measuring overload effects during fatigue crack growth in bainitic steel
TRANSCRIPT
-
7/24/2019 Measuring Overload Effects During Fatigue Crack Growth in Bainitic Steel
1/6
Measuring overload effects during fatigue crack growth in bainitic steel
by synchrotron X-ray diffraction
P. Lopez-Crespo a,, A. Steuwer b,c, T. Buslaps d, Y.H. Tai e, A. Lopez-Moreno f, J.R. Yates g, P.J. Withers h
a Department of Civil and Materials Engineering, University of Malaga, C/Dr Ortiz Ramos, s/n, 29071 Malaga, Spainb MAX IV Laboratory, Lund University, Box 118, SE-221 00 Lund, Swedenc NMMU, Gardham Avenue, 6031 Port Elizabeth, South Africad ESRF, 6 rue J Horowitz, 38000 Grenoble, Francee Rolls-Royce plc, PO Box 31, Derby DE24 8BJ, UKfDepartment of Materials Science and Metallurgy Engineering, University of Jaen, Campus Las Lagunillas, 23071 Jaen, Spaing Simuline Ltd., Derbyshire S18 1QD, UKh School of Materials, University of Manchester, Grosvenor St., Manchester M13 PL, UK
a r t i c l e i n f o
Article history:
Received 28 October 2013
Received in revised form 9 March 2014
Accepted 17 March 2014
Available online 26 March 2014
Keywords:
Overload effect
Fatigue crack closure
Residual stress
X-Ray diffraction
a b s t r a c t
In this work we present the results of in situ synchrotron X-ray diffraction measurements of fatigue
crack-tip strain fields following a 100% overload (OL) under plane strain conditions. The study is made
on a bainitic steel with a high toughness and fine microstructure. This allowed a very high (60lm) spatial
resolution to be achieved so that fine-scale changes occurring around the crack-tip were captured along
the crack plane at the mid-thickness of the specimen. We have followed the crack as it grew through the
plastic/residually stressed zone associated withthe OL crack location. We observed two effects; one when
the enhanced plastic zone is ahead of the crack and one after it has been passed. Regarding the former it
was found that the compressive stress at the crack-tip initially falls sharply, presumably due to the
increased plastic stretch caused by the OL. This is associated with a concomitant fall in peak tensile stress
atKmax, the elastic excursion between KminandKmaxremaining essentially unchanged from before OL.
Subsequently discontinuous closure as seen previously for plane stress caused by crack face contact at
the OL location limits the elastic strain range experienced by the crack tip and thereby retards crack
growth.
2014 Elsevier Ltd. All rights reserved.
1. Introduction
The concept of crack closure has been used to explain many
crack retardation effects in the fatigue of materials. Closure encom-
passes effects that cause the crack faces to close early during
unloading so that the crack-tip does not experience the full
crack-opening fatigue cycle. Plasticity induced crack closure isone of the most important mechanisms of crack closure, but is still
a hotly debated subject with some researchers suggesting it does
not occur at all[1],while others believe that it can only occur un-
der plane stress[2]. To date, experimental measurements of crack
closure for plane strain samples have been inconclusive relying on
either (i) measuring some secondary property of the cracked body
such as compliance or electrical resistance or (ii) measurement of
crack-opening displacements on the surface of the cracked body.
Third generation synchrotron X-ray facilities allow experimental
measurement of the strain field within the interior of the speci-
men. Recently it has been shown that it is possible to map in 2D
the strain fields around the crack-tip, both with neutron diffraction
[3]and synchrotron X-ray diffraction[46].
Croft et al.[6,7]have studied the crack-tip stress fields during
and after an overload event in 4 mm thick (approximately planestress) steel samples. In some well-designed experiments they
found evidence of discontinuous crack closure in the locality of
the overload event at distances as large as 1.5 mmbehind the crack
tip. This work has also been corroborated by synchrotron strain
mapping in 5 mm Ti6Al4V samples[8].
Under plane strain conditions the evidence obtained to date for
crack closure is not so clear. The highest spatial resolution crack-
tip strain measurements (25 lm) have been made on a very fine
grained AlLi alloy, but the low fracture toughness meant that un-
der plane strain conditions the plastic zone was very small [9].
Nevertheless it was possible to extract accurate measures of the
http://dx.doi.org/10.1016/j.ijfatigue.2014.03.015
0142-1123/ 2014 Elsevier Ltd. All rights reserved.
Corresponding author.
E-mail address: [email protected](P. Lopez-Crespo).
International Journal of Fatigue 71 (2015) 1116
Contents lists available at ScienceDirect
International Journal of Fatigue
j o u r n a l h o m e p a g e : w w w . e l s e v i e r . c o m / l o c a t e / i j f a t i g u e
http://dx.doi.org/10.1016/j.ijfatigue.2014.03.015mailto:[email protected]://dx.doi.org/10.1016/j.ijfatigue.2014.03.015http://www.sciencedirect.com/science/journal/01421123http://www.elsevier.com/locate/ijfatiguehttp://www.elsevier.com/locate/ijfatiguehttp://www.sciencedirect.com/science/journal/01421123http://dx.doi.org/10.1016/j.ijfatigue.2014.03.015mailto:[email protected]://dx.doi.org/10.1016/j.ijfatigue.2014.03.015http://-/?-http://-/?-http://-/?-http://-/?-http://-/?-http://-/?-http://-/?-http://-/?-http://-/?-http://-/?-http://-/?-http://crossmark.crossref.org/dialog/?doi=10.1016/j.ijfatigue.2014.03.015&domain=pdfhttp://-/?- -
7/24/2019 Measuring Overload Effects During Fatigue Crack Growth in Bainitic Steel
2/6
crack-tip stress intensity factor at minimum, maximum and over-
load stresses,Kmin,KmaxandKOL. More recent experiments have al-
lowed the measurement of crack-tip strains under plane stress
conditions with characteristically larger plastic zones [10,11].
Overload events have been studied at the surface via digital image
correlation[11]and microscopy[7,12]and in the interior, via syn-
chrotron X-ray diffraction[13]. However, the large grain sizes in-
volved in previous studies (50lm) did not allow sufficiently
high resolution strain mapping to resolve the changes occurring
immediately local to the crack-tip. The current work aims to exam-
ine the effect of overload during fatigue crack growth in a bainitic
steel. The attractions of using such a steel are firstly the high frac-
ture toughness which allows high levels of applied DKduring fati-
gue and hence a large plastic-zone, and secondly the very fine grain
size, which allows excellent resolution when mapping the strain
fields around the crack-tip. In the work that we report here, it
has been possible to track the crack-tip strain field in a single sam-
ple at various stages of crack growth past the overload event.
2. Experimental procedure
2.1. Material and specimen
A compact tension (CT) fatigue specimen was machined from
quenched and tempered steel similar to Q1N (HY80) [14]. Its
chemical composition is summarised inTable 1. The tensile prop-
erties are as follows: Yield Stress = 570 MPa and Ultimate Tensile
Stress = 663 MPa. The CT specimen had a width of 62.5 mm and
thickness (B) of 12 mm.
The steel has a typical bainitic microstructure, shown inFig. 1,
with an approximate grain size of 5 lm.
2.2. X-ray diffraction experimental setup
The crack-tip elastic strain fields were measured on beam line
ID15A at the European Synchrotron Radiation Facility (ESRF) inGrenoble, using the same arrangement as that described in [5]as
shown schematically in Fig. 2. The scattering angle was 2h= 5.
The strains were derived by analysing the shifts in the (211) dif-
fraction peak. The incident beam slits were opened to
60 60lm giving a lateral resolution (x,y) of 60 lm and a nominalgauge length through-thickness (z) of around 1.4 mm. This allowed
a 10 times greater resolution than in previous plastic zone map-
ping experiments under plane strain conditions[11]. Such a good
resolution was possible because of the very fine microstructure
of the bainitic steel used here (see Fig. 1), meaning that even at
such small gauge volumes, sufficient number of grains in the gauge
volume contribute to the diffracted signals to allow powder analy-
sis of the diffraction patterns[16].
There are a number of methods for identifying the stress free
lattice parameter, as discussed by Withers et al. [17]. The initial
selection of a stress free lattice parameter far from the crack tip
gave a residual strain of400 106 across the crack faces at Kmax.This may be the result of Poissons ratio effects or plastic anisot-
ropy [18]. Instead a stress-free lattice parameter was chosen so
as to give zero strain across the (open) crack faces at Kmaxfor the
baseline fatigue case (OL 1). It is noteworthy that previous stud-ies have seen similar slightly negative (compressive) strains in the
crack wake[5,6].
Great care was taken throughout the experiment to ensure that
any sample movement during loading toKmaxor unloading toKminwas taken account of. Since the crack-tip lies deep within the bulk
it is not possible to unambiguously determine the crack-tip loca-
tion. For an ideal crack tip stress field it would be a simple matter
to determine the crack position simply from the locationof the sin-gularity. However in reality the crack-tip is not so sharp, in part be-
cause of plasticity, in part because the gauge volume is 60lm wide
and in part because the crack will not be perfectly straight or par-
allel to the z direction noting that the gauge is 1.4 mm long
through thickness. This was studied by cooling the sample to liquid
nitrogen temperature and cracking it open once the XRD experi-
ment was finished. Beach markings indicated that at least for the
OL condition the crack front was essentially straight with around
1 mm difference in length from side to side.
For plane stress the plastic zone, rp, would be expected to be
approximately (DKI/rYS)2/2p= 380 lm or around approximately
(DKI/rYS)2/6p= 120 lm for plane strain (although considerably
smaller along the line of the crack) which is close to the gauge
dimension. From a gauge volume smoothing view point one couldtake the crack-tip to be located at the mid-height of the rising
Table 1
Chemical composition in weight % of Q1N steel. The balance is Fe.
Alloy C Si Mn P S Cr Ni Mo Cu
Q1N 0.16 0.25 0.31 0.010 0.008 1.42 2.71 0.41 0.10
Fig. 1. Optical micrograph of the bainitic steel used in the current work. The
micrograph was obtained at 1000X magnification.
Fig. 2. Schematic of the diffraction geometryshowinga CT specimen with the crack
plane horizontal, andthe two detectors measuring two directions of strain; note the
coordinate system forexx andeyyadopted after[5]. Given that h = 2.5, these strains
can be taken as representative of those in the loading (y) and crack growth (x)
directions.
12 P. Lopez-Crespo et al. / International Journal of Fatigue 71 (2015) 1116
-
7/24/2019 Measuring Overload Effects During Fatigue Crack Growth in Bainitic Steel
3/6
elastic strain curve. Here we have takenthe crack-tip to lie near the
top of the elastic strain response which is possibly around rp in
advance of the actual crack-tip position.
2.3. Fatigue experiment
The specimen was fatigue pre-cracked for 43,000 cycles at a fre-
quency of 10 Hz and stress intensity range DK= 28 MPapm andload ratio Kmin/Kmax= 0.05. Plane strain conditions were met at
the mid-plane through the thickness for all loads applied during
the experiment[19]. The crack length was measured perpendicu-
larly to the loading direction from the centre of the loading holes
[15]. Once the fatigue crack had grown to a length of 12.75 mm,
a 100% overload (OL) was applied. Strain measurements were
made on a number of occasions, namely, during the cycle immedi-
ately before the overload (OL 1); during the overload (OL); 20 cy-cles after the overload (OL + 20); 1000 cycles after the overload
(OL + 1000); 11,000 cycles after the overload (OL + 11,000); and
21,000 cycles after the overload (OL + 21,000). By measuring
around 50 strain points along the crack plane (y= 0), a profile of
the strain evolution behind and ahead of the crack-tip was pro-
duced for each of the stages of fatigue crack growth.
3. Results and discussion
Historically, it has been the practice to correlate long fatigue
crack propagation with a single fracture mechanics parameter
based on the applied range of stress intensity factor. As an engi-
neering approximation for constant amplitude loading this has
served engineers well for more than half a century. In the presence
of overloads, and cycles of varying amplitude, the correlations
breakdown as process of crack growth is more complex than sim-
ply described by the range of the elastic crack tip displacements.
The elegant work by Liu and others [20,21] clearly show the
non-linear relationship between external load and crack tip
displacements.The traditional approach to modifying the stress intensity factor
crack in the presence of closure is to define the level at which the
crack faces touch, represented by Kcl, and contrive an effective
range ofDK. This is often identified by a knee in the crack compli-
ance during unloading from Kmax, as recorded by a back-face strain
gauge[22]or by DIC[23]. A recent, and more promising, approach
has been proposed by James and co-workers [24] in which they use
four parameters to capture the different influences on the crack tip
stress field. The important feature of this work is the separation of
a retardation parameter, governed by the plastic enclave, from the
conventional elastic opening terms of the field.
To investigate the manner in which the crack-tip strain field
varies with unloading a series of measurements were made during
an unloading cycle (Kmax, 0.7Kmax, 0.2Kmax, Kmin) and the results areshown in Fig. 3a. Furthermore the changes in elastic strain through
the loading cycle are shown inFig. 3b with reference to the strains
at Kmin and compared directly with those expected for linear elastic
fracture mechanics:
eyyx KIffiffiffiffiffiffiffiffiffi2px
p 1 2t1 tE
where t is Poissons ratio, Eis Youngs modulus andxis the distance
along the crack line as shown in Fig. 2.FromFig. 3b it is clear that
the strains increase in accordance with linear elastic fracture
mechanics fromKminexcept in the immediate vicinity of the crack
tip where the linear elastic curve becomes singular while yielding
limits the elastic strains achieved in practice. The maximum elastic
strain is consistent with a multi-axial stress around (1650 MPa,1650 MPa, 990 MPa) suggesting multiaxial yielding at around
660 MPa (under the Tresca yield criterion) which is close to the
UTS for uniaxial loading. The plastic zone radius appears to bearound 100 lm along the x direction, in accordance with
predictions.
From these measurements it is clear that compressive stresses
have started to develop by the time the load has fallen to 0.2Kmax.
Although this compressive trough is normally depicted in terms of
stresses [25,26], it is common to represent it in terms of strain
when experiments are performed with XRD[11,27]. Indeed the
compressive zone appears to increase only marginally both in
terms of depth and extent with further unloading to Kminalthough
the peak tensile strain ahead of the crack at the edge of the plastic
zone continues to fall to a value of around 400 106 atKminlo-cated approximately 250 lm ahead of the crack tip. Very similar
behaviour has been observed in the region of the crack tip by Croft
et al.[7]. While it is difficult to ensure completely faithful registrybetween scans atKmaxandKmin, in both their work and our work
the compressive zone appears to span the crack-tip in our case
spanning a distance of around -360 lm. Of course stress equilib-
rium requires stress balance across thexz plane at zero load. At
Kmin the essentially residual stresses along the centreline (z= 0,
y= 0) do not appear to balance; the tensile residual stresses ahead
of the crack being larger than the compressive stresses in the vicin-
ity of the crack. That might suggest that crack is being held open
somewhat at the mid-plane (z= 0) by the material towards the sur-
faces of the CT specimen.
Fig. 4shows the evolution in the strain field in the crack open-
ing direction along the crack plane (y= 0) at mid-thickness (z= 0)
both at maximum and minimum loading as the sample undergoes
the overload event (at x = 0) and then as the crack grows past it.The peak tensile strain at overload is around 4750 106 which
-2 0 2 4 6
Distance from Crack-tip (mm)
-1000
0
1000
2000
3000
4000
Elasticstrain(10-
6) Kmax
0.7Kmax
0.2Kmax
Kmin
5000
8
-2 0 2 4 6
Distance from Crack-tip (mm)
8
-1000
0
1000
2000
3000
4000
Realativeelasticstrain(10-6)
5000
Kmax
0.7Kmax
0.2Kmax
Fig. 3. (a) The variation in the crack opening elastic strain measured mid-thickness
(z= 0)along the crack plane (y = 0) as the sample was unloaded from Kmaxto Kminat
OL 1 (1 cycle prior to overload), and (b)the change in elastic strainrelative to Kmincompared to the ideal elastic crack-tip response (dashed lines). Note that the crack-
tiplocation (x= 0) wastakento be the pointof maximumtensile strainand so could
berp (60120 lm) ahead of the actual crack tip position.
P. Lopez-Crespo et al. / International Journal of Fatigue 71 (2015) 1116 13
http://-/?-http://-/?-http://-/?-http://-/?- -
7/24/2019 Measuring Overload Effects During Fatigue Crack Growth in Bainitic Steel
4/6
represents a triaxial stress of around (1780 MPa, 1780 MPa,
1070 MPa) which is equivalent to mutli-axial yield at a level
around 710 MPa. It is clear that very little crack growth has takenplace after 20 and 1000 cycles, since in each case the tensile crack-
tip stress field is located at approximately the same location as for
the OL curve. After 10,000 additional cycles the crack has pro-
gressed around 300 lm while after an additional 20,000 cycles
the crack has progressed 1070 lm beyond the overload event. It
should be noted that during the loading cycles the CT sample
moves slightly in response to the applied load, however we have
tried to correct for movement in both the crack growth direction
(x) and perpendicular to it (y). Given the overload the broad and
shallow residual compressive residual stress at the crack at Kminat OL is surprising and may be because the scan line just misses
the actual crack tip location. For cycles (OL + 20 and OL + 1000) it
is clear that at Kmin there is an extensive compressive trough at
the crack location arising from the OL. This has the effect ofdepressing the tensile peak at Kmax compared to that just before
the overload (OL 1). With the crack having moved away fromthe OL location there is clear evidence of a compressive stress
(compression across the crack faces) at the OL location (x= 0) for
OL + 11,000 atKminand limited evidence of crack face compression
at OL + 21,000 as observed by Croft et al. for plane stress[7]. The
low level tensile strains (
-
7/24/2019 Measuring Overload Effects During Fatigue Crack Growth in Bainitic Steel
5/6
the synchrotron beam both vertically (y) and horizontally (x).
While great care was taken to ensure that the linescans were accu-
rately aligned with the crack tip, it is possible these differences in
the location and extent of the compressive zone are due to slight
differences in the locations of the linescans and 2D strain maps
are currently being acquired [11] to check this. Besides the changes
in the immediate vicinity of the crack tip, the compressive contact
stresses in the region of the crack tip at overload atKminare clearly
evident in Fig. 6 as positive peaks in the KmaxKmin curves for
OL + 11,000 and OL + 21,000. These compressive stresses appearto hold open the crack reducing the extent of the excursion in
elastic strain at the crack tip by 30% for OL + 11,000 and to a lesser
extent (25%) for OL + 21,000. Of course it is possible that compres-
sive stresses also arise behind the crack after 20 and 1000 cycles
but as is clear fromFig. 5b they are much less extensive.
The effect of the overload is studied in Fig. 7 in terms of the var-
iation in the maximum strain excursion (i.e. the maximum strain
difference betweenKmaxand Kmincurves inFig. 4), the maximum
tensile strain and the maximum compressive strain as a function
of crack growth past the OL crack tip location. It is clear that the
OL modifies the strain curve for each fatigue stage. Both maximum
compression and maximum tensile strain show a similar trend.
Similar trends were found in 4140 steel specimens also subjected
to 100% OL[12]. However further experiments are required to cor-
relate them with crack growth rates. Before the OL the compressive
zone peaks at 300 106 and deepens to1500 106 twentycycles after overload. It then falls back to the baseline fatigue value
(
300
106) by the time the crack has grown
1 mm beyond
the overload event (OL + 21,000). One of the most accepted theo-ries is that the effect of the OL extends for a length equal to the size
of the plastic zone [11,28,29], however the compressive zone here
(closure plus plastic zone) is considerably smaller along x, being
around 600lm.It is also worth noticing the striking resemblance between Fig. 7
and crack growth curves (see for exampleFig. 1in[7]), suggestive
of a relation between maximum compressive strain at Kmin or max-
imum tensile strain atKmaxand crack growth data while the crack
grows through the OL plastically affected zone. The increase in the
compressive strain at Kmin after OL is probably because of the in-
crease in plastic (stretch) deformation in the increased plastic zone
ahead of the crack rather than crack closure. The OL also increases
the crack opening thus promoting a decrease in crack face closure
behind the crack-tip as the faces come together on unloading.Whilst the four parameter stress field model of James et al.[24]
has yet to be applied to synchrotron data, it is clear from our work
that there would be value in doing so. The dramatic changes ob-
served in the strain field data after the OL event will provide valu-
able insight into the role of the changing retardation and shear
stress parameters on fatigue crack propagation.
4. Conclusions
Using very high spatial resolutions (60 lm inx,y) we have been
able to explore the elastic strains in the crack opening direction e
(x,y= 0), in the vicinity of a fatigue crack before and after an over-
load event. We have followed the crack until it is 1 mm beyond theoverload location. Unlike most of the work to date, the crack-tip
Fig. 5. Strain evolution mid-thickness along the crack plane at (a)Kmaxand (b)Kminfor all the stages of fatigue crack growth analysed. The coordinate along the crack
plane (horizontal axis) has been shifted so that the crack-tip positions coincide for
all cases.
Fig. 6. Plots showing the change in elastic strain between Kmax and Kmin as afunction of number of cycles, i.e.KmaxKmincurve.The dashed lines show the ideally
elastic curves.
-2000
-1000
0
1000
2000
3000
4000
5000
6000
-0.2 0.0 0.2 0.4 0.6 0.8 1.0 1.2
Distance from the OL (mm)
Relative
elasticstrain(10-6)
Max excursion
Max tensile
Max compression
Fig. 7. Maximum strain excursion (i.e. maximum strain difference betweenKmaxand Kmin curves in Fig. 4), maximum tensile strain and maximum compressive
strain as a function of crack growth from the OL location. The second data point for
each curve represents the OL.
P. Lopez-Crespo et al. / International Journal of Fatigue 71 (2015) 1116 15
http://-/?-http://-/?- -
7/24/2019 Measuring Overload Effects During Fatigue Crack Growth in Bainitic Steel
6/6