Está en la página 1de 8

Assessment of fatigue striation counting accuracy using

high resolution scanning electron microscope


Emmanuel Hershko
*
, Nir Mandelker, George Gheorghiu, Haim Sheinkopf,
Izack Cohen, Ofer Levy
Materials Division, Aircraft Maintenance Unit, Israeli Air Force, P.O. Box 02538, Israeli Defence Forces, Israel
Received 2 January 2007; accepted 8 January 2007
Available online 26 January 2007
Abstract
A well-known method for determining the number of fatigue load cycles prior to failure is to perform a striation count-
ing on the fractured surface.
The primary objective of this study is to evaluate the accuracy of a striation counting performed using a high-resolution
scanning electron microscope. Fatigue experiments were conducted on two aluminum alloy AA-2024-T3 specimens and two
low alloy steel AISI-4130-O specimens. We then performed a fractographical analysis of the fractured specimens and com-
pared the results to the experimental data. The second objective of this study is to determine several guidelines regarding the
method of striation counting process which will raise its accuracy. This study shows that high accuracy can be achieved by
counting fatigue striation using a scanning electron microscope, but one must be aware of several problems and diculties
which can occur during the counting process. Several guidelines that will raise the accuracy were determined.
2007 Elsevier Ltd. All rights reserved.
Keywords: Fatigue; Fractography; Fatigue life; Fatigue testing
1. Introduction
Apparently the most common mode of failure in metallic parts is fatigue. It is particularly important in
aircraft parts.
Determination of a maintenance policy after the occurrence of a fatigue failure in service is based on nding
the number of loading cycles which caused the failure. Then, this information need to be related to a measur-
able parameter, e.g. ight hours, landings, rotor revolutions, etc.
The counting of striations was suggested many years ago as a method to estimate the duration of crack
propagation. Several authors claim a one-to-one correspondence between striation and stress cycle [13].
The striation width was also suggested for estimation of the stress amplitude [4]. However, other researchers
found that there is no simple correlation between striations and stress cycles [5]. In particular such correlation
fails at low amplitude of stress intensity factor [4] Where striations are found it is generally true that each
1350-6307/$ - see front matter 2007 Elsevier Ltd. All rights reserved.
doi:10.1016/j.engfailanal.2007.01.005
*
Corresponding author. Tel.: +972 57 8179956; fax: +972 8 8685298.
E-mail address: e_hershko@hotmail.com (E. Hershko).
www.elsevier.com/locate/engfailanal
Engineering Failure Analysis 15 (2008) 2027
striation was produced by one load cycle, on the other hand, it is not generally true that every load cycle pro-
duces a striation. The strong correlation of striation spacing and crack growth rate/cycle is only valid at med-
ium crack growth rates [4]. Special techniques are needed to obtain a good accuracy [1]. It has been recently
claimed that very ne striations are sometimes observed that are ignored in ordinary failure analysis [6].
The primary objective of this study was to evaluate the accuracy of a striation counting by a high-resolution
Scanning Electron Microscope to determine the number of load cycles during crack propagation.
Fatigue experiments were conducted on two aluminum alloy AA-2024-T3 specimens and two low-alloy
steel AISI-4130-O specimens. In each experiment, the number of applied load cycles was recorded and frac-
tographic analysis of the fractured specimens was done and compared to the directly measured cycles.
It is worth noting that the information revealed by studying the fracture surfaces (particularly the number
of load cycles prior to failure) relates only to the propagation stage of the fatigue crack life. Therefore, it is not
possible to determine the number of load cycles which occurred during the initiation stage.
The second objective of this study was to develop guidelines regarding the method of striation counting in
order to improve its accuracy.
2. Experimental
2.1. Measurement of crack growth rate (da/dN)
The rst stage of the experiment was to physically measure the crack growth rate in a specimen. Standard
specimens were prepared in accordance with ASTM-E647 (standard test method for measurement of fatigue
crack growth rates) from the commonly used in the aircraft industry aluminum alloy AA-2024-T3 and low
alloy steel AISI-4130 annealed (two specimens from each material).
In order to measure the rate of crack growth, a COD (Crack Opening Displacement) device was attached to
the edge of the specimens (Fig. 1). This device measures the displacement at the specimens edge during each
load cycle. Then the information is used to calculate the length of the crack, based on the specimen geometry
and the materials elasticity.
The specimens were cyclically loaded on MTS tension compression machine (Fig. 2) at constant ampli-
tude until fracture occurred. The ratio of the minimum load to the maximum load (R) was 0.1.
During the experiment, we conducted a continuous monitoring of the crack length (a) as a function of the
number of load cycles (N). It allowed the calculation of the crack growth rate (da/dN).
2.2. Experiment parameters determination
The magnitude of the load P applied on the specimen was calculated based on its geometry and material
elasticity in order to induce a reasonable crack growth rate, as shown in Fig. 3 and Eq. (1).
DP
DK b

w
p
f a
0
=w
1
Fig. 1. A COD device attached to a specimen.
E. Hershko et al. / Engineering Failure Analysis 15 (2008) 2027 21
2.3. Critical crack length
Based on the load chosen to be used in the experiment, the critical crack length of the specimen was cal-
culated by comparing Eq. (2) with Eq. (3):
Fig. 2. The specimen and device loaded on the tension compression machine.
Fig. 3. A schematic illustration of the experiment specimen.
22 E. Hershko et al. / Engineering Failure Analysis 15 (2008) 2027
f a
cr
=w
DK
1c
b

w
p
DP
2
f a
cr
=w
2
a
cr
w

0:886 4:64
a
cr
w
13:32
a
cr
w

2
14:72
a
cr
w

3
5:6
a
cr
w

4

1
a
cr
w

3=2
3
2 3 ) a
cr
=w ) a
cr
4
The parameters used and the calculation results are shown in Table 1.
2.4. Fractographic analysis of fracture surfaces
The fractured specimens were rst examined using a stereomicroscope in order to determine the length of
the fatigue crack (i.e., the critical crack length), the location of the fracture origin/origins and the crack prop-
agation path.
Next, the fracture surfaces were examined by JEOL JSM-7000F Field Emission Scanning Electron Micro-
scope (SEM). Three separate paths connecting the main origin to the nal rupture zone were chosen and 10
areas equally spaced along each path were photographed. The number of fatigue striations (N) in a unit of
length (a) was counted in each area and graphs were generated showing the relationship between the striation
densities (dN/da) in each area as a function of the distance of that area from the origin (L).
Note: A separate graph was generated for each of the three paths chosen on each specimen.
These graphs were then compared with the direct mechanical measurements.
3. Results
3.1. Preparation of the specimens and measurement of crack growth rate
The results of the measurements performed using the COD are summarized in Table 2.
It is important to note that the total number of load cycles applied to the specimens was measured from the
start of measurable propagation occurred at the indicated distance from the crack origin.
3.2. Fractographical analysis of fracture surfaces
As stated above, 3 lines leading from the origin to the nal rupture zone were chosen on each fracture sur-
face (denoted top, middle and bottom) and the striation density was measured at 10 dierent locations along
the lines. Based on this data, graphs were generated showing the crack density (dN/da) as a function of the
distance from the origin (L) a separate graph was made for each line. The total number of striations (i.e.,
load cycles) in each specimen was then calculated by approximating the area under the corresponding graph.
Fig. 4 shows a typical area in the fracture surface of one of the aluminum specimens. Fig. 5 shows a typical
fracture surface of one of the steel specimens. Figs. 6 and 7 show the graphs representing the middle line in the
aluminum 1 and the steel 1 specimen, respectively.
Table 3 shows the total number of striations calculated for each line in each specimen.
Table 1
The parameters and calculation results
Aluminum specimen Steel specimen
a
0
[in] 0.375 1.2
W [in] 1.89 1.96
b [in] 0.316 0.472
K
1C psi

in
p

32,000 2,70,000
f(a
0
/w) 4.268 14.34
DK psi

in
p
12,000 32,900
DP [lb] 1100 1516
a
cr
1.096 1.76
E. Hershko et al. / Engineering Failure Analysis 15 (2008) 2027 23
Table 2
The results obtained from the fatigue experiment
Specimen Total number of load cycles Distance from origin (mm) Critical crack length (mm)
Aluminum 1 50,250 3.8 32
Aluminum 2 35,875 0 27
Steel 1 59,381 0.67 29
Steel 2 97,752 1.63 33
Fig. 4. Typical fracture surface of an aluminum sample and measurement lines.
Fig. 5. Typical fracture surface of a steel sample. One measurement line is shown.
24 E. Hershko et al. / Engineering Failure Analysis 15 (2008) 2027
Fig. 6. Experimental and fractographic results aluminum sample.
Fig. 7. Experimental and fractographic results steel sample.
Table 3
The total number of striations calculated for each line, in each specimen
Specimen Line Total number of striations Average total number of striations
Aluminum 1 Top 54,784 57,696
Middle 59,098
Bottom 59,207
Aluminum 2 Top 36,563 33,162
Middle 31,590
Bottom 31,334
Steel 1 Top 58,204 59,446
Middle 58,795
Bottom 61,340
Steel 2 Top 50,837 57,397
Middle 60,941
Bottom 60,412
E. Hershko et al. / Engineering Failure Analysis 15 (2008) 2027 25
4. Discussion
The relative errors (in percentage) of the average total counted striations with respect to the number of load
cycles applied to the specimen are shown in Table 4. The table also shows the relative error with regard to the
line where the total number of striations counted was closest to the total number of the applied load cycles.
In three out of the four specimens, a good conformance between the calculated number of striations and the
real number of cycles applied during crack propagation was found.
It is however notable that the conformance of the average error column with the real number of load cycles
is lower.
These errors are the result of the following problems and diculties that one must be aware of when count-
ing fatigue striations by a SEM:
1. One must always make sure that the area of the striations counting is planar and normal to the electron
beam. Any inclination of the inspected area will result in an inaccurate density measurement, which directly
distorts the total number of calculated striations.
2. It was found that counting 10 equally spaced areas along three dierent lines from the origin to the nal
rupture zone was a redundant procedure. No distinguishing characteristics were found inherent to any
one line on a specimen and no signicant variation of the striation density was observed. Hence, it is rec-
ommended to concentrate on one line, along which to examine 23 dierent areas at each location, in order
to minimize the errors.
3. It is imperative to make sure that the line one follows from the origin to the nal rupture zone exactly fol-
lows the path of the crack propagation. This can be achieved by carefully examining the shape and direc-
tion of the fatigue striations and fan lines on the fracture surface. Counting striations in locations from two
or more dierent paths will aect the results of the calculation.
4. While examining the fracture surfaces, it became evident that there were several degrees of striations. For
example, in steel specimens, large fatigue tears are visible in magnications of 25006000, ne fatigue tears
are visible in magnications of 10,00020,000 and even ner fatigue striations are visible in magnications
of 20,00045,000 (the same is true for aluminum specimens, but there the smallest striations are already
visible at 20,000 magnication). Although occasionally dicult to nd, only the nest striations truly rep-
resent a single load cycle. Counting the larger striations (at lower magnications) will inevitably yield
results far smaller than the actual number of load cycles applied to the specimen.
The inaccuracies found in the project were, therefore, a result of one or more of the above diculties. If all
the above conditions are met, counting striations using an electron microscope should yield the number of
load cycles with high accuracy.
5. Conclusions
This study shows that a scanning electron microscope may be a very eective tool to determine, accurately,
the number of fatigue cycles that propagated a crack. We used striation counting to evaluate the number of
fatigue cycles on aluminum alloy AA-2024-T3 and low alloy steel AISI-4130 annealed specimens.
We analyzed the results and identied several factors that may distort the results (inclination of the
inspected area, straight line from origin to the nal rupture zone and degree of striation counted). Proper
account for these issues will increase the reliability of the results.
Table 4
The relative errors (in percentage) of the counted striations with respect to the number of load cycles applied to the specimen
Specimen Average error percentage (%) Best line error percentage (%)
Aluminum 1 14.8 9
Aluminum 2 7.6 +1.9
Steel 1 +0.1 0.98
Steel 2 41.3 37.65
26 E. Hershko et al. / Engineering Failure Analysis 15 (2008) 2027
Our experiments were performed with a xed load value and frequency, in a controlled environment.
Therefore it is unclear how accurate this procedure will be in an actual failure investigation, where these values
may vary constantly. This should be the subject of a further research.
References
[1] Connors WC. Fatigue striation spacing analysis. Mater Charact 1994;33(3):24553.
[2] Odrico J, LeGall Ph. Striation counting on fatigue fractured light alloys. Rev Phys Appl 1974;9(July):67381.
[3] Khan Z, Rauf A, Yonas M. Prediction of fatigue crack propagation life in notched members under variable amplitude loading. J Mater
Eng Perform (USA) 1997;6(3):36573.
[4] Forth J, Schutz W, Crack Propagation Under Constant and Variable Stress Amplitudes: A Comparison of Calculations Based on the
Striation Spacing and Tests, AGARD (NATO), vol. CP-376, no. CP-376, November 1984. p. 17.117.9.
[5] Goswami T. Flange bolt failure analysis. J Mech Behav Mater (UK) 1999;10(4):20514.
[6] Gerberich WW, Nelson JC, Jungk JM. Fatigue life predictions in small volume components. In: Fatigue: A David L. Davidson
symposium as held at the 2002 TMS annual meeting, Seattle, WA, USA, 1721 February 2002. p. 12134.
E. Hershko et al. / Engineering Failure Analysis 15 (2008) 2027 27

También podría gustarte