Learning from the Past?

Fatigue Failures in Engineered

Systems
David K. Matlock
Advanced Steel Processing and Products Research Center
Colorado School of Mines
Golden, Colorado
The Hatfield Memorial Lecture
December 2, 2014
Why title: “Learning from
the Past?”
 Railroad Axle Failure: circa 1844
Fracture at change in diameter = stress concentration

• Mid-1800’s – Wöhler (and others)

• showed that fatigue occurs by crack growth from
surface defects
• Developed apparatus for repeated loading of railway
axles
• Contributions led to the S-N or “Wöhler” curve
• Result: improved understanding of fatigue.

Image - Original: Joseph Glynn, Paper No 617, Proc. ICE en:1844.

commons.wikimedia.org/wiki/File:Tender_fatigued_axle.JPG. en.wikipedia.org/wiki/August_W%C3%B6hler
Railroad Axle Failure: 2004
Conclusion: “..Fatigue
fracture originated at a
…surface profile irregularity...
likely introduced during axle
reconditioning..”

Final
Fracture
Railroad Axle Failure: 2010
Conclusion: “... The axle
failed in fatigue near the
mid-point of the axle body…”
 Presentation Overview
• Introduction: “what is fatigue?”
• “Modification of material strength and
fracture characteristics by the cyclic
application of load or stress, often leading
to fracture without prior component
shape change ”

• Present a “primer” on fatigue

• Case Studies
• Fatigue enhancement via metallurgy
• Design and application
 Fatigue Potential: Our Daily Lives
Transportation

www.netcarshow.com airplanesihaveknown.blogspot.com inhabitat.com

Recreation

sandiegomountainbikeskills.com
www.world-insider.com/usa-the-best-amusement-parks/

Energy

en.wikipedia.org/wiki/ www.lusas.com
awcwire.wordpress.com/2009/08/10/how-
Wind_turbine
wind-turbines-work/ taflab.berkeley.edu/ME168-FA13/ME168_Applications.htm
 How many cycles do we “experience”?

Passenger Car Engine

100,000 mile “typical” use
Average 40 mph @ 2000 rpm
300 million revolutions

Truck Wheels/Axles
1 million mile “typical” life
≈ 500 wheel rotations/mile
500 million rotations
 Primer on Fatigue

• Types of loading..
• Material property changes due
to cyclic loading….
• How to measure?
• How to control?
Examples of Cyclic Loading: Axial

Unidirectional Loading

Load or Stress Time

Reversed Loading
Load or Stress

Time
 Examples of Cyclic Loading: Bending
Reversed Bending
Example: Leaf Spring

Tension

Reversed Bending

Load or Stress
Compression
Time
Compression

Tension
 Examples of Cyclic Loading: Combined
Rotational Bending
Example: Drive shaft out of alignment

Tension Rotation

Applied Bending
Compression

Stress at point as shaft rotates

Load or Stress

Time
 Effects of Cyclic Loading on Strength
• Cyclic stress-strain behavior
• Measure load & displacement in sample cyclically
loaded from tension to compression

Pure Copper
Fully Annealed Cold Worked
Stress

Strain
J.D. Morrow, Cyclic Plastic Strain Energy and Fatigue of Metals. In: American Society for
Testing and Materials - ASTM STP 378. Internal Friction, Damping and Cyclic Plasticity 1965;
p. 45–87.
 Effects of Cyclic Loading on Strength
Monotonic σ-ε
SAE 4340 Steel
Cyclic σ-ε

R.W. Landgraf, Achievement of High Fatigue Resistance in Metals and Alloys, ASTM STP-467, 1970, p. 3.
 Effects of Cyclic Loading on Strength

Cyclic Hardening Cyclic Softening

Aluminum
Steel
2024-T6
SAE 4340

Stress
Stress

Strain Strain

R.W. Landgraf, J.D. Morrow, and T. Endo, J. Materials, JMLSA 4(1), ASTM 1969, P. 176.
Effects of Cyclic Loading on Strength
“Big-Picture” Conclusion:
Hard (i.e. high strength) materials
cyclically soften --- while soft (i.e. low
strength) materials cyclically harden!

Cyclic Softening

Cyclic Hardening
Stress

Strain
 Effects of Cyclic Loading on Fracture
Three “stages” of fatigue
• Crack nucleation - at point of high applied
stress – results from local plastic
deformation after multiple cycles
• Stable crack growth - on plane perpendicular
to the maximum tensile stress
• Final fracture - after crack grows to critical
length -- i.e. remaining material can no longer
support applied cyclic loads
• Total Fatigue Life:
NTotal = NNucleate + NGrowth + 1Overload
 Effects of Cyclic Loading on Fracture
Fatigue Crack Nucleation and Growth

R.A. Lund, “Fatigue Fracture

Appearances,” ASM Handbook, Vol. 10 mm
11, 2002. p. 627.
 Effects of Cyclic Loading on Fracture
Fatigue Crack Nucleation and Growth

10 µm

TEM Replica: Low Cycle Fatigue

7075 Al – T6 Aluminum
R.D. Sloan, Sloan Research Inds. Inc (Circa 1970)

10 mm
R.A. Lund, ASM Handbook,
Vol. 11, 2002
 Unidirectional
Tension-tension Loading

Stable Fatigue
Crack Growth

5 mm
Important points…..Effects of Cyclic Loading
• Strength altered
• Crack nucleation and growth leads to failure at
low stress (e.g. often less than yield stress)
• Stable crack growth exists prior to fracture
• Occurs without macroscopic geometry
change
• Grows on plane perpendicular to maximum
tensile stress
• Presence offers the opportunity to utilize
non-destructive testing techniques to
identify prior catastrophic failure
Evidence Fatigue is Critical
to Our Daily Lives
 1951
Starring
James Stewart

www.metacafe.com/watch/7743905/no_highway_in_the_sky_1951/ (accessed Nov 2014)

 Life Imitates Movie…
• De Havilland Comet 1
• Innovative airplane
• Commercial service www.telegraph.co.uk
• Initiated 1952
• Operated at 40,000 feet
• Cabin pressurized, 8000 ft equivalent
• Two catastrophic accidents 1954
• Royal Aircraft Establishment pressurization tests
confirmed cabin structural failure by fatigue
• Required significant redesign
• Opened the way for modern design and testing
concepts.

P.A. Withey, “Fatigue Failure of the De Havilland Comet I,” Engr. Fail. Anal., vol. 4, no. 2, 1997, pp. 147-154.
 Aloha Airlines, Flight 243
April 28, 1988

The National Transportation Safety Board (NTSB)

determined that the probable cause of the accident was
…… fatigue damage of the fuselage skin lap splice.
lessonslearned.faa.gov (accessed Nov 2014)
 Flight 232 - Sioux City, Iowa – July 19, 1989

• Turbofan engines - fan disk failure – Ti alloy.

• Undetected defect formed during initial manufacture (Dec. 1971).
• Defect caused the initiation of a fatigue crack
• Crack grew to a critical size ----- catastrophic failure
• Disk parts damaged hydraulic control systems
• Total service time = 41,009 hours and 15,503 cycles (i.e. flights)
lessonslearned.faa.gov (accessed Nov2014)
¾ inch (19 mm) diameter bolts
 “…fracture surfaces…of three
…bolts…indicated fatigue
Adopted
October 1, 1991 cracks initiating at multiple
sites along the thread roots on
diametrically opposite sides of
the bolts”

≈ 10 mm
Methods to Assess Fatigue
Fracture Properties

Fatigue Life Tests (S-N)

Fatigue Crack Growth
 Fatigue Life Curves
• Multiple standardized tests available
• Specialized tests designed to simulate
in-service conditions

ASPPRC, Colorado School of Mines, Golden, CO USA

 Fully Reversed Test; Frequency = 30 Hz

Video starts
after
2280 cycles

L.M. Rothleutner and D.K. Matlock, ASPPRC, Colorado School of

Mines, Golden, CO USA, 2014
 Failure life = 2750 cycles

L.M. Rothleutner and D.K. Matlock, ASPPRC, Colorado School of

Mines, Golden, CO USA, 2014
 Failure life = 2750 cycles

L.M. Rothleutner and D.K. Matlock, ASPPRC, Colorado School of 5 mm

Mines, Golden, CO USA, 2014
 Typical “S-N” Data
900 130
Direct cooled “Non-traditional”
NTB
Nominal Reversed Bending Stress (MPa)

2750 Bainitic Steel 120

Nominal Reversed Bending Stress (ksi)

800
cycles 0.34 C, 1.21 Mn, 0.66 Si, 0.09 V
25HRC; 15% retained austenite 110
700 100

90
600
8027
500
70

400 60

50
Baseline Fatigue Limit
300
(3) 40 Or
200 30 Endurance Limit
103 104 105 106 107 108
Cycles

M.D. Richards, M. Burnett, J.G. Speer, and D.K. Matlock, Metall. and Mat. Trans. A, 2013, vol. 441, pp. 270-285.
 Krouse-type Bending Fatigue
• Displacement controlled; constant frequency
• Large constant stress region
• Variable R
–1 to 1
• Flat samples

5 cm

ASPPRC, Colorado School of Mines, Golden, CO USA

 Bending Fatigue of Spring Steel

426 oC Temper

500 oC Temper

WQ and AC indicate cooling

after tempering
As-Quenched

N. Merlano, Effect of Tempering Conditions On The Fatigue and Toughness of 5160H Steel, MS Thesis, Colorado School of Mines, 1989
 Fatigue Crack Growth Analysis
• Fracture mechanics based approach
• Assume material contains a crack (flaw, notch,..)
• Machine standard sample
• Impose cyclic tensile load
• Measure change in crack length (da) with each cycle
(dN)
• Correlate:
• da/dN = f(ΔP) = f’(Δσ) = f’’(ΔK)
• Where:
• P = load
• σ = stress = (load/area)
• K = stress intensity factor ∝ σ·g(crack geometry)
 Fatigue Crack Growth Analysis
ΔP

da a

www.fracturemechanics.net T.L. Anderson, Fracture Mechanics: Fundamentals and

(accessed Nov 2014) Applications, CRC Press, Boca Raton, Florida, 1991, p. 603.
 Potential to Alter Stable Crack Growth

10-5

da/dN = 1.36 x 10-10 (∆K)2.25

Single
Function!
10-6
da
= A (∆K )m
da/dN (m/cycle)

dN

10-7 Tempered Martensitic

Steels
Applicability of data:
12 Ni STEEL
10-8 10 Ni STEEL Yield = 560 to 2070 MPa
HY-130 STEEL Ambient temperature
HY-80 STEEL
Dry air
10-9
1 10 100
Adapted from: J.M. Barsom and S.T. Rolfe, Fracture and
∆K (MPa√ m) Fatigue Control in Structures, 2nd Edition (1987), Prentice-Hall,
Englewood Cliffs, New Jersey, p. 287
Stable Fatigue
Crack Growth

Plastic
Zone

5 mm
ASPPRC, Colorado School of Mines, Golden, CO USA
 Interpretation of Single da/dN Function

D.K. Matlock, ASPPRC, Colorado School of Mines, Golden, CO USA, 2009.

 Interpretation of Single da/dN Function
Apply stress = plastic zone

A 1  KI 
2
rp =
6π  σ y 

D.K. Matlock, ASPPRC, Colorado School of Mines, Golden, CO USA, 2009.

 Interpretation of Single da/dN Function
Apply cyclic stress = plastic zone advances

A 1  KI 
2
rp =
6π  σ y 

da
dN
Crack advances

D.K. Matlock, ASPPRC, Colorado School of Mines, Golden, CO USA, 2009.

 Interpretation of Single da/dN Function
Apply cyclic stress = plastic zone advances

A 1  KI 
2
rp =
6π  σ y 

rp
= 10 to 2000
da da
dN
dN
Growth controlled by “cyclic stress strain”
“Hard materials cyclically soften”
“Soft materials cyclically harden”
D.K. Matlock, ASPPRC, Colorado School of Mines, Golden, CO USA, 2009.
 Interpretation of Single da/dN Function
Apply cyclic stress = plastic zone advances

Conclusion: Limited
A
opportunity to influence 1  KI  2
rp =  
fatigue life through control6of π  σ y 
fatigue crack growth rates via
rp
metallurgy modifications= -- 10 to 2000
da
 must address crackda
dN
dN nucleation!
Orcontrolled
Growth crack growth by design!
by “cyclic stress strain”
“Hard materials cyclically soften”
“Soft materials cyclically harden”
D.K. Matlock, ASPPRC, Colorado School of Mines, Golden, CO USA, 2009.
 Lessons Learned – Lab Tests
• Summary of approaches to produce structures
with enhanced fatigue performance
• Decrease surface cyclic tensile stress
• Remove Loads!!
• Remove Cycles!!
• Minimize stress concentrations
• Design
• Manufacturing
• Induce residual compressive stress
• Increase material strength ( EL ∝ UTS )
• Bulk or surface
• Maximize material “quality” i.e. minimize
inclusion contents, etc.
 Examples: Metallurgical
Modifications to Control Crack
Nucleation

• Process Control
• Deep Rolling - Shafts
• Alloy Control
• Steel Cleanliness – Bearings
• Microalloying - Gears
 Drivers: Future Automobile Engines
• Lighter weight + higher performance = higher stresses
• High-strength fatigue-resistant materials facilitate designs
 Deep Rolling: Crankshafts
M.D. Richards, PhD Thesis, Colorado School of Mines, USA, 2008.

Connecting Rod

www.driving-test-success.com/how-cars-work.htm en.wikipedia.org/wiki/Crankshaft
 Single Cylinder
Crankshaft

A. Fatemi, et al., “Fatigue Performance Evaluation of Forged Steel Vs. Ductile Cast
Iron Crankshafts: A comparative Study,” U. of Toledo, 2007, www.autosteel.org.
 Deep Rolling Laboratory Sample
Sample Diameter = 25 mm

M.D. Richards, M. Burnett, J.G. Speer,

and D.K. Matlock, “Effects of
Deformation Behavior on the Fatigue
Performance of Deep Rolled Medium
Carbon Steels,” Metallurgical and
Materials Transactions A, 2013, vol.
441, pp. 270-285

A. Fatemi, et al., “Fatigue Performance Evaluation of Forged Steel Vs. Ductile Cast
Iron Crankshafts: A comparative Study,” U. of Toledo, 2007, www.autosteel.org.
 Deep Rolling

M.D. Richards, The Effects of Deformation Behavior on the Fatigue Performance of Deep
Rolled Medium Carbon Steels, PhD Thesis, Colorado School of Mines, USA, 2008.
 Deformation during Deep Rolling
Roller

Geometry Change Due

to Deformation

Residual
Stress Notch
Constraint

Strain Deformation
Volume

• Radially symmetric, non-uniform strain

• Increases local strength
• Mechanically burnishes surface
• Develops residual stress
• Residual stress stability depends on response to
cyclic loading M.D. Richards, The Effects of Deformation Behavior on the Fatigue Performance of Deep Rolled
Medium Carbon Steels, PhD Thesis, Colorado School of Mines, USA, 2008.
 Test Methodology: R = 1, Freq. = 30 Hz
Sample Diameter = 25 mm

M.D. Richards, The Effects of Deformation Behavior on the Fatigue Performance of Deep
Rolled Medium Carbon Steels, PhD Thesis, Colorado School of Mines, USA, 2008.
 Baseline Fatigue Performance

900 130
4140
4140 Steel Three Steel
Nominal Reversed Bending Stress (MPa)

120

Nominal Reversed Bending Stress (ksi)

800
110 Alloys
700 100 Fatigue Ratio
Alloy
90 EL/UTS
600
80 4140 0.49
500
70 NTB 0.47
60 C38M 0.43
400
Baseline 50
300 (3)
(3) 40

200 30
103 10 4 5
10 6
10 10 7
108
Cycles

M.D. Richards, M. Burnett, J.G. Speer, and D.K. Matlock, “Effects of Deformation Behavior on the Fatigue Performance
of Deep Rolled Medium Carbon Steels,” Metallurgical and Materials Transactions A, 2013, vol. 441, pp. 270-285
 Deep Rolled Fatigue Performance
1100
150
41404140
Steel
Nominal Reversed Bending Stress (MPa)

1000

Nominal Reversed Bending Stress (ksi)

140
900 130 Nominal
Fatigue Ratio
Alloy
120 Endurance Limit
800 kt*Sf-DR/UTS
110
Sf-DR (MPa)
700 4140
100 469 0.74
(3)
600 90NTB 448 0.76
Deep Rolled 80
(2)
C38M 386 0.69
500 70
(3)

400 60
Baseline 50 Deep rolling
(3)
300 (3) 40 increases
200 30
104 105 106 107 endurance Limit
Cycles
by 50 to 60 %.
M.D. Richards, M. Burnett, J.G. Speer, and D.K. Matlock, “Effects of Deformation Behavior on the Fatigue Performance
of Deep Rolled Medium Carbon Steels,” Metallurgical and Materials Transactions A, 2013, vol. 441, pp. 270-285
 Processing to Optimize Fatigue Resistance
• Hypothesize – Fatigue resistance improved by
• Stabilization of cold worked dislocation
structure
• Stabilization of residual stress distribution
• Approaches to process modifications
• Age previously rolled samples
• Roll at dynamic strain aging temperatures
(up to about 350 oC)

M.D. Richards, M. Burnett, J.G. Speer, and D.K. Matlock, “Effects of Deformation Behavior on the Fatigue Performance
of Deep Rolled Medium Carbon Steels,” Metallurgical and Materials Transactions A, 2013, vol. 441, pp. 270-285
 Dynamic Strain Aging (DSA)
• Changes in Deformation Mechanisms
• Decrease dislocation mobility – pinning
• Increase dislocation density
• Change in dislocation structure from cellular to
diffuse tangles

ELONGATION
C.C. Li, and W. C. Leslie, “Effects of dynamic strain aging on the subsequent mechanical properties of carbon steels,” Metallurgical
Transactions A, December 1978, Volume 9, Issue 12, pp 1765-1775.
 Deep Rolled @ 340 oC
900
4140 Steel
4140

Nominal Reversed Bending Stress (MPa)

800

700
Deep Rolled - HT
600 (3)
(3)
Deep Rolled - RT
500 (2)
(3)

400
Baseline
300 (3)
(3)

200
103 104 105 106 107 108
Cycles
M.D. Richards, M. Burnett, J.G. Speer, and D.K. Matlock, “Effects of Deformation Behavior on the Fatigue Performance
of Deep Rolled Medium Carbon Steels,” Metallurgical and Materials Transactions A, 2013, vol. 441, pp. 270-285
 Summary: Deep Rolling

• Fatigue crack nucleation made more

difficult
• Deep rolling at elevated temperatures
increases EL by approximately 100%
• Processing at DSA temperatures proved
very cost effective to enhance fatigue
performance

M.D. Richards, M. Burnett, J.G. Speer, and D.K. Matlock, “Effects of Deformation Behavior on the Fatigue Performance
of Deep Rolled Medium Carbon Steels,” Metallurgical and Materials Transactions A, 2013, vol. 441, pp. 270-285
 Examples: Metallurgical
Modifications to Control Crack
Nucleation

• Process Control
• Deep Rolling - Shafts
• Alloy Control
• Steel Cleanliness – Bearings
• Microalloying - Gears
 Fatigue in Gears and Bearings

Drive Contact
Gear

Bending
Driven
Gear

commons.wikimedia.org/wiki/File:Spur_gears_animation.gif
 Rolling Contact Fatigue in Bearings
Cleaner Steel
1000 Today 1980

Fatigue Life (Millions of Revolutions)

Vacuum Arc Remelted
Improved Bottom Pour

Original Bottom Pour

100

Precipitation Deoxidation
+ Shrouding

10 Precipitation Deoxidation

Vacuum Carbon Deoxidation

1
0.0001 0.001 0.01 0.1 1 10
Total Length of Inclusion Stringers (mm/cm3)
C.V Darragh, “Engineered Gear Steels – A
Review,” 2001 Drives and Controls/Power
Stress profile adapted from L.E. Alban, Electronics Conference, pp. 21-26.
P. Kramer, An Investigation of Rolling-Sliding Systematic Analysis of Gear Failures,
Contact Fatigue Damage of Carburized Gear American Society for Metals, Metals Park, OH
Steels, MS Thesis, CSM 2013 (1985), pp. 94–106
 Bending Fatigue: Gear Steels
• Utilize higher temperature carburizing – more
efficient (vacuum, plasma)
• Issue, need to suppress grain growth & refine
austenite grain sizes to increase performance

• Utilize microalloy (Nb) precipitates to suppress

grain growth G. Krauss, D.K. Matlock, and A. Reguly, “Microstructural Elements and
Fracture of Hardened High-Carbon Steels”, Proc. of Thermec 2003, Trans
Tech Publications, Inc., Uetikon-Zurich, Switzerland, 2003, pp. 835-840
0.1 Nb
Nb-Ti Modified 8620 Steel :
Vacuum Carburized @1050 oC

100 µm
1400
0.06 Nb 1300 All Alloys - 114 ºC min-1

1200

1100

Stress (Mpa)
1000 0.1 Nb

100 µm 900

800 0.06 Nb
0.02 Nb
700
0.02 Nb
600 b
500
3 4 5 6 7
10 10 10 10 10
Cycles
R.E. Thompson, D.K. Matlock, and J.G. Speer, "The Fatigue Performance of High Temperature Vacuum
100 µm Carburized Nb Modified 8620 Steel," SAE Transactions, Journal of Materials and Manufacturing, Vol. 116,
Sect. 5 (2007) pp. 392-407.
 Selected Case Studies to
Illustrate Engineering Solutions
to Fatigue Failures……..
importance of design,
manufacturing, and
maintenance
 Design Example:
Fatigue Failure in
Bullwheel Axle Shaft
Lower Terminal – Bullwheel Axle Failure

Hub Sheave

Main Bullwheel
Shaft
D.K. Matlock, "Lift Fatigue,” Ski Area Management, vol. 23, no. 1, 1984, pp. 62
63, 80 (http://www.saminfo.com/article/lift-fatigue).
 Bullwheel Shaft
Dia = 5 ¼ inch
(13.3 cm)

Location
Crack
D.K. Matlock, "Lift Fatigue,” Ski Area Management, vol. 23, no. 1, 1984, pp. 62
63, 80 (http://www.saminfo.com/article/lift-fatigue).
 Hub Sheave

Main Bullwheel
Shaft
D.K. Matlock, "Lift Fatigue,” Ski Area Management, vol. 23, no. 1,
1984, pp. 62 63, 80 (http://www.saminfo.com/article/lift-fatigue).
 Hub Sheave

Main Bullwheel
Shaft
D.K. Matlock, "Lift Fatigue,” Ski Area Management, vol. 23, no. 1,
1984, pp. 62 63, 80 (http://www.saminfo.com/article/lift-fatigue).
Have we learned anything from
the past?

What about the future?

 Closing Comments
• So….
“Why do fatigue failures continue to occur?”
• Multiple “inputs” affect fatigue performance
• Design
• Material
• Manufacture
• Maintenance
• Application/Use
• Fatigue fractures will continue to occur!
 Closing Comments
• Opportunities exist for continued development
of high-performance “clean” materials
• Inspection
• Opportunities for “smart” NDE technologies
to identify cracks before catastrophic failure
• Continual “fatigue” education critical
• All “parties” involved must appreciate factors
which control fatigue life
• … still need good “Common Sense
Engineering”…