fatigue: its measurement & applications suresh gulati october 9, 2006
TRANSCRIPT
Fatigue: Its Measurement & Applications
Suresh Gulati
October 9, 2006
Definition
Materials, when stressed above threshold level, experience flaw growth and become weak, i.e. less durable. This phenomenon is called FATIGUE.
Types of Stress
• Constant Stress (static fatigue)
• Constant Stress Rate (dynamic fatigue)
• Cyclic Stress (cyclic/dynamic fatigue)
Fatigue Mechanism Summary
SiO2 + H2O +Stress Si (OH)4
Hence, stress becomes the catalyst, i.e. no stress, no fatigue !
Strong oxide bonds become weak hydroxyl bonds !!
Necessary Elements for Fatigue
1. Stress
2. Flaw
3. Water Vapor
4. Time
Role of Water Vapor
• Small molecule (< 3 Ao )
• Easily fits the cavity of 6-member SiO2
tetrahedra
• Need only few molecules of H2O
• Hence fatigue can occur at low RH
• Higher the humidity, higher the fatigue
• No fatigue in vacuum
Role of Stress
• Stress is the catalyst
• Higher the stress, higher the fatigue
Role of Temperature
• Vibrational energy of H2O molecule vs. T
• Chemical reaction of SiO2 with H2O vs. T
• Vibrational energy of glass forming oxides,SiO2, Na2O, CaO, B2O3, etc. vs. T
• Fatigue vs. Coeff. Thermal Expansion
• Fatigue at subzero temperature
Role of Stress Duration
• (SiO2 + H2O + stress) reaction requires time
• Short duration of time means less fatigue
• Long duration of stress means more fatigue
Measurement of Strength
• 4-point Bend Test (uniaxial)
• Tensile Test (uniaxial)
• Ring-on-Ring Test (biaxial)
• ASTM Standard
• Test Duration
• Test Environment
St. Venant Flexure
• M/I = E/R = /y
• MOR = f = 1.5 P (L – l ) / (b t2)
P = load at failureL = support spanl = load spanb = width of speciment = thickness of specimen
Strength Distribution
• Gaussian
• Weibull
Pf = 1 – exp [ - ( / o)m ]
Pf : failure probability at stress o : stress at Pf = 0.63
m : slope of (lnlnPf vs. ln )
Weibull Distribution
• Unimodal (single value of m, i.e. uniform flaw)• Multimodal (multiple values of m, i.e. multiple
families of flaw)
• Physical meaning of m: approx.(mean f / std.dev)
• Fatigue is difficult to measure for multiple families of flaw
Measurement of Dynamic Fatigue
d/dt
(MPa/s)
N
(MPa)
m t
(s)
3.0 20 75 15 25
0.3 20 68 18 250
0.03 20 59 16 2500
0.03 20 45 19 25000
0.003 20 30 17 250000
Weibull Distribution Plot
4000
4500
5000
5500
6000
6500
7000
7500
8000
1
2
5
10
20
40
60
80909598
99.5
PPAD plot: 05/31/06
Data Summary Set1: (19 Specs) Mean=7.01E3; Stdev=253; m=33.4; S
0=7.12E3
Set2: (19 Specs) Mean=6.05E3; Stdev=315; m=23.0; S0=6.19E3
Set3: (19 Specs) Mean=5.46E3; Stdev=321; m=21.0; S0=5.6E3
Set4: (19 Specs) Mean=4.89E3; Stdev=298; m=20.7; S0=5.01E3
Set5: (4 Specs) Mean=4.47E3; Stdev=120; m=44.7; S0=4.52E3
WeiPPAD 5.010 (01/29/03) (XOnProb) (MedianEst)
WEIBULL DISTRIBUTION
Fai
lure
Pro
ba
bili
ty
STRENGTH (psi)
Weibullized Eagle SG dynamic fatigue at each rate, fastest rate (#1) has the strongest distribution.
Computation of Fatigue Constant
• Let = median strength at (d/dt)1
• Let = median strength at (d/dt)2
• Then
1 / 2 = [ (d/dt)1 / (d/dt)2 ]1/(n +
1)
lnvs. ln (ddt) Plot
100µ 1m 10m 100m 1 10 100 1k 10k3k
4k
5k
6k
7k
8k
9k
RaMS plot: 4/13/2006
Power Law: n = 19.195% c.i.: 17.6, 20.981 Specimens [D = 3.71]
Str
en
gth
(p
si)
Stress Rate (psi/s)
1
n+1
Dynamic Fatigue Constants for Silicate Glasses (ambient environment)
Glass Code Composition Avg. n Value CTE (25-300C)
9061 CTV Panel 14 98
0080 SLS 16 80
7740 Pyrex 27 32
7059 Color Filter 28 38
1737 LCD 24 37
1723 SAS 30 36
7940 FS 37 5.5
7971 ULE 45 0.3
Plot of n vs. CTE
10 15 20 25 30 35 40 45 500
20
40
60
80
100
CTE ~ 1/n
CT
E (
10-7/C
)
Fatigue Constant, n
Physical Meaning of n
n Value Strength loss due to 10X lower (d/dt)
15 13.4 %
20 10.4 %
25 8.5 %
30 7.2 %
35 6.2 %
40 5.5 %
Measurement of Static fatigue
• Apply static (constant) stress to 4-point bend specimen and measure time tf till it fails
• Plot lnvs. tf
• Estimate n value from
1n tf1 = 2
n tf2
n = [ (ln tf2 – ln tf1) / (ln 1 – ln 2) ]
Plot of ln vs. ln tf
** *
* * * *
* * *
* * *
ln (time)
ln(stress)
*
*
**
1/n
1
Comparison of Static vs. Dynamic n Values
• Dynamic n > Static n
• Code 9061 glass: nd = 22, ns = 14
• CO diesel filter: nd = 30, ns = 15
Allowance for Fatigue Damage( LD Diesel Filter )
• Recall n tf = constant• New Filter: 1 = o, tf = to = 1 sec.• AT Filter for LDD application:
n = 70, Life = 200,000 KmRegen. Freq’y: 300 Km
max Duration: 75 sec. each regen.Find fatigue factor & usable strength
LD Diesel Filter (cont’d)
• Total regens = 200,000 Km / 300 KM = 667• Total stress duration = 667 x 75 s = 50,000 s.
• Useable strength = 2 as shown below
2n x 50,000 = 1
n x 1
2 = 1 [ 1 / 50,000 ]1/n
= 0.86 1 = 0.86 o
Fatigue Factor = 0.86
Allowance for Fatigue Damage( HD Diesel Filter )
• CO Filter for HDD applicationn =30, Life = 700,000 KmRegen. Freq’y = 400 Km
max Duration = 100 sTotal Regens = 1750Total Stress Duration = 175,000 s
2 = [ 1/175,000 ]1/30 = 0.67 o
Fatigue Factor = 0.67
Allowance for Stressed Area( HD Diesel Filter )
• Assume 9” diameter x 12” long filter
• Assume m = 15
• Area Factor = [ Aspec / Afilter ]1/m
• Aspec = 0.75” x 1” = 0.75 in2
• Afilter = 3.14 x D x L = 113 in2
• Area Factor = 0.72
Allowance for Acceptable Pf
( HD Diesel Filter )
• Failure Probability Factor = ( Pf )1/m,
• m = 15
• FPF = 0.54 for Pf = 0.0001 (0.01 % fail)
• FPF = 0.46 for Pf = 0.00001 (0.001% fail)
Final Useable Strength of CO Filter for HD Diesel Filter
• Useable Strength = o x FF x AF x FPF
= 0.26 o for Pf = 0.01%
= 0.22 o for Pf
=0.001%
Threshold Stress(fatigue free stress)
• At certain value of stress, known as threshold stress, the critical flaw does not propagate
• Materials can sustain threshold stress without becoming weak, hence we may call that “happy stress”
Static Fatigue Test for Estimating Threshold Stress
• Use 4-point bend test at 400C• Measure MOR of 15 specimens• Assume n =ndyn and estimate allowable stress (eas) for 9 x 12 filter
at Pf = 0.001%• Apply static stress = eas on 15 specimens
= eas + 50 psi on 15 specimens
= eas + 75 psi on 15 specimens• Hold above static stresses for 100 hours at 400C• Record any premature failures• Remove stress• Measure MOR of surviving specimens for each set• Compare MOR distributions before and after static stress• Expect max. loss of strength at highest static stress and none at eas
Static Fatigue Fixture with Multiple Specimens
Static Fatigue Fixture in 400C Oven
Comparison of Strength Distributions after Static Load Test CO-E1(axial) Static Load then Residual MOR at 400oC
100100 150 200 250200
250
300
350
400
450
500
Baseline
static breaks 664
Res
idua
l Str
engt
h (p
si)
Static Stress (psi)
Strength Data for CO 200/12
• Baseline: 353 psi (s = 0 )
• MOR res : 296 psi ( s = 150 psi )
• MOR res : 243 psi ( s = 175 psi )
• MOR res : 187 psi ( s = 200 psi )
Crack Growth during Static Loading
ar / ai = [ MORi / MORres ]2
• ar / ai = 1.42 @ 150 psi
= 2.11 @ 175 psi
= 3.56 @ 200 psi
Estimated n value: ~15
Static vs. Dynamic n Values
• Static value is lower than dynamic value due to longer time for stress corrosion reaction in humid environment
• Which n value should be used? This will depend on stress/time history for a given application
Applications of Fatigue Effect
• Color TV Panel
• Space Shuttle Windows
• Fiber Optics
• Catalytic Converters
• Diesel Filters
• LCD Panel
Acknowledgements
• John Helfinstine
• Janto Widjaja