline (mm) Stress component perpendicular to temperature gradient (tangent to mold surface) University of Illinois at Urbana-Champaign Metals Processing Simulation Lab Lance C. Hibbeler 19 Funnel Bending Effect A unique attribute of funnel molds is that the steel shell is significantly ifi bent as it slides down the mold Tension Compression Compression Tension Beam theory from solid mechanics can elucidate the important parameters in the phenomenon 2 h y This end pinned, u = 0 ε max, tensile ε max, compressive w x M y ε x = R [R.C. Hibbeler, Mechanics of Materials, 5e, 2003] University of Illinois at Urbana-Champaign Metals Processing Simulation Lab Lance C. Hibbeler 20

11 Strain Decomposition: Identify Bending Effect Centerline Inside Curve total thermal mechanical thermal elastic inelastic ε = ε + ε = ε + ε + ε University of Illinois at Urbana-Champaign Metals Processing Simulation Lab Lance C. Hibbeler 21 Analytical Bending Model: Comparison with Numerical Model Take the difference between bending a beam to the funnel radius at the meniscus and the funnel radius at some other depth: ε bending ( z) crown( z) rz ( ) = + δ( ) ( ) ( ) ( ) ( ) ( meniscus ) ( ) ( ) z δ z r z r z = = δ( z) r zmeniscus r z r z r z meniscus ( outer funnel width inner funnel width) 2 Strain (%) al Bending Mechanic Time Below Meniscus (s) Analytical Model Numerical Model 750 mm Outer Funnel Width 4 16 crown( z) mm Outer Funnel Width δ = distance from neutral axis shell thickness Compare with results of 2D model with the thermal effects subtracted h End of funnel length Distance Below Meniscus (mm) Bending Strain on Solidification Front University of Illinois at Urbana-Champaign Metals Processing Simulation Lab Lance C. Hibbeler 22

12 Analytical Parametric Study nding Strain (%) Mechanical Ben n Rate (%/s) anical Bending Strain Mecha Larger total funnel width = Larger radius = Lower bending strain and strain rate Shallower funnel = Larger radius = Lower bending strain and strain rate Longer Funnel = Increases strain near bottom of funnel (larger radius for more time), but lowers strain rate 750 mm Outer Funnel Width 850 mm Outer Funnel Width 950 mm Outer Funnel Width 130 mm Inner Funnel Width End of funnel length Distance Below Meniscus (mm) 750 mm Outer Funnel Width 850 mm Outer Funnel Width 950 mm Outer Funnel Width h End of funnel lengt Distance Below Meniscus (mm) nding Strain (%) Mechanical Ben Rate (%/s) nical Bending Strain Mecha mm Inner Funnel Width 130 mm Inner Funnel Width 0 mm Inner Funnel Width mm Outer Funnel Width Distance Below Meniscus (mm) End of funnel length 260 mm Inner Funnel Width 130 mm Inner Funnel Width 0 mm Inner Funnel Width h End of funnel length Distance Below Meniscus (mm) nding Strain (%) Mechanical Ben n Rate (%/s) anical Bending Strain Mecha 30 mm Crown at Top 20 mm Crown at Top 10 mm Crown at Top 8 mm Crown at Bottom End of funnel length Distance Below Meniscus (mm) mm Crown at Top 20 mm Crown at Top 10 mm Crown at Top End of nnel length fun Distance Below Meniscus (mm) nding Strain (%) Mechanical Ben Rate (%/s) mm Funnel Length mm Funnel Length 1050 mm Funnel Length Distance Below Meniscus (mm) nical Bending Strain mm Funnel Length 950 mm Funnel Length 1050 mm Funnel Length University of Illinois at Urbana-Champaign Metals Processing Simulation Lab Lance C. Hibbeler 23 Mecha Distance Below Meniscus (mm) Subsurface Hot Tears Typical solidification stresses put tension on the solidification front Tension increased by bending effect in inner curve region, thus higher risk of hot tearing Critical hot tearing strain quantified by Won: Won et al., Metall. Mat. Trans., 31B:4 (2000), pg. 779 εc = ε Δ T B Brittle temperature zone (BTZ): Δ T = T( f = 99%) T( f = 90%) B s s Average inelastic strain rate in BTZ: g ε ( fs = 99%) ε ( fs = 90%) ε = t( f = 99%) t( f = 90%) s s University of Illinois at Urbana-Champaign Metals Processing Simulation Lab Lance C. Hibbeler 24

13 Subsurface Hot Tears Extremely fine mesh required to apply Won model (0.06 mm element size is insufficient) Use 1D numerical model to calculate temperatures and inelastic strain profile history in flat regions of mold Add bending effect with analytical model Low-carbon steels exhibit strong numerical noise Use a higher carbon grade (0.07%C) to reduce effect High-carbon grades are also more crack-sensitive Define damage index as ratio of actual damage strain to critical damage strain (crack forms at unity) εdmg = ε( fs = 99%) ε( fs = 90%) D = εdmg εc University of Illinois at Urbana-Champaign Metals Processing Simulation Lab Lance C. Hibbeler 25 Subsurface Hot Tears No hot tears will form under normal operation Bending effect increases likelihood of cracks Most likely place is just a few mm subsurface Funnels more susceptible to hot tears: Narrower funnel width (higher bending strain) Deeper crowns (higher h bending strain) Longer (higher strain rate when mushy zone is large) (%/s) nelastic Strain Rate ( Avg. In rain (%) Won Critical Str Parallel Mold Nominal Funnel Mold 200 mm Wider Funnel 100 mm Longer Funnel Distance From Shell Surface (mm) Parallel Mold Nominal Funnel Mold 200 mm Wider Funnel 100 mm Longer Funnel Distance From Shell Surface (mm) Damage Strain (%) x (--) Damage Index Parallel Mold Nominal Funnel Mold 200 mm Wider Funnel 100 mm Longer Funnel Distance From Shell Surface (mm) Parallel Mold Nominal Funnel Mold 200 mm Wider Funnel 100 mm Longer Funnel Distance From Shell Surface (mm) University of Illinois at Urbana-Champaign Metals Processing Simulation Lab Lance C. Hibbeler 26

14 Implications on Funnel Design This effect is proportional to the funnel radius Larger radius = lower cracking tendency Also affected by funnel shape in casting direction Want more change in shape close to the meniscus when the mushy zone is still small Radiused style better than linear These subsurface cracks propagating through the shell are the likely mechanism behind a depression evolving into a breakout University of Illinois at Urbana-Champaign Metals Processing Simulation Lab Lance C. Hibbeler 27 Implications on Funnel Design A larger funnel radius provides: More uniform heat transfer Smaller bending effect in the transition region Lower tendency to form subsurface hot tears A better funnel (with respect to depression-type LFCs) has a wide funnel and small crown Decrease crown Increase outer funnel width Decrease inner funnel width ( f f ) 2 crown ( z ) outer funnel width inner funnel width rz ( ) = crown( z) Depression-related LFCs are also affected by fi friction, and narrow-face taper Many other phenomena can cause LFCs

15 Crack Simulation Domain Moving displacement with straight line enforced Insulated q=0 Stress free Solidification Solidification q mold Fixed Displacement Crack Insulated q=0 Crack H Jasti University of Illinois at Urbana-Champaign Metals Processing Simulation Lab Lance C. Hibbeler 29 Initiate a Depression with Nonuniform Heat Transfer Varies with distance away from the crack Varies with time in mold Fra action of qmol ld Fraction of qmold applied Distance from Crack (mm) Frac ction of qmold Fraction of qmold at crack edge over time Time (seconds) H Jasti University of Illinois at Urbana-Champaign Metals Processing Simulation Lab Lance C. Hibbeler 30

16 Depression Simulations Vertical direction inelastic strain, 0.75 mm tensile displacement superimposed on solidification shrinkage Heat flux locally decreased BY 50% Heat flux locally decreased BY 80% Greater decreases in heat flux lead to deeper depressions H Jasti University of Illinois at Urbana-Champaign Metals Processing Simulation Lab Lance C. Hibbeler 31 Depression Simulation Case with 80% reduction in heat flux produces reasonable depression shape However, comparison is with a cold sample Brimacombe et al., MMTB 1979 H Jasti University of Illinois at Urbana-Champaign Metals Processing Simulation Lab Lance C. Hibbeler 32

17 Study of Depression Behavior Greater decreases in heat flux lead to deeper depressions Differences not significant until mold exit H Jasti University of Illinois at Urbana-Champaign Metals Processing Simulation Lab Lance C. Hibbeler 33 Study of Depression Behavior Increasing superimposed tensile displacement makes depressions deeper Necking phenomenon 0.45 Effect of Applied Tension Strain on Depression Displacement from Mold Face (mm) Distance from Crack Edge (mm) 0.75mm displacement 0.25mm 0.50mm 1.0mm H Jasti University of Illinois at Urbana-Champaign Metals Processing Simulation Lab Lance C. Hibbeler 34

18 Study of Depression Behavior Decreasing applied heat flux under superimposed compressive displacement causes deep depressions via buckling old Face (mm) splacement from M Dis Effect of Heat flux Uniformity on Depression Creation with Compressive Strain Distance from Crack Edge (mm) 20% of Average Heat Flux at Crack 50% 80% 100% H Jasti University of Illinois at Urbana-Champaign Metals Processing Simulation Lab Lance C. Hibbeler 35 Cracking Potential Damage index increases with increasing superimposed tensile displacement and increasing drop in heat flux Cracking is imminent i under certain conditions (strong imposed tension) Damage I ndex Distance from Shell Surface (mm) Ideal Casting mm Dis and 50% HF drop at crack 0.75mm displacement 50% HF drop at crack Index Damage I Distance from Shell Surface (mm) 0.75mm Dis with HF drop to 20% at crack 0.75mm Dis over 4 secs and HF drop to 50% at crack during same 4 secs H Jasti University of Illinois at Urbana-Champaign Metals Processing Simulation Lab Lance C. Hibbeler 36

19 Conclusions - 1 Five families of LFCs have been observed I. Funnel molds: inner curve depressions Lessen with larger horizontal funnel radius II. Funnel molds: outer curve depressions Lessen by optimizing taper III. Heat transfer related IV. Fluid flow related near SEN V. Taper related near NF University of Illinois at Urbana-Champaign Metals Processing Simulation Lab Lance C. Hibbeler 37 Conclusions - 2 Depressions can be formed from severe local drops in heat flux Superimposed tension (insufficient taper) leads to slightly deeper depressions Superimposed compression (excessive taper) leads to much deeper depressions However, cracks require tension to form, so either subsurface cracks propagate through the shell or something is very wrong at the shell surface University of Illinois at Urbana-Champaign Metals Processing Simulation Lab Lance C. Hibbeler 38

20 Acknowledgements Continuous Casting Consortium Members (ABB, Arcelor-Mittal, Baosteel, Corus, LWB Refractories, Nucor Steel, Nippon Steel, Postech, Posco, ANSYS-Fluent) National Center for Supercomputing Applications (NCSA) at UIUC Abe cluster Dassault Simulia, Inc. (ABAQUS parent company) University of Illinois at Urbana-Champaign Metals Processing Simulation Lab Lance C. Hibbeler 39