Linköping Studies in Science and Technology. Thesis No. 1622 Licentiate Thesis Residual Stresses and Fatigue of Shot Peened Cast Iron Mattias Lundberg LiU-TEK-LIC-2013:56 Division of Engineering Materials Department of Management and Engineering Linköping University, SE-58183, Linköping, Sweden http://www.liu.se Linköping, October 2013
61
Embed
Residual Stresses and Fatigue of Shot Peened Cast Iron
This document is posted to help you gain knowledge. Please leave a comment to let me know what you think about it! Share it to your friends and learn new things together.
Transcript
Linköping Studies in Science and Technology. Thesis No. 1622
Licentiate Thesis
Residual Stresses and Fatigue
of Shot Peened Cast Iron
Mattias Lundberg
LiU-TEK-LIC-2013:56
Division of Engineering Materials
Department of Management and Engineering
Linköping University, SE-58183, Linköping, Sweden
http://www.liu.se
Linköping, October 2013
ii
Cover:
Fracture surface of compact graphite iron with a graphite nodule.
In the classification of grey cast iron it is important as a foundry to be able to
specify, not only, the mechanical properties of the cast component, but also, the
graphite flake size, especially for grey cast iron. The foundry should mention the
graphite flake size, according to a standard, in the cast component. In the standard
EN ISO 945:1994 the flake sizes are divided into eight different categories based on
the longest flake size, expanding from the longest flakes of 100 mm or more down
to the smallest category of flake size of less than 1.5 mm in size, see Figure 2. The
flake size refers to the perceived size in x100 magnification, thus the actual flake
size is 100 times smaller.
9
Figure 2: Graphite shapes in grey cast iron according to EN ISO 945:1994
Large flake size is associated with slow cooling rates in irons having high carbon
equivalent values. The attributed changes in properties of grey cast iron consisting
of large flake sizes are high thermal conductivity and damping capacity on the cost
of strength. Hypoeutectic iron subjected to rapid cooling (not quenching) generally
provides very small flake sizes, which results in a high tensile strength since the
short graphite flakes interrupt the matrix to a smaller extent. Finally we have come
to the category of graphite distributions according to ASTM A 247 and ISO 945:1994
as can be seen in Figure 3.
Figure 3: Graphite distribution in grey cast iron according to ASTM and ISO. Type A Random flake graphite in a uniform distribution. Type B Rosette flake graphite. Type C Kish graphite (hyper-eutectic compositions). Type D Undercooled flake graphite. Type E Interdendritic flake graphite (hypo-eutectic compositions).
Depending on the graphite morphology different mechanical properties can be
achieved with the same type of matrix. For example, a pearlitic flake cast iron with a
nominal flake length of 150 µm has much better damping qualities and lower
tensile strength than a pearlitic flake cast iron with a nominal flake length of 20 µm.
In cast iron the matrix is one of the following: ferritic, pearlitic, austenitic,
martensitic or bainitic (often called austempered). The most common matrix in cast
iron castings is ferrite or ferrite-pearlite. Ferrite is the soft low-carbon α-Fe phase
that has low tensile strength but excellent ductility. Ferrite is often found in
conjunction with undercooling. Pearlite is the eutectoid transformation where the
A B C D E
10
austenite transforms to a lamellar structure of ferrite and cementite. The hardness
and tensile strength of pearlite are higher than in ferrite but with lower ductility.
With a smaller lamellar spacing in the material, the hardness and tensile strength
increases. The lamellar spacing and thus the mechanical properties can be
controlled with a more rapid cooling or alloying elements. The cementite phase in
pearlite is an intermetallic compound (Fe3-C) and is very hard and brittle. As a
comparison of these microconstituents, an increase in hardness from 75 to 200 to
550 HB for the ferrite, pearlite and cementite respectively are accepted as standard
values. Tensile strength and elongation of ferrite are roughly 280 MPa and 60 %
respectively whereas the pearlite possesses a tensile strength up to 860 MPa and 10
% elongation.
11
3. Residual stresses The term residual stress used by engineers is related to local variations in strains
inside the material on a macroscopic or microscopic level without any external load
acting on the material. They arise from elastic response to an inhomogeneous
distribution of non-elastic stains. The most common sources of non-elastic strains,
and thus residual stresses, are plastic deformation, phase incompatibility and
thermal expansion strains. The strain variations can be converted into stresses
which are fundamentally easier to grasp and defined as the force per square meter.
The “conversion parameter” differs between the materials due to their different
response to loading [3,4].
3.1. XRD measurements of residual stresses
Residual stresses in a work piece can be divided into macro- and microstresses [3-
5]. By definition the macrostresses are the same in all phases present at the same
depth in the material. Consider the plastic deformation occurring at the surface
during machining operations where the deformation of the surface layer will be
constrained by the bulk where the plastic deformation is minimal, if existing.
Macrostresses are self-equilibrated through the cross-section of the work piece. In
multi-phase materials the differences in yield point and possible response to
mechanical load results in inhomogeneous strains in the material volume. Strong
phases constraining the weaker ones and giving rise to stresses on a microscale
level, these are the microstresses.
Cast residual stresses are present in the as-cast condition in all castings due to
differences in cooling rates in different parts of the component during solidification,
strength of the mould and cleaning of the cast work piece. The main contributors to
residual stresses that one needs to keep in mind when casting are the different
cooling rates in the work piece and the blast cleaning, or rather the energy of the
blast cleaning process. When casting complex geometries, the wall thickness often
differs much and this does not only affects the thermal residual stresses but also
the microstructure in the final component. When casting a component in grey cast
iron, a sand mould is commonly used. To get rid of the sand residues, sand blast
cleaning is often applied and depending on the energy output of the blasting
medium, considerable amount of residual stresses can be introduced at the surface.
12
The force associated with stresses in material are not measured, instead the
strain(s) are measured and later converted to stresses by different methods
depending on stress measurement used. Perhaps a more correct phrase is to say
that we measure changes in strains by some means, which are related to stresses.
Let´s continue with an introduction to basic residual stress measurements with x-
rays. To do so firstly we need to state the origin of the Cartesian coordinate systems
used; the specimen (S) and the laboratory (L) system, see Figure 4. Once the
constitutive equations that describe the two coordinate systems are expressed we
can apply these to the method used in this project to measure the residual stresses
with x-rays.
Figure 4: Coordinate systems used in stress measurement with x-ray diffraction
The stresses (and strains) are defined in the specimen coordinate system and L3
direction is the bisectrise between the incoming and the diffracted x-ray beam,
defined by the two angles ψ and φ. From this relation we can derive the strain
equation from the sample coordinates to our laboratory system, or vice versa. The
calculated stresses and strains will be expressed in the sample coordinate system.
By tensor transformation we can then write the normal strains in the sample
The believed outcome of these tests was to be able to choose appropriate shot
peening parameters used to increase the fatigue strength on grey cast iron under
axial loading as well as compact graphite iron and grey cast iron under bending
fatigue.
30
6.2. Fatigue testing
The fatigue testing performed within this thesis has been carried out by Ph D
Maqsood Ahmad at Volvo Powertrain. The fatigue testing was done in axial loading
with R=-1 in an electrical high frequency testing machine. Testing frequency was
around 125Hz with a failure criteria set to be a 3% change in stiffness. The fatigue
life was set to be 10 million cycles meaning that if the test piece survives 10 million
cycles then it is regarded to have “infinite” life, also called run-out sample. Both the
slope in the low-cycle-fatigue regime and the fatigue strength were to be
investigated.
The mechanically polished samples were set to be the reference which a change in
fatigue strength should be compared to after shot peening or some other post
treatment.
Two different sample geometries were used, having a kt of 1.05 and 1.33. The
argument of the low kt value instead of a “real” smooth sample having a defined
gauge length was for the simplicity of testing. By not having a defined gauge length
the crack should start and propagate at the thinnest section of the sample which
should lower the scattering in fatigue data. The usage of kt 1.33 was to test if the
material exhibits any notch sensitivity when subjected to post treatments.
The evaluation of the fatigue strength was made by using “staircase”-method also
commonly referred to as Dixon and Mood method proposed 1948 [64] which is
based on the maximum likelihood estimation. The method provides an
approximated formula to calculate both the mean (σfat) and the standard deviation
(σstd) of the fatigue strength assuming that the fatigue strength follows a normal
distribution. Also the stress levels needs to be equally spaced in order to use this
method.
In practise, first one specifies the stress spacing to be used, and then start the
testing at an assumed stress level of the fatigue strength. If the specimen is a run-
out then you increase the stress as defined before. On the other hand, if the
specimen failed before the defined fatigue life then you lower stress the same
amount. By doing so the statistical analyses of the data in terms of the fatigue
strength, standard deviation and confidence intervals becomes easier, but still not
very easy.
All evaluation of the fatigue data has been done by Ph D Maqsood Ahmad.
31
6.3. Residual stress profiling
A central part of the project, where this thesis has been carried out, is to measure
residual stresses in cast iron. Different surface treatments result in different surface
stresses and also different subsurface distributions of the residual stresses. This
distribution of stresses in the material affects the fatigue properties, and to get a
better knowledge on how the post treatment performed introduces residual
stresses can be of high importance. For example can the distribution be of good use
when analysing the components fatigue properties. To obtain the residual stress
profile with laboratory x-rays, material needs to be removed in small steps since the
penetration depth of the x-rays are very shallow. This should be done without
introducing any new residual stresses, like those obtained from polishing. To
remove material without introducing new residual stresses, electrolytic polishing is
the best method to use and it the one conducted in this project. Even though it is
practically impossible to remove material without affecting the stress field, with the
electrolytic polishing the residual stress distribution will be affected to a minimum.
When material is removed the stress field will change and if possible stress
corrections due to material removal should be done. In many cases this is practically
impossible due to component geometry without FEM-simulations. Another
problem, or concern, with electrolytic polish is that every type of edge, due to
masking or heterogeneous material, result in a potential peak which led to uneven
material removal. A problem to accept and to live with when working with cast iron,
to minimize the uncertainty of the amount of material removed several depth
measurements have been averaged for every depth.
As described and mentioned earlier in this thesis the sin2ψ-method has been used,
to measure the residual stresses in the cast iron studied. All measurements have
been performed in a four-circle goniometer Seifert X-ray machine. With this
method, only the residual stresses in the ferrite are measured. The measured
stresses are thus phase stress, the sum of macro and micro stresses in the ferrite.
The usage of a Cr-tube for generation of x-rays results in a 2θ angle of
approximately 157° for the {211}-diffraction plane.
To convert the diffraction data obtained to stresses with the sin2ψ-method, a value
on the XEC, ½S2, is needed. In this project that value is 5.81*10^-6. Due to the
heterogeneity of this material, and that the residual stresses in one of the three
phases are measured, lead to calibration of the XEC ½S2. This calibration has
resulted in somewhat un-expected results and need more work to confirm and
validate the results. The results and the discussion are excluded in this thesis and
the calibration is on-going.
32
6.4. Microstructural studies
To gain a better understanding on the material behaviour, microstructural
investigations in a Hitachti SU-70 FESEM scanning electron microscope have been
done. The SEM is equipped with Nordley detector from Oxford Instrument and
Channel 5 software, which was used for EBSD mapping of the ferritic phase. From
the mapping a low angle grain boundary (LAGB) density can be calculated from the
obtained data. The LAGB density plot can then be used to quantify the amount of
deformation in the material. Deformation of the microstructure due to polishing
and shot peening as well fatigue accumulated deformation after testing have been
investigated. New techniques (for these materials) to quantify the deformation
behaviour have been tested, i.e. EBSD mapping, to determine the material response
and deformation due to loading. With electron channelling contrast imagine (ECCI),
plastic deformation in the material results in clear contrast in the image, as can be
seen in the upper left part of Figure 10. These two techniques provide the
researcher with a lot of information regarding the deformation in the material.
Figure 10: ECCI of shot peened grey cast iron. The speckled pattern in the upper left is a result from plastic deformation by shot peening. In the middle there is graphite and in the lower right part of the picture, the un-affected material is shown.
SEM investigations have been done on the first test peening that were done on flat
cylindrical specimens to get a better understanding on the plastic deformation done
by the shot peening conditions tested. The first fatigue testing series were
investigated with both ECCI and EBSD. Also fracture surfaces were investigated. The
33
second fatigue test series with the gentle shot peening condition has also been
investigated with ECCI and EBDS as well as some of those that were short time
annealed.
34
35
7. Appended paper summary
PAPER I
Residual Stresses in Shot Peened Grey and Compact Iron
It is well recognized that shot peening results depend on both the target material
and a number of peening parameters. An industrial process is often controlled by
choosing peening media, i.e. type of shots and shot size, Almen intensity and degree
of peening coverage. Based on the limited publications on shot peening and shot
blasting of cast irons found in the literature, twelve unique combinations of shot
size, intensity and coverage were chosen. The different shot peening treatments
combining different shot sizes, different degrees of coverage and Almen intensities
were applied to a grey cast iron and a compact graphite iron having essentially a
pearlitic matrix. The induced surface residual stresses and subsurface residual stress
distributions as well as plastically deformed depths were investigated by x-ray
diffraction.
Relatively high compressive residual stresses have been induced on the surface of
all the samples. For both GI and CGI, the largest surface residual stress was found
for peening with the smallest shots (S170), low intensity (0.17 mmA) and 100%
peening coverage and the lowest surface stress for peening with the largest shots
(S550), high intensity (0.29 mmC) and 300% peening coverage. An increased
coverage from 100% to 300% had a minor effect on the subsurface residual stress
distribution as measurements on selected GI samples show. This is explained by a
small effect of further peening on the cyclic deformation behaviour of the cast iron.
Peening with a higher intensity, on the other hand, strongly affected the residual
stress depth profiles. Also the depth of the compressive zone was greatly increased
for both GI and CGI due to a larger depth of plastic deformation. When increasing
the shot size, the plastically affected depth does not change pronounced with
similar peening intensities, but the residual stress profile and the amount of
residual stresses at depth will somewhat differ between the two materials. The
better response of CGI to shot peening can be related to its microstructure,
especially graphite morphology.
36
PAPER II
Shot Peening Induced Plastic Deformation in Cast Iron – Influence of Graphite
Morphology
To increase the fatigue strength, shot peening of the component is often conducted
and proven to be efficient on e.g. steel and aluminium. When shot peening, the
shots interact with the surface and result in high level of local plastic deformation.
Due to different graphite morphology it is therefore suspected that grey cast iron
and compact graphite iron response differently to identical shot peening
parameters. While the limited literature available in this field mostly deals with shot
peening of nodular graphite cast irons, thus the influence of graphite morphology
on shot peening results is not thoroughly investigated. The purpose with this work
is to obtain better knowledge on the different plastic behaviours between grey and
compact iron. Using electron backscatter diffraction (EBSD) and electron
channelling contrast imaging (ECCI) the microstructural changes in the peening
affected zone are quantified. The same parameters were used as in Paper I on both
materials. With EBSD mapping, LABG density can be calculated, giving information
about the plastic deformation distribution from the peening process in depth
material.
The better response of CGI to shot peening in the form of larger plastic deformation
and higher compressive stresses in the subsurface was attributed to the difference
in microstructure, especially graphite morphology, and different capability for
plastic deformation of the matrix. Flake graphite inclusions at certain depth can
effectively hinder the propagation of plastic deformation into the interior of the
sample. Other graphite inclusions do not cause a large reduction of plastic strains of
the matrix. All the observed graphite morphologies can locally raise plastic
deformation in its surrounding matrix. A higher degree of plastic deformation in the
pearlite dominant subsurface regions of compacted cast iron compared to grey cast
iron was found. This can be related to the differences in microstructure such as
thickness of pearlitic lamellar and density of LAGBs.
PAPER III
In-situ SEM/EBSD Study of Deformation and Fracture Behaviour of Flake Cast Iron
Since flake cast iron exhibits brittle failure under tension loading, due to its many
graphite tips acting as notches or small cracks that are considered to be initiation
points for cracks in the material. The common knowledge on how cracks propagate
37
in flake cast irons is that the crack initiation point is one or several of the many
graphite tips, from the tips the crack propagates along the graphite flakes in the
graphite-matrix interface. At strains just before break down, the crack propagates
through the matrix and connecting graphite tips. In this study the crack initiation
point(s) and crack propagation in a grey cast iron were studied during axial loading
in a SEM. Thanks to a specially designed in-situ stage, analysis of the crack initiation
and its propagation as well as crystallographic changes was investigated. The
microstructural features associated with different strain levels were investigated
with SE-mode and EBSD.
From the performed experiment, the fracture behaviour of flake cast iron can be
summarized in a few steps of main events. The first noticeable microstructural
feature due to the axial loading is the opening of graphite. Graphite flakes opens
inside, not all flakes, and this feature can be found in graphite flakes independent of
its orientation to the applied load all over the surface. A second important and
noticeable event is the delamination of the graphite-matrix interface. Parallel to
this event one can see that the graphite that had opened inside opens even more.
The third cracking feature is the local plasticity (microplasticity) at graphite tips lying
perpendicular to the loading. Fourth and last important feature, before break down,
is the bulge of the matrix at the delaminated interface, where the crack later will
propagate. Also some of the graphite with an opening inside had also generated a
delamination of the graphite-matrix interface. Development of bulges start once
the plastic deformation at graphite tips is evidence, but strains resulting in stresses
close to and above σys0.2% are needed for a bulge to develop. At applied loadings just
before rapture the fracture behaviour of flake cast iron results in multiple cracking,
which makes it possible for the main crack to “jump” relatively large distances due
to the network of flaky graphite. The weakest points in the sample will then link
together which results in a rough topography of the fracture surface. As a final
occurrence of fracture behaviour in flake cast iron subjected to axial loading the
crack propagates in one of the following alternatives:
• Through the opening inside the graphite.
• Through the delaminated interface.
• Via the closest distance between graphite tips where the local plastic
deformation results in a ductile-like fracture appearance.
• At the pearlitic grain boundaries leaving a ductile-like fracture.
• Straightforward through the matrix resulting in a cleavage fracture.
• Along the ferrite-cementite interface giving cleavage fracture.
38
With EBSD changes in grain orientation and LAGBs in flake cast iron due to axial
loading was not detected for the set-up used. Analyse of the fractured surface
revealed both ductile and cleavage fracture and an increase in fracture topology
where multiple cracks was easily detected.
PAPER IV
Fatigue Strength of Machined and Shot Peened Grey Cast Iron
A common opinion is that cast iron, especially grey cast iron, is not as notch
sensitive as steel and has therefore not been treated by shot peening to suppress
crack initiation. For a heterogeneous material that also is brittle, just like grey cast
iron, the shot peening parameters needed to induce beneficial surface residual
stresses can be problematic to identify. Fatigue testing under uniaxial loading with
an R value of -1, on mechanically polished and shot peened specimens, has been
performed to determine the fatigue strength at 107. Two different types of
specimen geometries were tested, one smooth and one notched specimen having kt
equal to 1.05 resp. 1.33. With large shots and high peening intensity (S330 shots
and 0.16 mm C peening intensity) the fatigue strength clearly decreased whereas
small shots and low peening intensity (S70-H shots and 0.07 mmA peening
intensity) might have lowered the fatigue strength. A short time annealing at 285°C
after gentle shot peening increased the fatigue strength. The results are discussed
and explained based on x-ray diffraction measurements, i.e. residual stress and full
width at half maximum profiles, as well as microstructural investigations in SEM.
The results showed that in comparison with the mechanically polished specimens,
the gentle shot peening specimens had a similar, if not slightly lower, fatigue
strength but the heavy shot peening had clearly lower fatigue strength. This could
be largely attributed to a negative effect of tensile residual stresses in the
subsurface layer, reviled by stress correction due to material removal from the
obtained residual stress profiles. Damage in the surface layer in form of
microcracking and surface roughness for the heavy shot peened specimens could
also contribute to the lower fatigue strength. The positive results on the fatigue
strength found after a short time annealing at 285°C on specimens being gently shot
peened could be a combination of ageing, associated with Cottrell atmosphere, and
recovery of the material. This resulted in increased fatigue strength of 10%,
compared to the mechanically polished specimen. However, further works is
needed to confirm this.
39
8. Conclusions The research presented in this thesis deals with microstructural changes in cast iron
after different surface treatments, and its effect on axial fatigue strength. Different
surface treatments (in this case) refer to the different shot peening conditions
tested so far in the project. To study the changes, a reference condition is needed,
and the mechanically polished surface was set to be the reference condition in this
project. The changes have been investigated with x-rays (residual stress profiles and
FWHM profiles) and in a SEM.
To find shot peening parameters that will increase the fatigue strength of a low
ductile material is not easy. Trying to find parameters to a heterogeneous material
subjected to shot peening that result in increased fatigue strength is also difficult.
Thus finding proper parameters to shot peen cast iron, which has low ductility and
is heterogeneous due to the graphite, is not straight forward. The general
considered benefits from shot peening have been detected using x-rays in the cast
iron specimens subjected to the shot peening parameters used. That is high
compressive residual stresses at the surface, reaching a relatively large depth and a
clear work hardening of the surface. Despite this, the expected beneficial effects
from shot peening have not been observed in axial fatigue loading.
From the specimens studied it becomes clear that the fracture process is very
complicated, with multiple cracking and crack networking and crack linking, which
slows down the analyses. Making it more problematic to state clear conclusions
since so many things contribute, different deformation mechanisms in and between
the phases depending on the loading, to the observed phenomenon.
Grey cast iron specimens shot peened very gently and annealed at 285°C for 30
minutes showed an increase in fatigue strength in axial loading with R=-1. Only the
gentle shot peening might have resulted in a low increase in fatigue strength but
after annealing the fatigue strength was clearly increased perhaps due to interstitial
diffusion in a Cottrell atmosphere locking dislocations and some recovery of the