Reducing Uncertainty: Reflections on Establishing Life Limits 2014 ASTM JoDean Morrow Lecture on Fatigue of Materials New Orleans, LA 11 November 2014 J.M. Larsen1, S.K. Jha2, M.J. Caton1, R. John1, A.H. Rosenberger1, D.J. Buchanan3, C.J. Szczepanski5, W.J. Porter3, A.L. Hutson3, P.J. Golden1, J.R. Jira1, S. Mazdiyasni1, V. Sinha4 Integrity Service Excellence Air Force Research Laboratory Wright-Patterson Air Force Base, OH 45433 1AFRL/RXC, 2Universal 3University Technology Corporation of Dayton Research Institute, 4UES, Inc.., 5Special Metals Corp. Approved for CaseNo. No.88-ABW-2013-0906 88ABW-2015-0198 Approved forpublic publicrelease: release: Case 1 In-house and Collaborative Team Government Mike Caton Lt. Chris Fetty Pat Golden Lt. Sigfried Herring Jay Jira Reji John Jim Larsen Siamack Mazdiyasni Ryan Morrissey Andy Rosenberger Mike Shepard Chris Szczepanski Lt. Steve Visalli On-site Contractor (UDRI) Bob Brockman Marc Huelsman Dennis Buchanan David Johnson Kezhong Li John Porter Herb Stumph Pete Phillips On-site Contractor (GDIT) Universal Technology Corp. (UTC) Sushant Jha Universal Energy Systems (UES) Vikas Sinha University of Texas at San Antonio Harry Millwater University of Michigan Wayne Jones Tresa Pollock Christ Torbet Ohio State University Alison Polasik Hamish Fraser Mike Mills Jim Williams Statistical Engineering Inc. Chuck Annis, Jr., P.E. Independent Consultant Tom Cruse Mike Dent Approved for public release: Case No. 88ABW-2015-0198 2 Outline Life management of high performance turbine engines – Today and tomorrow Alloys explored: Ti-10V-2Fe-3Al Ti-6Al-2Sn-4Zr-6Mo () Fatigue variability and uncertainty Ti-6Al-2Sn-4Zr-6Mo (L-) – Examples Ti-6Al-2Sn-4Zr-2Mo () • Ti-6Al-2Sn-4Zr-6Mo () Ti-6Al-4V • IN100 Gamma TiAl Waspaloy (Wrought) Future opportunities IN100 (P/M: fine grain) – Life management & design IN100 (P/M: coarse grain) – Verification & validation René-88 DT (P/M) – Optimize Performance, Safety, Reliability, IN718 (Wrought) Maintainability, Affordability, Utilization Ni Single Crystal 1484 Al 7075-T651 Acknowledgements: Al-Cu-Mg-Ag alloy AFRL/RX & AFRL/HQ AFOSR -- Multi-Scale Structural Mechanics and Prognosis (Dr. David Stargel) AFOSR -- Structural Mechanics (Dr. Victor Giurgiutiu) DARPA/DSO – Engine System Prognosis (Dr. Leo Christodoulou) Approved for public release: Case No. 88ABW-2015-0198 3 Outline Life management of high performance turbine engines – Today and tomorrow Alloys explored: Ti-10V-2Fe-3Al Ti-6Al-2Sn-4Zr-6Mo () Fatigue variability and uncertainty Ti-6Al-2Sn-4Zr-6Mo (L-) – Examples Ti-6Al-2Sn-4Zr-2Mo () • Ti-6Al-2Sn-4Zr-6Mo () Ti-6Al-4V • IN100 Gamma TiAl Waspaloy (Wrought) Future opportunities IN100 (P/M: fine grain) – Life management & design IN100 (P/M: coarse grain) – Verification & validation René-88 DT (P/M) – Optimize Performance, Safety, Reliability, IN718 (Wrought) Maintainability, Affordability, Utilization Ni Single Crystal 1484 Al 7075-T651 Acknowledgements: Al-Cu-Mg-Ag alloy AFRL/RX & AFRL/HQ AFOSR -- Multi-Scale Structural Mechanics and Prognosis (Dr. David Stargel) AFOSR -- Structural Mechanics (Dr. Victor Giurgiutiu) DARPA/DSO – Engine System Prognosis (Dr. Leo Christodoulou) Approved for public release: Case No. 88ABW-2015-0198 4 Design Certification Methodology to Assure Integrity Throughout the Life Cycle Usage (e.g. Stress) Propulsion System Integrity Program (PSIP) - MIL-STD-3024 Untapped Performance Mean Max Safe Life Typical log Life (e.g. Cycles or TACs) • Design and certify all components are within this “safe” zone. • All components are “not safe” if one in 1000 is predicted to initiate a crack “Safe Life” has been standard practice for engine rotors for over 50 years. …………………….. Used to compensate for uncertainty/lack of knowledge For Official Use Only (FOUO) Approved for public release: Case No. 88ABW-2015-0198 5 Traditional Life Prediction Process Stress-life (S-N) Fatigue Tests – All conditions Fit S-N data with Multi-Condition Regression • Data-Driven 99.9% B50/B.1 = Scatter Factor (material + condition + model) • Distribution w.r.t. mean behavior 50% • Potentially Condition 1 untapped performance 0.1% Condition 2 B.1 B50 Actual/Predicted Lifetime (A/P) Condition n Fleet Scale-up B0.1 Lifetime Component Scale-up • Needs generation of new database for new material or microstructure • Difficult to incorporate effects of residual stress, mission, microstructure, etc. B0.1 Approved for public release: Case No. 88ABW-2015-0198 6 Propulsion System Integrity Program Life-Cycle Design Philosophy (PSIP; MIL-STD-3024) Low-Cycle-Fatigue Design Criteria (safe life) Based on statistical lower bound 1 in 1000 components predicted to initiate a 0.8 mm crack Deterministic 1 or 2 safety inspections during service life • • aC Mean Typical Lower Bound Crack Length Usage (e.g. Stress) • • Damage-Tolerant Design Criteria (fracture mechanics) a* ai log Life (e.g. Cycles or TACs) Cycles (or Equivalent) Both design criteria are met at all critical locations on a component Approved for public release: Case No. 88-ABW-2013-0906 Approved for public release: Case No. 88ABW-2015-0198 7 Move Engine Lifing from Safe-Life Approach to Retirement For Cause LCF Initiation Distribution Traditional “Safe-Life” Retirement Approach Manage to -3 Lower Bound -3 0 Before 1980s RFC program 10000 20000 30000 40000 Life (Time or Cycles) 50000 -3 60000 70000 Number of Parts 0 10000 10000 Traditional “Safe-Life” Retirement Approach Manage to -3 Lower Bound 20000 30000 40000 50000 60000 70000 B0.1 = 8000 TAC LCF Initiation Distribution -3 0 After 1980s RFC program Life (Time or Cycles) B0.1 = 4000 TAC Retire all components when 1 in 1000 is predicted to fail Retire all components when 1 in 1000 is predicted to fail Economic/Risk Limit = Definition of Retirement for Cause After ERLE program Traditional “Safe-Life” Retirement Approach Manage to -3 Lower Bound 20000 30000 40000 Life (Time or Cycles) B0.1 = 12000 TAC 50000 60000 70000 0 2000 4000 6000 8000 10000 12000 14000 16000 18000 20000 22000 24000 26000 28000 30000 32000 34000 36000 38000 40000 42000 44000 46000 48000 50000 52000 54000 56000 58000 60000 62000 64000 66000 68000 70000 Number of Parts Retire all components when 1 in 1000 is predicted to fail Number of Parts LCF Initiation Distribution Penetrate the LCF Distribution Approved for public release: Case No. 88ABW-2015-0198 Life (Time or Cycles) 8 Prognosis will Enable Transformation in Asset Management Dr. Leo Christodoulou Yes Service Failure physics, damage evolution, predictive models Prognosis Database: Mission History, Maintenance, Life Extension, and Design Failure Occurrences Usage (Duty Cycles) State Awareness Interrogation “Book Life” Today “Book Life” Tomorrow Reduce and Manage Uncertainty NO Retire Prognosis Translates Knowledge and Information Richness to Physical Capability Approved for public release: Case No. 88ABW-2015-0198 9 Background •Current design and life management of turbine engine materials – Extensive fatigue testing required to produce large databases – Statistically-based life limits by extrapolation from the mean behavior •Next-generation design and life management – Design Target Risk: • • DoD: < 5*10-8 failures/engine flight hour FAA: < 1*10-9 failures/flight – Safety, reliability, affordability – Reduced life-cycle cost – Reduction in uncertainty in materials life-cycle prediction – Reduce requirements for materials testing •Overarching science and technology initiatives – DoD Engineered Resilient Systems – Materials Genome Initiative (MGI) – Integrated Computational Materials Engineering (ICME) – Big Data Approved for public release: Case No. 88-ABW-2013-0906 Approved for public release: Case No. 88ABW-2015-0198 10 Opportunity: Physics-Based Description of Fatigue Variability Traditional (Empirical) Description Physics-Based Description of Fatigue Variability Fatigue variability described as deviation from the expected mean-behavior Fatigue variability described as separation of the mean and the life-limiting behavior Overall mean behavior Mean behavior Distribution in the lifelimiting mechanism (crack-growth controlled) Variability described w.r.t. the overall mean behavior max max Variability in the meandominating response Nf (Cycles) POF = 0.1% life limit (Book life) Life-limit based on the uncertainty in the worst-case mechanism Large degree of uncertainty associated with life prediction Failure Occurrence Failure Occurrence Nf (Cycles) POF = 0.1% life limit Usage (Duty cycles) Approved for public release: Case No. 88ABW-2015-0198 Crack growth related peak (life-limiting mechanism) Mean-lifetime dominating peak Total variability Duty cycles 11 N. E. Frost, K. J. Marsh, and L. P. Pook "Metal fatigue, 1974." Oxford University Press, Oxford. Approved for public release: Case No. 88-ABW-2013-0906 Approved for public release: Case No. 88ABW-2015-0198 12 Life-limiting Fatigue Total Fatigue Life = NTotal Small-Crack Crack Initiation 0? Growth Long-Crack Growth NTotal ? ? ? Ni NP,small NP,long Ni NP,small NP,long NTotal Low-Cycle-Fatigue Life Limits: A New Understanding Life-limiting low-cycle-fatigue life is governed by the growth of a dominant crack from an initial crack size defined by the microstructural features & mechanisms that control crack formation. For Official Use Only (FOUO) Approved for public release: Case No. 88ABW-2015-0198 13 Outline Life management of high performance turbine engines – Today and tomorrow Alloys explored: Ti-10V-2Fe-3Al Ti-6Al-2Sn-4Zr-6Mo () Fatigue variability and uncertainty Ti-6Al-2Sn-4Zr-6Mo (L-) – Examples Ti-6Al-2Sn-4Zr-2Mo () • Ti-6Al-2Sn-4Zr-6Mo () Ti-6Al-4V • IN100 Gamma TiAl Waspaloy (Wrought) Future opportunities IN100 (P/M: fine grain) – Life management & design IN100 (P/M: coarse grain) – Verification & validation René-88 DT (P/M) – Optimize Performance, Safety, Reliability, IN718 (Wrought) Maintainability, Affordability, Utilization Ni Single Crystal 1484 Al 7075-T651 Acknowledgements: Al-Cu-Mg-Ag alloy AFRL/RX & AFRL/HQ AFOSR -- Multi-Scale Structural Mechanics and Prognosis (Dr. David Stargel) AFOSR -- Structural Mechanics (Dr. Victor Giurgiutiu) DARPA/DSO – Engine System Prognosis (Dr. Leo Christodoulou) Approved for public release: Case No. 88ABW-2015-0198 14 Ti-6Al-2Sn-4Zr-6Mo (Ti-6-2-4-6) Approved for public release: Case No. 88-ABW-2013-0906 Approved for public release: Case No. 88ABW-2015-0198 15 Lifetime Distribution Ti-6-2-4-6, RT, R = 0.05, = 20 Hz and 20 kHz Approved for public release: Case No. 88-ABW-2013-0906 Approved for public release: Case No. 88ABW-2015-0198 16 Confidence Bounds on B0.1 Lifetime -- All Data 99.99 Probability of failure (%) 99.9 max = 820 MPa 99 95 90 80 70 50 30 20 10 5 1 All data points 721 cycles 95% confidence intervals .1 .01 100 3 10 4 10 10 5 10 6 10 7 8 10 9 10 Lifetime, N (Cycles) f Approved for public release: Case No. 88-ABW-2013-0906 Approved for public release: Case No. 88ABW-2015-0198 17 Bimodal Model 99.9 Probability of failure (%) max 99 95 90 80 70 = 820 MPa f t ( N ) pl f l ( N ) (1 pl ) f m ( N ) 50 30 20 10 5 1 Data Bimodal fit Lower bound Upper bound 4565 cycles .1 100 10 3 4 10 5 10 10 6 10 7 10 8 9 10 N (Cycles) f Approved for public release: Case No. 88-ABW-2013-0906 Approved for public release: Case No. 88ABW-2015-0198 18 Confidence Bounds on B0.1 Lifetime Limiting Condition of pl → 1 99.99 Probability of failure (%) 99.9 max = 820 MPa 99 95 90 80 70 50 30 20 10 5 1 f t ( N ) pl f l ( N ) (1 pl ) f m ( N ) pl 1 Life-limiting distribution 5660 cycles 95% confidence intervals .1 .01 100 10 3 4 10 5 10 10 6 10 7 10 8 9 10 Lifetime, N (Cycles) f Approved for public release: Case No. 88-ABW-2013-0906 Approved for public release: Case No. 88ABW-2015-0198 19 Lifetime Distribution Ti-6-2-4-6, RT, R = 0.05, = 20 Hz and 20 kHz Approved for public release: Case No. 88-ABW-2013-0906 Approved for public release: Case No. 88ABW-2015-0198 20 CDF Space Probability of Occurence (%) Effect of Stress Level on Mean vs. Life-Limiting Behavior 99 Meandominating behavior 95 90 80 70 50 1040 MPa 925 MPa 900 MPa 860 MPa 820 MPa 700 MPa 650 MPa 600 MPa 550 MPa 30 20 10 5 1 4 10 Life-limiting behavior 5 10 6 10 7 10 10 8 10 9 10 10 Cycles to Failure Approved for public release: Case No. 88-ABW-2013-0906 Approved for public release: Case No. 88ABW-2015-0198 21 Bimodal Fatigue Behavior Ti-6Al-2Sn-4Zr-6Mo; RT Approved for public release: Case No. 88-ABW-2013-0906 Approved for public release: Case No. 88ABW-2015-0198 22 99 95 90 ys= 1140 MPa 80 70 50 1040 MPa 925 MPa 900 MPa 860 MPa 820 MPa 700 MPa 650 MPa 600 MPa 550 MPa 30 20 10 5 1 4 10 5 10 6 10 10 7 10 8 10 9 10 10 Cycles to Failure Failure Occurrence Probability of Occurence (%) Probability of Occurrence of LifeLimiting Failures Probability of Life-Limiting Failures Crack-growth-controlled B0.1 density (Critical lifetimes heterogeneity level) Mean-dominating density (Smaller heterogeneity scales) Empirically-derived density Duty cycles Stress (MPa) Approved for public release: Case No. 88-ABW-2013-0906 Approved for public release: Case No. 88ABW-2015-0198 23 Alternate Life-Prediction Approach 99.99 behavior separate with decreasing and respond differently to operating variables. • Life Prediction based on variability in the worst‐case mechanism. Probability of Failure (%) • The Mean and the worst‐case All points Type I Type II 99.9 99 max 95 90 80 70 = 860 MPa Type I 50 30 20 10 5 Type II Reduction in uncertainty 1 .1 when compared to the traditional approach. • Improved reliability of design life. 10 4 1 in 1000 Life limits 10 5 Cycles to Failure, N Variability in crack growth 6 10 7 10 f Variability in crack Initiation + growth Failure Occurrence • Significant reduction in uncertainty .01 1000 Duty cycles FOR OFFICIAL USE ONLY Approved for public release: Case No. 88ABW-2015-0198 24 Alternate Life-Prediction Approach 99.99 behavior separate with decreasing and respond differently to operating variables. • Life Prediction based on variability in the worst‐case mechanism. Probability of Failure (%) • The Mean and the worst‐case All points Type I Type II Simulated, Type I 99.9 99 95 90 80 70 max = 860 MPa Type I 50 30 20 10 5 Type II Reduction in uncertainty 1 .1 when compared to the traditional approach. • Improved reliability of design life. 10 4 1 in 1000 Life limits 10 5 Cycles to Failure, N Variability in crack growth 6 10 7 10 f Variability in crack Initiation + growth Failure Occurrence • Significant reduction in uncertainty .01 1000 Duty cycles FOR OFFICIAL USE ONLY Approved for public release: Case No. 88ABW-2015-0198 25 Mechanism-Based Probabilistic Prediction of Limiting Life Crack Initiation Size Predicted Life-Limiting Distribution 35 99.99 area p 99.9 Crcak nucleation area Probability of Failure (%) Occurrence frequency 30 25 20 15 10 5 0 0 20 40 60 80 100 120 140 160 2 area; Crack nucleation area (m ) 180 max 99 95 90 80 70 50 30 20 10 5 10 -4 10 -5 10 -6 Experimental .1 Life-limiting population .01 4 10 10 da/dN (m/cycle) Long crack 10 10 -8 10 -9 da f (K ) dN max af -10 Ti-6-2-4-6 = 860 MPa max R = 0.05 = 20 Hz T = 23°C -11 10 -12 1 6 10 10 7 f 10 10 5 N (Cycles) Small cracks ( = 860 MPa) Power-law fits -7 Experimental (Life limiting) Predicted (Life limiting) 1 p Small-Crack Growth Variability = 860 MPa 10 K (MPa-m ) 1/2 Np a ai da f (K ) • Prediction of limiting life of Ti6Al-2Sn-4Zr-6Mo • Monte Carlo simulation based on microstructural features and small-crack growth 100 Approved for public release: Case No. 88-ABW-2013-0906 Approved for public release: Case No. 88ABW-2015-0198 26 Confidence Bounds on B0.1 Lifetime Limiting Condition of pl → 1 99.99 Probability of failure (%) 99.9 max = 820 MPa 99 95 90 80 70 50 30 20 10 5 1 f t ( N ) pl f l ( N ) (1 pl ) f m ( N ) pl 1 Life-limiting distribution 5660 cycles 95% confidence intervals .1 .01 100 10 3 4 10 5 10 10 6 10 7 10 8 9 10 Lifetime, N (Cycles) f Approved for public release: Case No. 88-ABW-2013-0906 Approved for public release: Case No. 88ABW-2015-0198 27 Crack-Growth-Controlled Failures 99.99 Probability of Failure (%) 99.9 max = 820 MPa 99 95 90 80 70 50 30 20 10 5 B0.1 lifetime Predicted life-limiting distribution 1 Crack-growthcontrolled failures .1 .01 100 1000 10 4 10 5 10 6 7 10 8 10 10 9 Lifetime, N (Cycles) f Approved for public release: Case No. 88-ABW-2013-0906 Approved for public release: Case No. 88ABW-2015-0198 28 Bimodal Fatigue Behavior Ti-6Al-2Sn-4Zr-6Mo; RT Approved for public release: Case No. 88-ABW-2013-0906 Approved for public release: Case No. 88ABW-2015-0198 29 How Can this Understanding Affect the Life-Cycle Design Philosophy? An Integrated Design Criterion B0.1 Lifetime Predicted Distribution in a vs. N 820 MPa Limiting Damage Tolerance Curve Cycles Approved for public release: Case No. 88-ABW-2013-0906 Approved for public release: Case No. 88ABW-2015-0198 30 Life-Limiting Failure IPF map N1 N1 N2 F1 F1 F1 N2 Faceted p Loading axis Crackinitiation facet Basal plane trace Specimen tilt = 30° • max = 860 MPa; Nf = 49,893 cycles • Facet inclination = 31° Methods: • Quantitative tilt microscopy using MEXTM • FIB sectioning through crack-initiation facet (in some cases) • EBSD analysis of the crack-initiation region Approved for public release: Case No. 88-ABW-2013-0906 Approved for public release: Case No. 88ABW-2015-0198 31 Summary of Mean vs. Life-Limiting Configurations Surface-Initiated Mechanisms Faceted grains Resolved along the loading axis Facet inclination w.r.t. the loading axis (°) Facet inclination Resolved along the facet normal 50 45 40 35 Neighboring grains 30 Life-limiting 25 4 10 Mean-dominating 5 10 6 10 7 10 Lifetime (Cycles) Approved for public release: Case No. 88-ABW-2013-0906 Approved for public release: Case No. 88ABW-2015-0198 32 Probability of occurrence Hypothesis: Hierarchy of Fatigue Deformation Heterogeneities Basal plane Inclination ≤ 30 , p soft Heterogeneity level Approved for public release: Case No. 88-ABW-2013-0906 Approved for public release: Case No. 88ABW-2015-0198 33 Microstructure-Based Prediction of Life-Limiting Fatigue Mechanisms in Ti-6-2-4-6 Using the Hierarchy Model of Heterogeneity levels Microstructure Model 0.0 1.0 0.5 0.0 1 • CP-FEM • Definition of method1 • Statistically representative volume element • Smaller than labscale specimen heterogeneity parameter distribution heterogeneity parameter 1C. 2R. , str, etc. 0 • Model the • Ellipsoid packing 1.0 3 2 Deformation parameter model2 0.5 P(Life-limiting failure) Fatigue Model Probability of occurrence Compute Fatigue Heterogeneity Parameter Hierarchy Model Probability of lifelimiting mechanism • Simulate fatigue specimens (lab scale) using the hierarchy model • Spatial distribution given by the Poisson point process • Interrogate for lifelimiting criterion P. Przybyla and D . L. McDowell, International Journal of Plasticity, 2010 A. Brockman, et al. Approved for public release: Case No. 88-ABW-2013-0906 Approved for public release: Case No. 88ABW-2015-0198 34 Summary Ti-6-2-4-6 • Study fundamental drivers of fatigue lifetime distribution – Stresses and lifetimes representative of engine rotors designs • Total fatigue lifetime (NT) : NT = Ni + NSC + NLC – Ni is the dominant term only in the mean lifetime as the stress level is decreased – Ni approaches 0 cycles for the life-limiting failures • The minimum lifetime was spent almost completely in the growth of a crack that began on the microstructural scale • How can one preclude the rare conditions that lead to Ni 0? – Microstructure, surface treatments (e.g., residual stresses), etc. – Need to quantify the probability of life-limiting failure (Ni 0) • Suggests alternative interpretation for integrated life-cycle design and management of turbine-engine rotor materials and components Approved for public release: Case No. 88-ABW-2013-0906 Approved for public release: Case No. 88ABW-2015-0198 35 Outline Life management of high performance turbine engines – Today and tomorrow Alloys explored: Ti-10V-2Fe-3Al Ti-6Al-2Sn-4Zr-6Mo () Fatigue variability and uncertainty Ti-6Al-2Sn-4Zr-6Mo (L-) – Examples Ti-6Al-2Sn-4Zr-2Mo () • Ti-6Al-2Sn-4Zr-6Mo () Ti-6Al-4V • IN100 Gamma TiAl Waspaloy (Wrought) Future opportunities IN100 (P/M: fine grain) – Life management & design IN100 (P/M: coarse grain) – Verification & validation René-88 DT (P/M) – Optimize Performance, Safety, Reliability, IN718 (Wrought) Maintainability, Affordability, Utilization Ni Single Crystal 1484 Al 7075-T651 Acknowledgements: Al-Cu-Mg-Ag alloy AFRL/RX & AFRL/HQ AFOSR -- Multi-Scale Structural Mechanics and Prognosis (Dr. David Stargel) AFOSR -- Structural Mechanics (Dr. Victor Giurgiutiu) DARPA/DSO – Engine System Prognosis (Dr. Leo Christodoulou) Approved for public release: Case No. 88ABW-2015-0198 36 Model-Based Fatigue Life-Limits Mean-Based → Life-Limiting-Mechanism-Based Model-Based Probability of Life-Limiting Mechanism (Ni = 1) Probability of Life-Limiting Mechanism Crack-growth lifetime distribution Model- (life-limiting distribution) based B0.1 Mean-dominating P(Life-limiting mechanism) max = 1150 MPa; Nf = 2,210 Critical microstructural neighborhood for Ni = 1 Mechanistic Understanding distribution Volume da/dN Model of LifeLimiting Distribution PDF Data-based approach Small cracks • • Life-limiting trend is different from the mean-behavior trend Model-based predictions focus on the life-limiting behavior Method also enables incorporation of new material, microstructure, residual stress, mission, etc. Probability Crack Initiation Size • Distribution in Life-Limiting Mechanism Life-limiting distributions K B0.1 Nf (Life-Limiting) Approved for public release: Case No. 88ABW-2015-0198 37 Mechanism Mapping for Kt = 1 w.r.t. Stress Level Fine Grain IN100 (650°C) Coarse Grain IN100 (650°C) 1150 1250 Subsurf. NMP Surf. pore 1100 Mean lifetime Surface NMP (MPa) 1150 IN100 650°C 1050 max 1100 Surface NMP Subsurface NMP Surface pore Subsurface pore Surf. NMP max (MPa) Surface 1200 pore 1050 1000 1000 950 100 Subsurface NMP 1000 4 10 10 N (Cycles) 5 10 6 7 10 950 100 1000 f max = 1150 MPa; Nf = 2,210 Surface NMP Surface pore 4 10 10 N (Cycles) 5 10 6 10 7 f Subsurface NMP Surface pore Approved for public release: Case No. 88ABW-2015-0198 Subsurface NMP 38 Incorporation of Crack-Initiation Mechanism in Life Prediction Experimental Observations of Mechanism Variations Non-metallic Particle (NMP) Mean lifetime Mixed mode (MPa) 1200 Plate 300 1150 50 40 30 20 10 0 0 IN100 650°C 1100 Step 1 max Surface pore Simulation of Crack-Initiating Features 1250 1050 200 100 Pore 100 Specimens 200 1000 max = 1150 MPa; Nf = 2,210 Surface NMP 950 100 Step 2 1000 4 10 10 N (Cycles) 5 10 f Subsurface NMP Transgranular 20 m 0 40 m Transgranular 6 7 10 99.999 99.99 99.9 99 95 90 80 70 50 30 20 10 5 1 .1 .01 .001 100 Probability of Failure (%) 10 m 300 1100 MPa Data 1000 4 5 10 10 Cycles to Failure, N 10 6 f Model Prediction and Validation • There are competing mechanisms for crack-initiation • Incorporating these mechanisms in life prediction models can lead to lower uncertainty and better utilization of residual useful life For Official Use Only Forrelease: OfficialCase Use No. Only Approved for public 88ABW-2015-0198 39 Model-Based Fatigue Limits Non-metallic particle Pores P-life-limiting Feature volume 0.1 0.01 Lab-scale specimen Interrogate simulated specimens for microstructural condition representing Ni =1 1 l Simulated Plate Lab-scale specimen Component feature volume Probability of finding a condition leading to life-limiting mechanism, P Probability of Occurrence of Life-Limiting Mechanism 0.001 0.0001 0.01 0.1 1 10 100 3 Surface layer volume (mm ) • Model-based probability of occurrence of life-limiting mechanism (Ni = 1) • Volumetric effect on the probability of occurrence enables scale-up to component feature volumes For Official Use Only Approved for public release: Case No. 88ABW-2015-0198 40 Model-Based Fatigue Life Limits Smooth Geometry Life-limiting distribution 99.99 1 P-life-limiting 538°C, Subsurface initiation 99.9 566°C, Subsurface initiation max = 1000 MPa T = 538°C Lab-scale specimen 0.01 0.001 Feature volume 0.1 Probability of Failure (%) l Probability of finding a condition leading to life-limiting mechanism, P Probability of occurrence of life-limiting mechanism 0.1 1 593°C, Subsurface initiation 95 90 80 70 50 30 20 10 5 593°C, Surface initiation 1 621°C, Subsurface initiation 650°C, Subsurface initiation 677°C, Subsurface initiation 593°C, predicted lifelimiting distribution max = 1000 MPa .1 0.0001 0.01 99 10 100 .01 100 4 1000 5 10 10 10 6 Cycles to Failure, N 3 Surface layer volume (mm ) f max = 1150 MPa; Nf = 2,210 Life-limiting mechanism: Surface NMP Initiation • Life-limiting mechanism ≡ Crack initiation from • • surface NMP 1 out of 76 specimens failed by surface NMP at 1000 MPa (T = 538 – 677°C) Reasonable agreement between data and predictions of the predicted probability of occurrence and the life-limiting distribution Courtesy of John Leugers, AFRL/RW Public Release #88ABW‐2012‐2266 20 m Approved for public release: Case No. 88ABW-2015-0198 41 Mechanism-Based Prediction of LifeLimiting Distribution Predictions Inputs NMP crack-initiation size distribution Probability of Failure (%) 7 99.99 99.9 Fine Grain 6 Frequency Coarse Grain 5 4 3 2 1 .1 .01 100 95 11 0 12 5 14 0 15 5 17 0 Initiation Size (m) Variability in small-crack growth rate 10 -2 650°C; 0.33 Hz; R = 0.05 -3 da/dN (mm/cycle) 10 -4 10 -5 Long cracks (No dwell) Small cracks, pore crack initiation (1150 MPa) -6 10 Small crack, NMP crack initiation (1150 MPa) -7 10 4 6 8 10 1/2 K (MPa-m ) 30 50 70 99.99 99.9 Probability of Failure (%) 80 65 50 35 20 0 10 1150 MPa 99 95 90 80 70 50 30 20 10 5 1 Predicted life-limiting distribution 1000 4 10 10 Cycles to Failure, N 5 10 6 f 1100 MPa 99 95 90 80 70 50 30 20 10 5 1 .1 .01 100 Predicted life-limiting distribution 1000 4 10 10 Cycles to Failure, N 5 10 6 f Approved for public release: Case No. 88ABW-2015-0198 42 Comparison to Data-Based Method Predictions Inputs NMP crack-initiation size distribution Probability of Failure (%) 7 99.99 99.9 Fine Grain 6 Frequency Coarse Grain 5 4 3 2 1 95 11 0 12 5 14 0 15 5 17 0 Variability in small-crack growth rate 10 -2 650°C; 0.33 Hz; R = 0.05 -3 da/dN (mm/cycle) 10 -4 10 -5 Long cracks (No dwell) Small cracks, pore crack initiation (1150 MPa) -6 Small crack, NMP crack initiation (1150 MPa) -7 10 4 6 8 10 1/2 K (MPa-m ) 30 50 70 99.99 99.9 Probability of Failure (%) 80 65 50 35 20 Initiation Size (m) 10 Overconservative .1 .01 100 0 10 1150 MPa 99 95 90 80 70 50 30 20 10 5 1 1150 MPa (10 random tests) Predicted life-limiting distribution 1000 4 5 10 10 Cycles to Failure, N 10 6 f 1100 MPa 99 95 90 80 70 50 30 20 10 5 1 .1 .01 100 Anticonservative 1100 MPa (15 tests) Predicted life-limiting distribution 1000 4 10 10 Cycles to Failure, N 5 6 10 f Approved for public release: Case No. 88ABW-2015-0198 43 Mechanism-Based Prediction of LifeLimiting Distribution Predictions Inputs NMP crack-initiation size distribution Probability of Failure (%) 7 99.99 99.9 Fine Grain 6 Frequency Coarse Grain 5 4 3 2 1 .1 .01 100 95 11 0 12 5 14 0 15 5 17 0 Initiation Size (m) Variability in small-crack growth rate 10 -2 650°C; 0.33 Hz; R = 0.05 -3 da/dN (mm/cycle) 10 -4 10 -5 Long cracks (No dwell) Small cracks, pore crack initiation (1150 MPa) -6 10 Small crack, NMP crack initiation (1150 MPa) -7 10 4 6 8 10 1/2 K (MPa-m ) 30 50 70 99.99 99.9 Probability of Failure (%) 80 65 50 35 20 0 10 1150 MPa 99 95 90 80 70 50 30 20 10 5 1 1150 MPa, Experiment Life-limiting points Predicted life-limiting distribution 1000 4 10 10 Cycles to Failure, N 5 6 10 f 1100 MPa 99 95 90 80 70 50 30 20 10 5 1 .1 .01 100 1100MPa, 20 tests Predicted life-limiting distribution 1000 4 5 10 10 Cycles to Failure, N 6 10 f Approved for public release: Case No. 88ABW-2015-0198 44 Understanding Crack Growth at Fracture Critical Locations Machining, shot peening, glass-bead peening, blend repair => surface residual stresses Shot Peening Benefit 5.0 IN100 (cg): 650°C 0.333 Hz, R = 0.05 Kt,net = 1.8 Crack Length (mm) 4.0 net = 680.6 MPa 3.0 LSG, bore LSG, face SP = 6A, bore SP = 6A, face 2.0 1.0 0.0 0 3000 6000 9000 12000 Total Cycle Count, N • Notched specimens simulate fracture-critical features of components – Simulate crack growth under stress gradients (notches) – Simulate crack growth with shot peened residual stresses DISTRIBUTION C: Distribution authorized to US Government agencies and their contractors (Critical Technology), XX October 2013. Other requests for this document shallrelease: be referredCase to AirNo. Force88ABW-2015-0198 Research Laboratory, AFRL/RXCM. Approved for public 45 Model-Based Fatigue Life Limits Benefit of Surface Residual Stress Residual Stress MPa 200 æ -200 æ æ æ æ æ æ æ æ æ æ æ æææ æ -600 æ æ ææ æ æ æ -800 æ æ -400 æ ææ æ -1000 0.0 æ æ ææ ææ Measured shotpeen RS profiles 0.2 0.3 Distance mm 0.1 æ æ æ ææ 0.4 0.5 3.5x10 4 Benefit of RS 99 95 90 80 70 50 30 20 10 5 1 650°C 900 MPa B0.1 Without SP residual stress .1 .01 1000 Without RS With SP RS 3x10 4 B0.1 Lifetime (Cycles) Probability of Failure (%) 99.99 99.9 æ æ æ æ æ æ ææ æ æ æ æ æ ææ 0 With SP residual stress 4 10 10 5 10 6 With shotpeen RS 2.5x10 4 2x10 4 1.5x10 4 900 MPa 1x10 4 5x10 3 0 300 Without RS 350 Cycles to Failure, N 400 450 500 550 Temperature (°C) 600 650 700 f • Benefit of shot-peen residual stress can be readily incorporated in the proposed model-based life limits Courtesy of John Leugers, AFRL/RW Public Release #88ABW‐2012‐2266 Approved for public release: Case No. 88ABW-2015-0198 46 Applicability to Notched Geometries -MotivationNotch Locations are often Life Limiting • Air Hole • Bolt hole • Tang • Snap Fillet • … Point Solution @ 650°C for Kt = 1.89 99.999 Kt = 1.89 99.99 T= 650˚C; f=0.33 Hz; R=0.05 99.9 99 95 90 Percent Elastic‐Plastic Notch Analysis 80 70 50 30 20 10 5 800 MPa 900 MPa 1 Prediction 800 MPa 900 MPa .1 .01 .001 10 3 Mechanical Specimen 10 4 10 5 10 6 Cycles to Failure All lifing methods have to predict notch life For Official Use Only Approved for public release: Case No. 88ABW-2015-0198 47 Model-Based Fatigue Life-Limits Process for Components Stress Analysis K solution for fracturecritical features K Component Feature 1 Feature 2 Bolt hole a Surface RS Microstructure Mission Fillet • Model-based B0.1 method can be scaled up to a component or feature Variables such surface RS, microstructure, and mission are inputs to the model B0.1 limit Life-limiting distribution Nf (life-limiting) For Official Use Only Approved for public release: Case No. 88ABW-2015-0198 P(Life-limiting mechanism) • Probability Model-based B0.1 Model-based probability of lifelimiting mechanism (Ni = 1) Volume 48 Outline Life management of high performance turbine engines – Today and tomorrow Alloys explored: Ti-10V-2Fe-3Al Ti-6Al-2Sn-4Zr-6Mo () Fatigue variability and uncertainty Ti-6Al-2Sn-4Zr-6Mo (L-) – Examples Ti-6Al-2Sn-4Zr-2Mo () • Ti-6Al-2Sn-4Zr-6Mo () Ti-6Al-4V • IN100 Gamma TiAl Waspaloy (Wrought) Future opportunities IN100 (P/M: fine grain) – Life management & design IN100 (P/M: coarse grain) – Verification & validation René-88 DT (P/M) – Optimize Performance, Safety, Reliability, IN718 (Wrought) Maintainability, Affordability, Utilization Ni Single Crystal 1484 Al 7075-T651 Acknowledgements: Al-Cu-Mg-Ag alloy AFRL/RX & AFRL/HQ AFOSR -- Multi-Scale Structural Mechanics and Prognosis (Dr. David Stargel) AFOSR -- Structural Mechanics (Dr. Victor Giurgiutiu) DARPA/DSO – Engine System Prognosis (Dr. Leo Christodoulou) Approved for public release: Case No. 88ABW-2015-0198 49 Model-Based Life-limits: Deconstruct Uncertainty to Capture Benefits Surface Residual Stresses Microstructural Hierarchies Surface NMP max = 1150 MPa; Nf = 2,210 Surface pore 5.0 Transgranular IN100 (cg): 650°C 0.333 Hz, R = 0.05 Kt = 1.8, net = 680.6 MPa Subsurface NMP Crack Length (mm) 4.0 Mixed mode Crystallographic Notch Analysis LSG, bore LSG, face SP = 6A, bore SP = 6A, face 3.0 2.0 1.0 0.0 0 3000 6000 9000 12000 Total Cycle Count, N Model‐based life limit Transgranular Mission Usage % Max Stress 100 Simulate lifetime 3D Effects, etc. 80 60 40 20 0 Time For Official Use Only Approved for public release: Case No. 88ABW-2015-0198 50 Multi-scale Physics and Mechanics of Materials Fatigue Life Limits Mechanisms Simulations 0.0 1.0 0.5 0.0 0.5 1.0 3 2 1 0 What controls life-limit uncertainty? For Official Use Only Approved for public release: Case No. 88ABW-2015-0198 51 Integrated Computational Materials Engineering (ICME) for Life Macro-scale • Fatigue crack development and growth from a life-limiting location in a component • Detecting “large” cracks 10,000 m Crack origin Meso-scale Top-down approach – determine the physics of fatigue damage and lifetime variability • Fracture modes, small-crack growth, fracture morphology, and local neighborhood • Characterizing smaller flaws Micro-scale • Crack-initiating microstructural arrangements and mechanisms • NDE of microstructure features Nano-scale • Slip Probabilistic lifeprediction on the component-scale by integrating lab-scale information Slip traces Approved for public release: Case No. 88ABW-2015-0198 mechanisms promoting crack initiation 52 Model-based Life-limit Approach Implications -- Based on Predicted Risk Reliability • Deconstruct Uncertainty • Microstructure-based lifing Maintainability • Integrated life cycle • Optimize for maintainability Affordability • Much less testing • NDE: Tailored POD Sustainment of Legacy Engines • Understand & reduce life-cycle uncertainty Digital material lifecycle and design • Optimize for full life Life-cycle Design • Materials / microstructures • Components / features Verification & Validation • Probabilistic risk • Validation material science Manufacturing • Optimized processes • Digital Thread life-cycle data For Official Use Only (FOUO) Approved for public release: Case No. 88ABW-2015-0198 New Engines • Minimize life-cycle uncertainty 53 Related Publications • • • • • • • • • • • • S. K. Jha, C. J. Szczepanski, R. John, and J. M. Larsen, “Demonstration of a Method for Predicting the Probability of LifeLimiting Fatigue Failures,” to be submitted, Engineering Fracture Mechanics S. K. Jha, C. J. Szczepanski, R. John, and J. M. Larsen, “Deformation heterogeneities and their role in life limiting fatigue failures in a two-phase titanium alloy,” Acta Materialia, Vol. 82, pp. 378-395, 2015. A. L. Hutson, S. K. Jha, W. J. Porter, and J. M. Larsen, “Activation of life-limiting fatigue damage mechanisms in Ti-6Al2Sn-4Zr-6Mo,” International Journal of Fatigue, Vol. 66, pp. 1-10, 2014. S. K. Jha, R. John, and J. M. Larsen, “Incorporating small fatigue crack growth in probabilistic life prediction: Effect of stress ratio in Ti-6Al-2Sn-4Zr-6Mo,” International Journal of Fatigue, Vol. 51, pp. 83-95, 2013. J. M. Larsen, S. K. Jha, C. J. Szczepanski, M. J. Caton, R. John, A. H. Rosenberger, D. J. Buchanan, P. J. Golden, and J. R. Jira, “Reducing uncertainty in fatigue life limits of turbine engine alloys,” International Journal of Fatigue, Vol. 57, pp. 103-112, 2013. C. J. Szczepanski, S. K. Jha, P. A. Shade, R. Wheeler, and J. M. Larsen, “Demonstration of an in situ microscale fatigue testing technique on a titanium alloy,” International Journal of Fatigue, Vol. 57, pp. 131-139, 2013. C. J. Szczepanski, P. A. Shade, M. A. Groeber, J. M. Larsen, S. K. Jha, and R. Wheeler, “Development of a microscale fatigue testing technique,” Advanced Materials and Processes, Vol. 171, pp. 18-21, 2013. M. E. Burba, M. J. Caton, S. K. Jha, and C. J. Szczepanski, “Effect of aging treatment on fatigue behavior of an Al-Cu-MgAg alloy,” Metallurgical and Materials Transactions A, Vol. 44, pp. 4954-4967, 2013. S. K. Jha, C. J. Szczepanski, P. J. Golden, W. J. Porter, III, and R. John, “Characterization of fatigue crack initiation facets in relation to lifetime variability in Ti-6Al-4V,” International Journal of Fatigue, Vol. 42, pp. 248-257, 2012. C. J. Szczepanski, S. K. Jha, J. M. Larsen, and J. W. Jones, “The role of local microstructure on small fatigue crack propagation in an a+b titanium alloy, Ti-6Al-2Sn-4Zr-6Mo,” Metallurgical and Materials Transactions A, Vol. 43, pp. 40974112, 2012. A. H. Rosenberger, D. J. Buchanan, D. A. Ward, and S. K. Jha, “The variability of fatigue in notched bars of IN100,” Superalloys 2012, pp. 143-148, 2012. S. K. Jha, C. J. Szczepanski, C. P. Przybyla, and J. M. Larsen, “The hierarchy of fatigue mechanisms in the long-lifetime regime,” VHCF-5, pp. 505-512, 2011. Approved for public release: Case No. 88ABW-2015-0198 54 Related Publications • • • • • • • • • • • • C. J. Szczepanski, S. K. Jha, J. M. Larsen, and J. W. Jones, “The role of microstructure on sequential stages of the very high cycle fatigue behavior of an a+b titanium alloy, Ti-6Al-2Sn-4Zr-6Mo,” VHCF-5, pp. 225-230, 2011. M. J. Caton and S. K. Jha, “Small fatigue crack growth and failure mode transitions in a Ni-base superalloy at elevated temperature,” International Journal of Fatigue, Vol. 32, pp. 1461-1472, 2010. R. John, D. J. Buchanan, M. J. Caton, and S. K. Jha, “Stability of shot peen residual stresses in IN100 subjected to creep and fatigue loading,” Procedia Engineering, Vol. 2., pp. 1887-1893, 2010. S. K. Jha, R. John, and J. M. Larsen, “Nominal vs local shot-peening effects on fatigue lifetime in Ti-6Al-2Sn-4Zr-6Mo,” Metallurgical and Materials Transactions A, Vol. 40, pp. 2675-2684, 2009. R. John, D. J. Buchanan, S. K. Jha, and J. M. Larsen, “Stability of shot-peen residual stresses in an a+b titanium alloy,” Scripta Materialia, Vol. 61, pp. 343-346, 2009. S. K. Jha, H. R. Millwater, and J. M. Larsen, “Probabilistic sensitivity analysis in life prediction of an a + b titanium alloy,” Fatigue and Fracture of Engineering Materials and Structures, Vol. 32, pp. 493-504, 2009. S. K. Jha, J. M. Larsen, and A. H. Rosenberger, “Towards a physics-based description of fatigue variability behavior in probabilistic life prediction,” Engineering Fracture Mechanics, Vol. 76, pp. 681-694, 2009. C. J. Szczepanski, S. K. Jha, J. M. Larsen, and J. W. Jones, “Microstructural influences on very-high-cycle fatigue-crack initiation inTi-6246,” Metallurgical and Materials Transactions A, Vol. 39, pp. 2841-2851, 2008. S. K. Jha, M. J. Caton, and J. M. Larsen, “Mean vs. life-limiting fatigue behavior of a nickel-based superalloy,” Superalloys-2008, pp. 565-572, 2008. W. J. Porter III, K. Li, M. J. Caton, S. K. Jha, B. B. Bartha, and J. M. Larsen, “Microstructural conditions contributing to fatigue variability in P/M nickel-base superalloys,” Superalloys-2008, pp. 541-548, 2008. S. K. Jha, M. J. Caton, and J. M. Larsen, “A new paradigm of fatigue variability behavior and implications for life predictions,” Materials Science and Engineering A, Vol. 468, pp. 23-32, 2007. S. K. Jha and J. M. Larsen, “Random heterogeneity scale and probabilistic description of the long-lifetime regime of fatigue,” VHCF-4, pp. 385-396, 2007. Approved for public release: Case No. 88ABW-2015-0198 55 Related Publications • • • • • • • • • • • • C. J. Szczepanski, S. K. Jha, J. M. Larsen, and J. W. Jones, “The role of microstructure on the fatigue lifetime variability in an a+b titanium alloy, Ti-6Al-2Sn-4Zr-6Mo,” VHCF-4, pp. 37-44, 2007. S. K. Jha, J. M. Larsen, and A. H. Rosenberger, “The role of competing mechanisms in the fatigue-life variability of a titanium and gamma-TiAl alloy,” JOM, Vol. 57, pp. 50-54, 2005. S. K. Jha, M. J. Caton, J. M. Larsen, A. H. Rosenberger, K. Li, and W. J. Porter, “Superimposing mechanisms and their effect on the variability in fatigue lives of a nickel-based superalloy,” Materials Damage Prognosis, TMS, pp. 343-350, 2005. S. K. Jha, J. M. Larsen, and A. H. Rosenberger, “The role of competing mechanisms in fatigue life variability of a nearly fully-lamellar g-TiAl based alloy,” Acta Materialia, Vol. 53, pp. 1293-1304, 2005. K. S. Ravi Chandran and S. K. Jha, “Duality of the S-N fatigue curve caused by competing failure modes in a titanium alloy and the role of Poisson defect statistics,” Acta Materialia, Vol. 53, pp. 1867-1881, 2005. C. Annis, J. M. Larsen, A. H. Rosenberger, S. K. Jha, and D. H. Annis, “RFTh, a random fatigue threshold probability density for Ti6246,” Materials Damage Prognosis, TMS, pp. 151-156, 2005. C. J. Szczepanski, A. Shyam, S. K. Jha, J. M. Larsen, C. J. Torbet, S. J. Johnson, and J. W. Jones, “Characterization of the role of microstructure on the fatigue life of Ti-6Al-2Sn-4Zr-6Mo using ultrasonic fatigue,” Materials Damage Prognosis, TMS, pp. 315-320, 2005. S. K. Jha, J. M. Larsen, and A. H. Rosenberger, “The role of fatigue variability in life prediction of an a+b titanium alloy,” Materials Damage Prognosis, TMS, pp. 1955-1960, 2005. S. K. Jha, J. M. Larsen, A. H. Rosenberger, and G. A. Hartman, “Mechanism-based variability in fatigue life of Ti-6Al-2Sn4Zr-6Mo,” Journal of ASTM International, Vol. 1, 2004. M. J. Caton, S. K. Jha, A. H. Rosenberger, and J. M. Larsen, “Divergence of mechanisms and the effect on the fatigue life variability of Rene’88DT,” Superalloys-2004, pp. 305-312, 2004. S. K. Jha, J. M. Larsen, A. H. Rosenberger, and G. A. Hartman, “Dual fatigue failure modes in Ti-6Al-2Sn-4Zr-6Mo and consequences on probabilistic life prediction,” Scripta Materialia, Vol. 48, pp. 1637-1642, 2003. S. K. Jha and K. S. Ravi Chandran, “An unusual fatigue phenomenon: duality of the S-N fatigue curve in the b titanium alloy Ti-10V-2Fe-3Al,” Scripta Materialia, Vol. 48, pp. 1207-1212, 2003. Approved for public release: Case No. 88ABW-2015-0198 56